AMD 5th Gen EPYC 9005 Turin vs. NVIDIA GH200 Performance Benchmarks

AMD 5th Gen EPYC "Turin" Linux benchmarks by Michael Larabel for a future article on launch day of EPYC 9005 series Linux performance.

Compare your own system(s) to this result file with the Phoronix Test Suite by running the command: phoronix-test-suite benchmark 2410298-NE-NVIDIAGRA46
Jump To Table - Results

View

Do Not Show Noisy Results
Do Not Show Results With Incomplete Data
Do Not Show Results With Little Change/Spread
List Notable Results
Show Result Confidence Charts

Limit displaying results to tests within:

AV1 2 Tests
BLAS (Basic Linear Algebra Sub-Routine) Tests 3 Tests
C++ Boost Tests 5 Tests
Timed Code Compilation 5 Tests
C/C++ Compiler Tests 11 Tests
Compression Tests 3 Tests
CPU Massive 20 Tests
Creator Workloads 18 Tests
Cryptography 4 Tests
Database Test Suite 3 Tests
Encoding 5 Tests
Fortran Tests 6 Tests
Game Development 5 Tests
HPC - High Performance Computing 21 Tests
Imaging 3 Tests
Common Kernel Benchmarks 2 Tests
LAPACK (Linear Algebra Pack) Tests 2 Tests
Machine Learning 3 Tests
Molecular Dynamics 9 Tests
MPI Benchmarks 10 Tests
Multi-Core 33 Tests
NVIDIA GPU Compute 3 Tests
Intel oneAPI 6 Tests
OpenMPI Tests 16 Tests
Programmer / Developer System Benchmarks 8 Tests
Python 2 Tests
Raytracing 2 Tests
Renderers 4 Tests
Scientific Computing 12 Tests
Software Defined Radio 2 Tests
Server 5 Tests
Server CPU Tests 13 Tests
Single-Threaded 2 Tests
Video Encoding 5 Tests
Common Workstation Benchmarks 2 Tests

Statistics

Show Overall Harmonic Mean(s)
Show Overall Geometric Mean
Show Geometric Means Per-Suite/Category
Show Wins / Losses Counts (Pie Chart)
Normalize Results
Remove Outliers Before Calculating Averages

Graph Settings

Force Line Graphs Where Applicable
Convert To Scalar Where Applicable
Disable Color Branding
Prefer Vertical Bar Graphs
No Box Plots
On Line Graphs With Missing Data, Connect The Line Gaps

Additional Graphs

Show Perf Per Core/Thread Calculation Graphs Where Applicable
Show Perf Per Clock Calculation Graphs Where Applicable

Multi-Way Comparison

Condense Multi-Option Tests Into Single Result Graphs
Condense Test Profiles With Multiple Version Results Into Single Result Graphs

Table

Show Detailed System Result Table

Run Management

Highlight
Result
Hide
Result
Result
Identifier
Performance Per
Dollar
Date
Run
  Test
  Duration
EPYC 9965
October 04
  11 Hours, 35 Minutes
EPYC 9575F
October 02
  10 Hours, 4 Minutes
NVIDIA GH200 144G HBM3e
October 28
  7 Hours, 42 Minutes
Invert Hiding All Results Option
  9 Hours, 47 Minutes

Only show results where is faster than
Only show results matching title/arguments (delimit multiple options with a comma):
Do not show results matching title/arguments (delimit multiple options with a comma):

EPYC 9965

Processor: AMD EPYC 9965 192-Core @ 2.25GHz (192 Cores / 384 Threads), Motherboard: AMD VOLCANO (RVOT1000D BIOS), Chipset: AMD Device 153a, Memory: 12 x 64GB DDR5-6000MT/s Samsung M321R8GA0PB1-CCPKC, Disk: 2 x 1920GB KIOXIA KCD8XPUG1T92, Graphics: ASPEED, Network: Broadcom NetXtreme BCM5720 PCIe

OS: Ubuntu 24.04, Kernel: 6.10.0-phx (x86_64), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xb101018
Java Notes: OpenJDK Runtime Environment (build 21.0.3-ea+7-Ubuntu-1build1)
Python Notes: Python 3.12.2
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS; IBPB: conditional; STIBP: always-on; RSB filling; PBRSB-eIBRS: Not affected; BHI: Not affected + srbds: Not affected + tsx_async_abort: Not affected

EPYC 9575F

Changed Processor to AMD EPYC 9575F 64-Core @ 3.30GHz (64 Cores / 128 Threads).

Processor Change: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xb002110

NVIDIA GH200 144G HBM3e

Processor: ARMv8 Neoverse-V2 @ 3.47GHz (72 Cores), Motherboard: Pegatron JIMBO P4352 (00022432 BIOS), Memory: 1 x 480GB LPDDR5-6400MT/s NVIDIA 699-2G530-0236-RC1, Disk: 1000GB CT1000T700SSD3, Graphics: NVIDIA GH200 144G HBM3e 143GB, Network: 2 x Intel X550

OS: Ubuntu 24.04, Kernel: 6.8.0-47-generic-64k (aarch64), Display Driver: NVIDIA, OpenCL: OpenCL 3.0 CUDA 12.6.65, Compiler: GCC 13.2.0 + CUDA 11.8, File-System: ext4, Screen Resolution: 1920x1200

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-dIwDw0/gcc-13-13.2.0/debian/tmp-nvptx/usr --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto --without-cuda-driver -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Python Notes: Python 3.12.6
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

AMD 5th Gen EPYC 9005 Turin vs. NVIDIA GH200 Performance BenchmarksProcessorMotherboardChipsetMemoryDiskGraphicsNetworkOSKernelCompilerFile-SystemScreen ResolutionDisplay DriverOpenCLEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3eAMD EPYC 9965 192-Core @ 2.25GHz (192 Cores / 384 Threads)AMD VOLCANO (RVOT1000D BIOS)AMD Device 153a12 x 64GB DDR5-6000MT/s Samsung M321R8GA0PB1-CCPKC2 x 1920GB KIOXIA KCD8XPUG1T92ASPEEDBroadcom NetXtreme BCM5720 PCIeUbuntu 24.046.10.0-phx (x86_64)GCC 13.2.0ext41920x1200AMD EPYC 9575F 64-Core @ 3.30GHz (64 Cores / 128 Threads)ARMv8 Neoverse-V2 @ 3.47GHz (72 Cores)Pegatron JIMBO P4352 (00022432 BIOS)1 x 480GB LPDDR5-6400MT/s NVIDIA 699-2G530-0236-RC11000GB CT1000T700SSD3NVIDIA GH200 144G HBM3e 143GB2 x Intel X5506.8.0-47-generic-64k (aarch64)NVIDIAOpenCL 3.0 CUDA 12.6.65GCC 13.2.0 + CUDA 11.8OpenBenchmarking.orgKernel Details- Transparent Huge Pages: madviseCompiler Details- EPYC 9965: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v - EPYC 9575F: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v - NVIDIA GH200 144G HBM3e: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-dIwDw0/gcc-13-13.2.0/debian/tmp-nvptx/usr --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto --without-cuda-driver -v Processor Details- EPYC 9965: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xb101018- EPYC 9575F: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xb002110- NVIDIA GH200 144G HBM3e: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)Java Details- EPYC 9965, EPYC 9575F: OpenJDK Runtime Environment (build 21.0.3-ea+7-Ubuntu-1build1)Python Details- EPYC 9965: Python 3.12.2- EPYC 9575F: Python 3.12.2- NVIDIA GH200 144G HBM3e: Python 3.12.6Security Details- EPYC 9965: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS; IBPB: conditional; STIBP: always-on; RSB filling; PBRSB-eIBRS: Not affected; BHI: Not affected + srbds: Not affected + tsx_async_abort: Not affected - EPYC 9575F: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS; IBPB: conditional; STIBP: always-on; RSB filling; PBRSB-eIBRS: Not affected; BHI: Not affected + srbds: Not affected + tsx_async_abort: Not affected - NVIDIA GH200 144G HBM3e: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

EPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3eLogarithmic Result OverviewPhoronix Test SuiteJohn The RipperminiBUDEMemcachedOpenSSLNAS Parallel BenchmarksGROMACSXmrigStockfishLiquid-DSPASKAPsrsRAN ProjectASTC Encoderlibavif avifenc7-Zip CompressionQuantLibSpeedbOpenVINOPrimesieveRocksDBParallel BZIP2 CompressionSecureMarkPennantCoremarkTimed Node.js CompilationTimed LLVM CompilationHelsingZstd CompressionWebP Image EncodeHigh Performance Conjugate GradientPyBenchTimed Godot Game Engine CompilationminiFEAlgebraic Multi-Grid BenchmarkNumpy BenchmarkClickHouseQMCPACKCloverLeafGraphicsMagickNWChemTimed Gem5 CompilationOpenFOAMLAMMPS Molecular Dynamics SimulatorXcompact3d Incompact3dTimed Linux Kernel Compilation

EPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3eLogarithmic Per Watt Result OverviewPhoronix Test SuiteminiBUDEJohn The RipperNAS Parallel BenchmarksGROMACSASTC EncoderStockfishOpenSSLSpeedbLiquid-DSPASKAP7-Zip CompressionClickHouseRocksDBNumpy BenchmarkQuantLibLAMMPS Molecular Dynamics SimulatorGraphicsMagickWebP Image EncodeZstd CompressionXmrigCoremarkAlgebraic Multi-Grid BenchmarkHigh Performance Conjugate GradientMemcachedSecureMarksrsRAN ProjectminiFEP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.MP.W.G.M

AMD 5th Gen EPYC 9005 Turin vs. NVIDIA GH200 Performance Benchmarksopenvkl: vklBenchmarkCPU ISPCrelion: Basic - CPUnwchem: C240 Buckyballhpcg: 144 144 144 - 60openssl: RSA4096openssl: RSA4096speedb: Read While Writingbrl-cad: VGR Performance Metricsecuremark: SecureMark-TLSxmrig: GhostRider - 1Mstockfish: Chess Benchmarkbuild-linux-kernel: allmodconfigincompact3d: X3D-benchmarking input.i3dluxcorerender: Orange Juice - CPUopenssl: SHA512luxcorerender: DLSC - CPUopenssl: AES-128-GCMopenssl: SHA256openssl: AES-256-GCMopenssl: ChaCha20openssl: ChaCha20-Poly1305ospray: particle_volume/scivis/real_timetensorflow: CPU - 512 - ResNet-50build-nodejs: Time To Compileclickhouse: 100M Rows Hits Dataset, Third Runclickhouse: 100M Rows Hits Dataset, Second Runclickhouse: 100M Rows Hits Dataset, First Run / Cold Cachelammps: 20k Atomsbuild-gem5: Time To Compileospray: particle_volume/ao/real_timebuild-llvm: Ninjaffmpeg: libx265 - Platformopenradioss: Chrysler Neon 1Mopenfoam: drivaerFastback, Medium Mesh Size - Execution Timeopenfoam: drivaerFastback, Medium Mesh Size - Mesh Timeaskap: tConvolve MPI - Griddingaskap: tConvolve MPI - Degriddingnumpy: blender: Barbershop - CPU-Onlybuild-godot: Time To Compileopenradioss: INIVOL and Fluid Structure Interaction Drop Containertensorflow: CPU - 256 - ResNet-50qmcpack: Li2_STO_aecompress-zstd: 19, Long Mode - Decompression Speedcompress-zstd: 19, Long Mode - Compression Speedospray-studio: 1 - 4K - 16 - Path Tracer - CPUospray: gravity_spheres_volume/dim_512/pathtracer/real_timecompress-zstd: 19 - Decompression Speedcompress-zstd: 19 - Compression Speedavifenc: 0memcached: 1:100openvino: Face Detection FP16-INT8 - CPUopenvino: Face Detection FP16-INT8 - CPUopenvino: Noise Suppression Poconet-Like FP16 - CPUopenvino: Noise Suppression Poconet-Like FP16 - CPUopenvino: Person Detection FP16 - CPUopenvino: Person Detection FP16 - CPUospray-studio: 3 - 4K - 1 - Path Tracer - CPUopenvino: Machine Translation EN To DE FP16 - CPUopenvino: Machine Translation EN To DE FP16 - CPUopenvino: Person Vehicle Bike Detection FP16 - CPUopenvino: Person Vehicle Bike Detection FP16 - CPUopenvino: Road Segmentation ADAS FP16-INT8 - CPUopenvino: Road Segmentation ADAS FP16-INT8 - CPUopenvino: Face Detection Retail FP16-INT8 - CPUopenvino: Face Detection Retail FP16-INT8 - CPUopenvino: Age Gender Recognition Retail 0013 FP16-INT8 - CPUopenvino: Age Gender Recognition Retail 0013 FP16-INT8 - CPUopenvino: Vehicle Detection FP16-INT8 - CPUopenvino: Vehicle Detection FP16-INT8 - CPUospray-studio: 1 - 4K - 1 - Path Tracer - CPUopenvino: Weld Porosity Detection FP16-INT8 - CPUopenvino: Weld Porosity Detection FP16-INT8 - CPUgraphics-magick: Sharpenspeedb: Rand Readgraphics-magick: Enhancedgraphics-magick: Noise-Gaussianliquid-dsp: 1 - 256 - 512rocksdb: Rand Readnamd: STMV with 1,066,628 Atomsospray-studio: 3 - 4K - 32 - Path Tracer - CPUospray-studio: 3 - 4K - 16 - Path Tracer - CPUhelsing: 14 digitospray-studio: 1 - 4K - 32 - Path Tracer - CPUquantlib: Multi-Threadedminibude: OpenMP - BM2minibude: OpenMP - BM2namd: ATPase with 327,506 Atomswebp: Quality 100, Lossless, Highest Compressionblender: Pabellon Barcelona - CPU-Onlynpb: SP.Cliquid-dsp: 256 - 256 - 512john-the-ripper: WPA PSKliquid-dsp: 256 - 256 - 57liquid-dsp: 128 - 256 - 512liquid-dsp: 256 - 256 - 32john-the-ripper: Blowfishjohn-the-ripper: bcryptliquid-dsp: 1 - 256 - 32blender: Classroom - CPU-Onlycompress-7zip: Decompression Ratingcompress-7zip: Compression Ratingastcenc: Exhaustiveopenradioss: Cell Phone Drop Testastcenc: Very Thoroughamg: primesieve: 1e13npb: LU.Csrsran: PUSCH Processor Benchmark, Throughput Totalgromacs: MPI CPU - water_GMX50_barecloverleaf: clover_bm64_shortnpb: IS.Dsrsran: PUSCH Processor Benchmark, Throughput Threadgpaw: Carbon Nanotubepennant: sedovbigcoremark: CoreMark Size 666 - Iterations Per Secondopenfoam: drivaerFastback, Small Mesh Size - Mesh Timesvt-av1: Preset 4 - Bosphorus 4Kopenfoam: drivaerFastback, Small Mesh Size - Execution Timeuvg266: Bosphorus 4K - Slowuvg266: Bosphorus 4K - Mediumnpb: CG.Cblender: Junkshop - CPU-Onlypybench: Total For Average Test Timesblender: Fishy Cat - CPU-Onlykvazaar: Bosphorus 4K - Slowblender: BMW27 - CPU-Onlykvazaar: Bosphorus 4K - Mediumsvt-av1: Preset 13 - Bosphorus 4Koidn: RTLightmap.hdr.4096x4096 - CPU-Onlyastcenc: Thoroughembree: Pathtracer ISPC - Asian Dragon Objpennant: leblancbigsvt-av1: Preset 8 - Bosphorus 4Kuvg266: Bosphorus 4K - Very Fastminife: Smalltensorflow: CPU - 256 - AlexNetoidn: RT.hdr_alb_nrm.3840x2160 - CPU-Onlynpb: MG.Coidn: RT.ldr_alb_nrm.3840x2160 - CPU-Onlyembree: Pathtracer ISPC - Crownembree: Pathtracer ISPC - Asian Dragonlammps: Rhodopsin Proteinsvt-av1: Preset 12 - Bosphorus 4Kprimesieve: 1e12compress-pbzip2: FreeBSD-13.0-RELEASE-amd64-memstick.img CompressionEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e4241169.698159460.70373351713.385251.023940015834839036553315447.6362007178248.285196.31866535.228750920617324.02251276003091322361364029022832805930231551003421237105314145266361.7737247.22117.822598.32587.17576.7666.773140.59361.687399.43360.1192.04176.39806160.6828744845.246390.5723.8174.2491.05377.76201.1294.5911469.09.29853064.80001520.519.453.2659290017.46414.32231.1410.8810251.07149.81640.0963278.071227.238.6311033.6127.343497.174.9737263.740.73208220.837.5212591.545358.3822159.64445969890739495296205277509583540645.03185201491006739.76517091566481.4428.04710701.17519.899360.6125.67152747.87318163333317746678185800000228170000080642666673860823864743989600020.311001858100715011.531222.0318.7713297634500015.976313412.108940.923.14224.117503.01129.627.5603.8069654247085.81199310.01525.33694550.9655.6951319.4611.2664611.0268.618.2068.82244.5753.22128.7562230.97362.379513104.63870.4865294.52085.026.73112588.996.74210.4002267.698655.019248.6971.3960.8912382388223.8071225.763.90051673091.245479.710783281140649049145111601.9182135967203.796217.38342322.264556808188715.361289299058233113637325283119369924499073306028605349926015231036.4455254.02133.857809.87809.27779.3154.786130.79936.4032110.52979.53109.72233.115195.0910648448.043732.6962.28138.3683.43589.66234.6872.9331970.811.81617435.65232081.225.540.85113808558.42225.86141.438.387401.3149.09650.80120341.94761.695.525717.1914.412198.482.8521512.070.25173795.434.247427.2710154.613533.20301526671792336291282746675057335253.80313413241919661.82735274297665.1244.0726101.79313.335160.8345.75148567.2518179333338577025442600000182293333352658666671989541990785380166738.215270376552416.203517.6410.1197319761125025.397303092.494804.714.65121.127043.18224.529.8746.2211083884226.12849712.45725.6042428.8232.6065470.2319.8448219.1240.3214.6341.30283.7022.4072.3773117.90443.238866121.05182.8969114.42050.635.04159641.785.06111.2850137.408442.850294.4112.1231.2634081214.742.32251055054.515332.5102410422668994015.598146858246.660240.2858024949067647860912252127086724440007515680506310149255169543108427679423202.937596.49594.01581.5958.143166.097168.969362.2671283.97750910933.815401.6699.30123.99669.6091305.98.641266.615.2108.2133049705.542708.946.5690.20199.48549.2532.71161.98110.9832.89547.01351.9051.0734.66519.023.674896.13117.25153.4738.63465.82428569021111369397345873356833946356.972261141.554.6561366.3950.5424153.1525363000076130266973333324826333345024000007475674749607903334246434366004.06606.5885225645466731.46444257.151848.26.00228.411931.21756.4303042396678.99625216.7391737.37380627521.8772446.41204.47377946786.363701.2259.8412.6671.682797OpenBenchmarking.org

OpenVKL

OpenVKL is the Intel Open Volume Kernel Library that offers high-performance volume computation kernels and part of the Intel oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgItems / Sec, More Is BetterOpenVKL 2.0.0Benchmark: vklBenchmarkCPU ISPCEPYC 9575FEPYC 99659001800270036004500SE +/- 0.67, N = 3SE +/- 0.33, N = 323884241MIN: 187 / MAX: 30690MIN: 349 / MAX: 46699
OpenBenchmarking.orgItems / Sec Per Watt, More Is BetterOpenVKL 2.0.0Benchmark: vklBenchmarkCPU ISPCEPYC 9575FEPYC 996536912156.72212.597
OpenBenchmarking.orgWatts, Fewer Is BetterOpenVKL 2.0.0CPU Power Consumption MonitorEPYC 9575FEPYC 996570140210280350Min: 38.22 / Avg: 355.24 / Max: 401.22Min: 38.06 / Avg: 336.66 / Max: 409.431. EPYC 9575F: Approximate power consumption of 240263 Joules per run.2. EPYC 9965: Approximate power consumption of 228705 Joules per run.

RELION

RELION - REgularised LIkelihood OptimisatioN - is a stand-alone computer program for Maximum A Posteriori refinement of (multiple) 3D reconstructions or 2D class averages in cryo-electron microscopy (cryo-EM). It is developed in the research group of Sjors Scheres at the MRC Laboratory of Molecular Biology. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterRELION 4.0.1Test: Basic - Device: CPUEPYC 9575FEPYC 996550100150200250SE +/- 3.85, N = 9SE +/- 3.21, N = 12223.81169.701. (CXX) g++ options: -fopenmp -std=c++11 -O3 -rdynamic -ldl -ltiff -lfftw3f -lfftw3 -lpng -ljpeg -lmpi_cxx -lmpi
OpenBenchmarking.orgWatts, Fewer Is BetterRELION 4.0.1CPU Power Consumption MonitorEPYC 9575FEPYC 996570140210280350Min: 38.25 / Avg: 389.21 / Max: 399.75Min: 38.2 / Avg: 375.69 / Max: 395.471. EPYC 9575F: Approximate power consumption of 88092 Joules per run.2. EPYC 9965: Approximate power consumption of 64869 Joules per run.

NWChem

NWChem is an open-source high performance computational chemistry package. Per NWChem's documentation, "NWChem aims to provide its users with computational chemistry tools that are scalable both in their ability to treat large scientific computational chemistry problems efficiently, and in their use of available parallel computing resources from high-performance parallel supercomputers to conventional workstation clusters." Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterNWChem 7.0.2Input: C240 BuckyballEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e300600900120015001594.01225.71214.7-m64-m641. (F9X) gfortran options: -lnwctask -lccsd -lmcscf -lselci -lmp2 -lmoints -lstepper -ldriver -loptim -lnwdft -lgradients -lcphf -lesp -lddscf -ldangchang -lguess -lhessian -lvib -lnwcutil -lrimp2 -lproperty -lsolvation -lnwints -lprepar -lnwmd -lnwpw -lofpw -lpaw -lpspw -lband -lnwpwlib -lcafe -lspace -lanalyze -lqhop -lpfft -ldplot -ldrdy -lvscf -lqmmm -lqmd -letrans -ltce -lbq -lmm -lcons -lperfm -ldntmc -lccca -ldimqm -lga -larmci -lpeigs -l64to32 -lopenblas -lpthread -lrt -llapack -lnwcblas -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz -lcomex -ffast-math -std=legacy -fdefault-integer-8 -finline-functions -O2
OpenBenchmarking.orgWatts, Fewer Is BetterNWChem 7.0.2CPU Power Consumption MonitorEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e70140210280350Min: 38.3 / Avg: 392.8 / Max: 400.06Min: 38.12 / Avg: 387.9 / Max: 409.94Min: 41.49 / Avg: 226.74 / Max: 281.44

High Performance Conjugate Gradient

HPCG is the High Performance Conjugate Gradient and is a new scientific benchmark from Sandia National Lans focused for super-computer testing with modern real-world workloads compared to HPCC. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOP/s, More Is BetterHigh Performance Conjugate Gradient 3.1X Y Z: 144 144 144 - RT: 60NVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F1428425670SE +/- 0.14, N = 3SE +/- 0.02, N = 3SE +/- 0.04, N = 342.3260.7063.901. (CXX) g++ options: -O3 -ffast-math -ftree-vectorize -lmpi_cxx -lmpi
OpenBenchmarking.orgGFLOP/s Per Watt, More Is BetterHigh Performance Conjugate Gradient 3.1X Y Z: 144 144 144 - RT: 60EPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e0.05380.10760.16140.21520.2690.1630.1760.239
MinAvgMaxEPYC 9575F38.3391.5399.9EPYC 996538.2345.0387.5NVIDIA GH200 144G HBM3e40.6177.3218.6OpenBenchmarking.orgWatts, Fewer Is BetterHigh Performance Conjugate Gradient 3.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 99831 Joules per run.2. EPYC 9965: Approximate power consumption of 215367 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 59263 Joules per run.

OpenSSL

OpenSSL is an open-source toolkit that implements SSL (Secure Sockets Layer) and TLS (Transport Layer Security) protocols. This test profile makes use of the built-in "openssl speed" benchmarking capabilities. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgverify/s, More Is BetterOpenSSL 3.3Algorithm: RSA4096NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965700K1400K2100K2800K3500KSE +/- 1377.88, N = 3SE +/- 743.25, N = 3SE +/- 310.57, N = 31055054.51673091.23351713.3-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenBenchmarking.orgverify/s Per Watt, More Is BetterOpenSSL 3.3Algorithm: RSA4096EPYC 9575FNVIDIA GH200 144G HBM3eEPYC 99652K4K6K8K10K4395.325640.208963.31
MinAvgMaxEPYC 9575F38.5380.7400.8EPYC 996538.3373.9396.9NVIDIA GH200 144G HBM3e39.6187.1196.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenSSL 3.3CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 116607 Joules per run.2. EPYC 9965: Approximate power consumption of 116045 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 57864 Joules per run.

OpenBenchmarking.orgsign/s, More Is BetterOpenSSL 3.3Algorithm: RSA4096NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996520K40K60K80K100KSE +/- 20.10, N = 3SE +/- 116.72, N = 3SE +/- 46.31, N = 315332.545479.785251.0-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl

Speedb

Speedb is a next-generation key value storage engine that is RocksDB compatible and aiming for stability, efficiency, and performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterSpeedb 2.7Test: Read While WritingNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99655M10M15M20M25MSE +/- 87939.63, N = 15SE +/- 91725.44, N = 15SE +/- 505667.36, N = 15102410421078328123940015-lpthread-lpthread1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti
OpenBenchmarking.orgOp/s Per Watt, More Is BetterSpeedb 2.7Test: Read While WritingEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 996516K32K48K64K80K28692.8256303.8476045.35
MinAvgMaxEPYC 9575F38.5375.8400.8EPYC 996538.7314.8417.4NVIDIA GH200 144G HBM3e43.4181.9215.8OpenBenchmarking.orgWatts, Fewer Is BetterSpeedb 2.7CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 23977 Joules per run.2. EPYC 9965: Approximate power consumption of 20148 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 11702 Joules per run.

BRL-CAD

BRL-CAD is a cross-platform, open-source solid modeling system with built-in benchmark mode. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgVGR Performance Metric, More Is BetterBRL-CAD 7.38.2VGR Performance MetricEPYC 9575FEPYC 99652M4M6M8M10M140649083483901. (CXX) g++ options: -std=c++17 -pipe -fvisibility=hidden -fno-strict-aliasing -fno-common -fexceptions -ftemplate-depth-128 -m64 -ggdb3 -O3 -fipa-pta -fstrength-reduce -finline-functions -flto -ltcl8.6 -lnetpbm -lregex_brl -lz_brl -lassimp -ldl -lm -ltk8.6
OpenBenchmarking.orgVGR Performance Metric Per Watt, More Is BetterBRL-CAD 7.38.2VGR Performance MetricEPYC 9575FEPYC 996520K40K60K80K100K3698.5783197.80
MinAvgMaxEPYC 9575F38.4380.3401.3EPYC 996538.4100.3250.2OpenBenchmarking.orgWatts, Fewer Is BetterBRL-CAD 7.38.2CPU Power Consumption Monitor110220330440550

VGR Performance Metric

NVIDIA GH200 144G HBM3e: The test quit with a non-zero exit status. E: ERROR: Could not find the BRL-CAD raytracer

SecureMark

SecureMark is an objective, standardized benchmarking framework for measuring the efficiency of cryptographic processing solutions developed by EEMBC. SecureMark-TLS is benchmarking Transport Layer Security performance with a focus on IoT/edge computing. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgmarks, More Is BetterSecureMark 1.0.4Benchmark: SecureMark-TLSNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F110K220K330K440K550KSE +/- 785.55, N = 3SE +/- 1741.58, N = 3SE +/- 602.41, N = 32668993655334914511. (CC) gcc options: -pedantic -O3
OpenBenchmarking.orgmarks Per Watt, More Is BetterSecureMark 1.0.4Benchmark: SecureMark-TLSEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e140028004200560070004523.075300.986404.02
MinAvgMaxEPYC 9575F38.192.796.6EPYC 996537.780.882.7NVIDIA GH200 144G HBM3e38.941.744.4OpenBenchmarking.orgWatts, Fewer Is BetterSecureMark 1.0.4CPU Power Consumption Monitor204060801001. EPYC 9575F: Approximate power consumption of 26422 Joules per run.2. EPYC 9965: Approximate power consumption of 22925 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 12017 Joules per run.

Xmrig

Xmrig is an open-source cross-platform CPU/GPU miner for RandomX, KawPow, CryptoNight and AstroBWT. This test profile is setup to measure the Xmrig CPU mining performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgH/s, More Is BetterXmrig 6.21Variant: GhostRider - Hash Count: 1MNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99653K6K9K12K15KSE +/- 0.17, N = 3SE +/- 508.75, N = 15SE +/- 53.68, N = 34015.511601.915447.6-maes-maes1. (CXX) g++ options: -fexceptions -fno-rtti -O3 -Ofast -static-libgcc -static-libstdc++ -rdynamic -lssl -lcrypto -luv -lpthread -lrt -ldl -lhwloc
OpenBenchmarking.orgH/s Per Watt, More Is BetterXmrig 6.21Variant: GhostRider - Hash Count: 1MNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965112233445528.6930.5248.04
OpenBenchmarking.orgWatts, Fewer Is BetterXmrig 6.21CPU Power Consumption MonitorEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e70140210280350Min: 38.21 / Avg: 380.2 / Max: 401.2Min: 38.1 / Avg: 321.57 / Max: 344.6Min: 41.25 / Avg: 139.96 / Max: 146.691. EPYC 9575F: Approximate power consumption of 35486 Joules per run.2. EPYC 9965: Approximate power consumption of 22295 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 35502 Joules per run.

Stockfish

This is a test of Stockfish, an advanced open-source C++11 chess benchmark that can scale up to 1024 CPU threads. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgNodes Per Second, More Is BetterStockfish 16.1Chess BenchmarkNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996580M160M240M320M400MSE +/- 1172057.21, N = 15SE +/- 2577463.33, N = 15SE +/- 2768351.23, N = 398146858182135967362007178-m64 -msse -msse3 -mpopcnt -mavx2 -mbmi -mavx512f -mavx512bw -mavx512vnni -mavx512dq -mavx512vl -msse4.1 -mssse3 -msse2 -mbmi2-m64 -msse -msse3 -mpopcnt -mavx2 -mbmi -mavx512f -mavx512bw -mavx512vnni -mavx512dq -mavx512vl -msse4.1 -mssse3 -msse2 -mbmi21. (CXX) g++ options: -lgcov -lpthread -fno-exceptions -std=c++17 -fno-peel-loops -fno-tracer -pedantic -O3 -funroll-loops -flto -flto-partition=one -flto=jobserver
OpenBenchmarking.orgNodes Per Second Per Watt, More Is BetterStockfish 16.1Chess BenchmarkNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965200K400K600K800K1000K375426.89490782.75931347.77
MinAvgMaxEPYC 996538.5388.7420.8EPYC 9575F38.7371.1401.2NVIDIA GH200 144G HBM3e42.7261.4290.6OpenBenchmarking.orgWatts, Fewer Is BetterStockfish 16.1CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 35889 Joules per run.2. EPYC 9575F: Approximate power consumption of 24988 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 18038 Joules per run.

Timed Linux Kernel Compilation

This test times how long it takes to build the Linux kernel in a default configuration (defconfig) for the architecture being tested or alternatively an allmodconfig for building all possible kernel modules for the build. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Linux Kernel Compilation 6.8Build: allmodconfigEPYC 9965NVIDIA GH200 144G HBM3eEPYC 9575F50100150200250SE +/- 1.72, N = 3SE +/- 0.59, N = 3SE +/- 0.48, N = 3248.29246.66203.80
MinAvgMaxEPYC 9575F38.5352.3401.2EPYC 996538.2307.8388.2NVIDIA GH200 144G HBM3e38.9229.8270.8OpenBenchmarking.orgWatts, Fewer Is BetterTimed Linux Kernel Compilation 6.8CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 74930 Joules per run.2. EPYC 9965: Approximate power consumption of 79520 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 59825 Joules per run.

Xcompact3d Incompact3d

Xcompact3d Incompact3d is a Fortran-MPI based, finite difference high-performance code for solving the incompressible Navier-Stokes equation and as many as you need scalar transport equations. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterXcompact3d Incompact3d 2021-03-11Input: X3D-benchmarking input.i3dNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996550100150200250SE +/- 0.29, N = 3SE +/- 1.26, N = 3SE +/- 0.30, N = 3240.29217.38196.321. (F9X) gfortran options: -cpp -O2 -funroll-loops -floop-optimize -fcray-pointer -fbacktrace -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz
MinAvgMaxEPYC 9575F38.2389.8399.8EPYC 996538.4357.8389.5NVIDIA GH200 144G HBM3e42.2172.9253.2OpenBenchmarking.orgWatts, Fewer Is BetterXcompact3d Incompact3d 2021-03-11CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 86804 Joules per run.2. EPYC 9965: Approximate power consumption of 73462 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 43344 Joules per run.

LuxCoreRender

LuxCoreRender is an open-source 3D physically based renderer formerly known as LuxRender. LuxCoreRender supports CPU-based rendering as well as GPU acceleration via OpenCL, NVIDIA CUDA, and NVIDIA OptiX interfaces. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgM samples/sec, More Is BetterLuxCoreRender 2.6Scene: Orange Juice - Acceleration: CPUEPYC 9575FEPYC 9965816243240SE +/- 0.25, N = 4SE +/- 0.69, N = 1522.2635.22MIN: 19.56 / MAX: 27.46MIN: 29.07 / MAX: 50.81
OpenBenchmarking.orgM samples/sec Per Watt, More Is BetterLuxCoreRender 2.6Scene: Orange Juice - Acceleration: CPUEPYC 9575FEPYC 99650.02210.04420.06630.08840.11050.0610.098
MinAvgMaxEPYC 9575F38.2367.3401.1EPYC 996538.0361.2413.4OpenBenchmarking.orgWatts, Fewer Is BetterLuxCoreRender 2.6CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 24332 Joules per run.2. EPYC 9965: Approximate power consumption of 24198 Joules per run.

OpenSSL

OpenSSL is an open-source toolkit that implements SSL (Secure Sockets Layer) and TLS (Transport Layer Security) protocols. This test profile makes use of the built-in "openssl speed" benchmarking capabilities. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbyte/s, More Is BetterOpenSSL 3.3Algorithm: SHA512EPYC 9575FNVIDIA GH200 144G HBM3eEPYC 996520000M40000M60000M80000M100000MSE +/- 40355477.92, N = 3SE +/- 557471050.78, N = 4SE +/- 148724346.57, N = 3455680818874949067647887509206173-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenBenchmarking.orgbyte/s Per Watt, More Is BetterOpenSSL 3.3Algorithm: SHA512EPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e50M100M150M200M250M118475697.80235679194.64248755218.22
MinAvgMaxEPYC 9575F38.6384.6400.9EPYC 996538.5371.3391.7NVIDIA GH200 144G HBM3e41.1199.0243.4OpenBenchmarking.orgWatts, Fewer Is BetterOpenSSL 3.3CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 70385 Joules per run.2. EPYC 9965: Approximate power consumption of 68073 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 36707 Joules per run.

LuxCoreRender

LuxCoreRender is an open-source 3D physically based renderer formerly known as LuxRender. LuxCoreRender supports CPU-based rendering as well as GPU acceleration via OpenCL, NVIDIA CUDA, and NVIDIA OptiX interfaces. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgM samples/sec, More Is BetterLuxCoreRender 2.6Scene: DLSC - Acceleration: CPUEPYC 9575FEPYC 9965612182430SE +/- 0.19, N = 4SE +/- 0.37, N = 1515.3624.02MIN: 14.74 / MAX: 17.33MIN: 21.26 / MAX: 31.63
OpenBenchmarking.orgM samples/sec Per Watt, More Is BetterLuxCoreRender 2.6Scene: DLSC - Acceleration: CPUEPYC 9575FEPYC 99650.01490.02980.04470.05960.07450.0410.066
MinAvgMaxEPYC 9575F38.3374.7401.1EPYC 996538.2366.0402.7OpenBenchmarking.orgWatts, Fewer Is BetterLuxCoreRender 2.6CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 24264 Joules per run.2. EPYC 9965: Approximate power consumption of 23763 Joules per run.

OpenSSL

OpenSSL is an open-source toolkit that implements SSL (Secure Sockets Layer) and TLS (Transport Layer Security) protocols. This test profile makes use of the built-in "openssl speed" benchmarking capabilities. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbyte/s, More Is BetterOpenSSL 3.3Algorithm: AES-128-GCMNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965500000M1000000M1500000M2000000M2500000MSE +/- 121548246.52, N = 3SE +/- 1089753853.94, N = 3SE +/- 1643367333.46, N = 360912252127012892990582332512760030913-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenBenchmarking.orgbyte/s Per Watt, More Is BetterOpenSSL 3.3Algorithm: AES-128-GCMNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99651400M2800M4200M5600M7000M2139935369.453601051103.716477196397.29
MinAvgMaxEPYC 996538.5387.9419.8EPYC 9575F38.0358.0377.9NVIDIA GH200 144G HBM3e41.2284.6293.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenSSL 3.3CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 70993 Joules per run.2. EPYC 9575F: Approximate power consumption of 65640 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 52564 Joules per run.

OpenBenchmarking.orgbyte/s, More Is BetterOpenSSL 3.3Algorithm: SHA256NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996550000M100000M150000M200000M250000MSE +/- 97000421.20, N = 3SE +/- 38504291.58, N = 3SE +/- 865484415.00, N = 386724440007113637325283223613640290-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenBenchmarking.orgbyte/s Per Watt, More Is BetterOpenSSL 3.3Algorithm: SHA256EPYC 9575FNVIDIA GH200 144G HBM3eEPYC 9965130M260M390M520M650M295380807.41440489220.25591912038.17
MinAvgMaxEPYC 9575F38.5384.7400.8EPYC 996538.4377.8402.0NVIDIA GH200 144G HBM3e41.2196.9273.1OpenBenchmarking.orgWatts, Fewer Is BetterOpenSSL 3.3CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 70531 Joules per run.2. EPYC 9965: Approximate power consumption of 69134 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 36358 Joules per run.

OpenBenchmarking.orgbyte/s, More Is BetterOpenSSL 3.3Algorithm: AES-256-GCMNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965500000M1000000M1500000M2000000M2500000MSE +/- 82190365.25, N = 3SE +/- 947677989.11, N = 3SE +/- 3767964201.90, N = 351568050631011936992449902283280593023-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenBenchmarking.orgbyte/s Per Watt, More Is BetterOpenSSL 3.3Algorithm: AES-256-GCMNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99651300M2600M3900M5200M6500M1809281985.063325699800.965989469249.57
MinAvgMaxEPYC 996538.4381.2412.5EPYC 9575F38.4358.9382.8NVIDIA GH200 144G HBM3e43.5285.0293.6OpenBenchmarking.orgWatts, Fewer Is BetterOpenSSL 3.3CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 69763 Joules per run.2. EPYC 9575F: Approximate power consumption of 65685 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 52634 Joules per run.

OpenBenchmarking.orgbyte/s, More Is BetterOpenSSL 3.3Algorithm: ChaCha20NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965300000M600000M900000M1200000M1500000MSE +/- 3236084.95, N = 3SE +/- 24295011.71, N = 3SE +/- 312050224.99, N = 31492551695437330602860531551003421237-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenBenchmarking.orgbyte/s Per Watt, More Is BetterOpenSSL 3.3Algorithm: ChaCha20NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965800M1600M2400M3200M4000M768573496.681908650939.883660529028.05
MinAvgMaxEPYC 996538.5423.7461.6EPYC 9575F38.6384.1401.2NVIDIA GH200 144G HBM3e43.4194.2217.3OpenBenchmarking.orgWatts, Fewer Is BetterOpenSSL 3.3CPU Power Consumption Monitor1202403604806001. EPYC 9965: Approximate power consumption of 77539 Joules per run.2. EPYC 9575F: Approximate power consumption of 70413 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 35862 Joules per run.

OpenBenchmarking.orgbyte/s, More Is BetterOpenSSL 3.3Algorithm: ChaCha20-Poly1305NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965200000M400000M600000M800000M1000000MSE +/- 3328662.15, N = 3SE +/- 312635540.30, N = 3SE +/- 170992251.22, N = 31084276794234992601523101053141452663-m64-m641. (CC) gcc options: -pthread -O3 -lssl -lcrypto -ldl
OpenBenchmarking.orgbyte/s Per Watt, More Is BetterOpenSSL 3.3Algorithm: ChaCha20-Poly1305NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965500M1000M1500M2000M2500M535429561.191303373595.942506574135.73
MinAvgMaxEPYC 996538.8420.2459.2EPYC 9575F38.3383.1400.8NVIDIA GH200 144G HBM3e41.5202.5225.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenSSL 3.3CPU Power Consumption Monitor1202403604806001. EPYC 9965: Approximate power consumption of 77028 Joules per run.2. EPYC 9575F: Approximate power consumption of 70099 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 37464 Joules per run.

OSPRay

Intel OSPRay is a portable ray-tracing engine for high-performance, high-fidelity scientific visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgItems Per Second, More Is BetterOSPRay 3.1Benchmark: particle_volume/scivis/real_timeEPYC 9575FEPYC 99651428425670SE +/- 0.01, N = 3SE +/- 0.05, N = 336.4561.77
OpenBenchmarking.orgItems Per Second Per Watt, More Is BetterOSPRay 3.1Benchmark: particle_volume/scivis/real_timeEPYC 9575FEPYC 99650.04280.08560.12840.17120.2140.1050.190
MinAvgMaxEPYC 9575F38.3346.8394.5EPYC 996538.2325.9379.2OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay 3.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 65436 Joules per run.2. EPYC 9965: Approximate power consumption of 54541 Joules per run.

TensorFlow

This is a benchmark of the TensorFlow deep learning framework using the TensorFlow reference benchmarks (tensorflow/benchmarks with tf_cnn_benchmarks.py). Note with the Phoronix Test Suite there is also pts/tensorflow-lite for benchmarking the TensorFlow Lite binaries if desired for complementary metrics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 512 - Model: ResNet-50EPYC 9965EPYC 9575F60120180240300SE +/- 0.31, N = 3SE +/- 0.17, N = 3247.22254.02
OpenBenchmarking.orgimages/sec Per Watt, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 512 - Model: ResNet-50EPYC 9575FEPYC 99650.18430.36860.55290.73720.92150.7720.819
MinAvgMaxEPYC 9575F38.4329.2346.8EPYC 996538.2301.7329.7OpenBenchmarking.orgWatts, Fewer Is BetterTensorFlow 2.16.1CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 75493 Joules per run.2. EPYC 9965: Approximate power consumption of 71906 Joules per run.

Device: CPU - Batch Size: 512 - Model: ResNet-50

NVIDIA GH200 144G HBM3e: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'absl'

Timed Node.js Compilation

This test profile times how long it takes to build/compile Node.js itself from source. Node.js is a JavaScript run-time built from the Chrome V8 JavaScript engine while itself is written in C/C++. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Node.js Compilation 21.7.2Time To CompileNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99654080120160200SE +/- 0.15, N = 3SE +/- 0.26, N = 3SE +/- 0.70, N = 3202.94133.86117.82
MinAvgMaxEPYC 9575F38.6313.9401.2EPYC 996537.9257.7406.9NVIDIA GH200 144G HBM3e39.2181.0279.3OpenBenchmarking.orgWatts, Fewer Is BetterTimed Node.js Compilation 21.7.2CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 44045 Joules per run.2. EPYC 9965: Approximate power consumption of 32304 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 38253 Joules per run.

ClickHouse

ClickHouse is an open-source, high performance OLAP data management system. This test profile uses ClickHouse's standard benchmark recommendations per https://clickhouse.com/docs/en/operations/performance-test/ / https://github.com/ClickHouse/ClickBench/tree/main/clickhouse with the 100 million rows web analytics dataset. The reported value is the query processing time using the geometric mean of all separate queries performed as an aggregate. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgQueries Per Minute, Geo Mean, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, Third RunNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F2004006008001000SE +/- 2.50, N = 3SE +/- 2.05, N = 3SE +/- 1.25, N = 3596.49598.32809.87MIN: 59.29 / MAX: 7500MIN: 91.19 / MAX: 6000MIN: 65.43 / MAX: 10000
OpenBenchmarking.orgQueries Per Minute, Geo Mean Per Watt, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, Third RunEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e2468104.1814.2027.980
MinAvgMaxEPYC 9575F38.3193.7368.0EPYC 996538.2142.4305.3NVIDIA GH200 144G HBM3e40.074.7282.8OpenBenchmarking.orgWatts, Fewer Is BetterClickHouse 22.12.3.5CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 24083 Joules per run.2. EPYC 9965: Approximate power consumption of 23304 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 12558 Joules per run.

OpenBenchmarking.orgQueries Per Minute, Geo Mean, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, Second RunEPYC 9965NVIDIA GH200 144G HBM3eEPYC 9575F2004006008001000SE +/- 4.38, N = 3SE +/- 2.22, N = 3SE +/- 3.75, N = 3587.17594.01809.27MIN: 92.02 / MAX: 6000MIN: 59.29 / MAX: 7500MIN: 65.65 / MAX: 8571.43

OpenBenchmarking.orgQueries Per Minute, Geo Mean, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, First Run / Cold CacheEPYC 9965NVIDIA GH200 144G HBM3eEPYC 9575F2004006008001000SE +/- 3.03, N = 3SE +/- 2.14, N = 3SE +/- 3.67, N = 3576.76581.59779.31MIN: 89.55 / MAX: 5454.55MIN: 58.94 / MAX: 6666.67MIN: 66.01 / MAX: 7500

LAMMPS Molecular Dynamics Simulator

LAMMPS is a classical molecular dynamics code, and an acronym for Large-scale Atomic/Molecular Massively Parallel Simulator. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgns/day, More Is BetterLAMMPS Molecular Dynamics Simulator 23Jun2022Model: 20k AtomsEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 99651530456075SE +/- 0.37, N = 3SE +/- 0.14, N = 3SE +/- 0.46, N = 354.7958.1466.771. (CXX) g++ options: -O3 -lm -ldl
OpenBenchmarking.orgns/day Per Watt, More Is BetterLAMMPS Molecular Dynamics Simulator 23Jun2022Model: 20k AtomsEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e0.04770.09540.14310.19080.23850.1410.1680.212
MinAvgMaxEPYC 996537.8396.7422.8EPYC 9575F38.4388.8399.6NVIDIA GH200 144G HBM3e40.1274.7292.0OpenBenchmarking.orgWatts, Fewer Is BetterLAMMPS Molecular Dynamics Simulator 23Jun2022CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 54084 Joules per run.2. EPYC 9575F: Approximate power consumption of 62864 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 42483 Joules per run.

Timed Gem5 Compilation

This test times how long it takes to compile Gem5. Gem5 is a simulator for computer system architecture research. Gem5 is widely used for computer architecture research within the industry, academia, and more. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Gem5 Compilation 23.0.1Time To CompileNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F4080120160200SE +/- 0.26, N = 3SE +/- 1.20, N = 3SE +/- 0.56, N = 3166.10140.59130.80
MinAvgMaxEPYC 9575F38.5211.8401.3EPYC 996537.8163.6402.4NVIDIA GH200 144G HBM3e39.4118.8239.4OpenBenchmarking.orgWatts, Fewer Is BetterTimed Gem5 Compilation 23.0.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 28661 Joules per run.2. EPYC 9965: Approximate power consumption of 23835 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 20441 Joules per run.

OSPRay

Intel OSPRay is a portable ray-tracing engine for high-performance, high-fidelity scientific visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgItems Per Second, More Is BetterOSPRay 3.1Benchmark: particle_volume/ao/real_timeEPYC 9575FEPYC 99651428425670SE +/- 0.01, N = 3SE +/- 0.02, N = 336.4061.69
OpenBenchmarking.orgItems Per Second Per Watt, More Is BetterOSPRay 3.1Benchmark: particle_volume/ao/real_timeEPYC 9575FEPYC 99650.03830.07660.11490.15320.19150.0950.170
MinAvgMaxEPYC 9575F38.5381.7396.3EPYC 996538.4362.8379.4OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay 3.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 54580 Joules per run.2. EPYC 9965: Approximate power consumption of 43780 Joules per run.

Timed LLVM Compilation

This test times how long it takes to compile/build the LLVM compiler stack. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed LLVM Compilation 16.0Build System: NinjaNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99654080120160200SE +/- 0.09, N = 3SE +/- 0.08, N = 3SE +/- 0.77, N = 3168.97110.5399.43
MinAvgMaxEPYC 9575F38.4313.1401.2EPYC 996538.2254.0404.4NVIDIA GH200 144G HBM3e39.3167.1239.8OpenBenchmarking.orgWatts, Fewer Is BetterTimed LLVM Compilation 16.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 38199 Joules per run.2. EPYC 9965: Approximate power consumption of 28707 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 32140 Joules per run.

FFmpeg

This is a benchmark of the FFmpeg multimedia framework. The FFmpeg test profile is making use of a modified version of vbench from Columbia University's Architecture and Design Lab (ARCADE) [http://arcade.cs.columbia.edu/vbench/] that is a benchmark for video-as-a-service workloads. The test profile offers the options of a range of vbench scenarios based on freely distributable video content and offers the options of using the x264 or x265 video encoders for transcoding. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterFFmpeg 7.0Encoder: libx265 - Scenario: PlatformEPYC 9965EPYC 9575F20406080100SE +/- 0.06, N = 3SE +/- 0.09, N = 360.1179.531. (CXX) g++ options: -O3 -rdynamic -lpthread -lrt -ldl -lnuma
OpenBenchmarking.orgFPS Per Watt, More Is BetterFFmpeg 7.0Encoder: libx265 - Scenario: PlatformEPYC 9575FEPYC 99650.11430.22860.34290.45720.57150.4920.508
MinAvgMaxEPYC 9575F38.3161.8226.0EPYC 996537.9118.2149.3OpenBenchmarking.orgWatts, Fewer Is BetterFFmpeg 7.0CPU Power Consumption Monitor601201802403001. EPYC 9575F: Approximate power consumption of 18122 Joules per run.2. EPYC 9965: Approximate power consumption of 17383 Joules per run.

OpenRadioss

OpenRadioss is an open-source AGPL-licensed finite element solver for dynamic event analysis OpenRadioss is based on Altair Radioss and open-sourced in 2022. This open-source finite element solver is benchmarked with various example models available from https://www.openradioss.org/models/ and https://github.com/OpenRadioss/ModelExchange/tree/main/Examples. This test is currently using a reference OpenRadioss binary build offered via GitHub. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterOpenRadioss 2023.09.15Model: Chrysler Neon 1MEPYC 9575FEPYC 996520406080100SE +/- 0.09, N = 3SE +/- 0.41, N = 3109.7292.04
MinAvgMaxEPYC 9575F38.7350.4400.0EPYC 996538.4300.9408.4OpenBenchmarking.orgWatts, Fewer Is BetterOpenRadioss 2023.09.15CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 45079 Joules per run.2. EPYC 9965: Approximate power consumption of 38413 Joules per run.

OpenFOAM

OpenFOAM is the leading free, open-source software for computational fluid dynamics (CFD). This test profile currently uses the drivaerFastback test case for analyzing automotive aerodynamics or alternatively the older motorBike input. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterOpenFOAM 10Input: drivaerFastback, Medium Mesh Size - Execution TimeNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996580160240320400362.27233.12176.40-mcpu=native -lfoamToVTK -ldynamicMesh -lfileFormats-m64 -lfiniteVolume -lmeshTools -lparallel -lregionModels-m64 -lfiniteVolume -lmeshTools -lparallel -lregionModels1. (CXX) g++ options: -std=c++14 -O3 -ftemplate-depth-100 -fPIC -fuse-ld=bfd -Xlinker --add-needed --no-as-needed -llagrangian -lgenericPatchFields -lOpenFOAM -ldl -lm
MinAvgMaxEPYC 9575F38.2388.3399.8EPYC 996538.1377.5408.6NVIDIA GH200 144G HBM3e41.1200.9291.9OpenBenchmarking.orgWatts, Fewer Is BetterOpenFOAM 10CPU Power Consumption Monitor110220330440550

OpenBenchmarking.orgSeconds, Fewer Is BetterOpenFOAM 10Input: drivaerFastback, Medium Mesh Size - Mesh TimeEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e4080120160200160.6895.0983.98-m64 -lfiniteVolume -lmeshTools -lparallel -lregionModels-m64 -lfiniteVolume -lmeshTools -lparallel -lregionModels-mcpu=native -lfoamToVTK -ldynamicMesh -lfileFormats1. (CXX) g++ options: -std=c++14 -O3 -ftemplate-depth-100 -fPIC -fuse-ld=bfd -Xlinker --add-needed --no-as-needed -llagrangian -lgenericPatchFields -lOpenFOAM -ldl -lm

ASKAP

ASKAP is a set of benchmarks from the Australian SKA Pathfinder. The principal ASKAP benchmarks are the Hogbom Clean Benchmark (tHogbomClean) and Convolutional Resamping Benchmark (tConvolve) as well as some previous ASKAP benchmarks being included as well for OpenCL and CUDA execution of tConvolve. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMpix/sec, More Is BetterASKAP 1.0Test: tConvolve MPI - GriddingNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F10K20K30K40K50KSE +/- 58.46, N = 3SE +/- 526.44, N = 13SE +/- 369.83, N = 310933.844845.248448.01. (CXX) g++ options: -O3 -fstrict-aliasing -fopenmp
OpenBenchmarking.orgMpix/sec Per Watt, More Is BetterASKAP 1.0Test: tConvolve MPI - GriddingNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F30609012015072.88144.36146.79
MinAvgMaxEPYC 9575F38.3330.0400.0EPYC 996538.2310.7396.7NVIDIA GH200 144G HBM3e41.4150.0265.8OpenBenchmarking.orgWatts, Fewer Is BetterASKAP 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 8141 Joules per run.2. EPYC 9965: Approximate power consumption of 20670 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 7052 Joules per run.

OpenBenchmarking.orgMpix/sec, More Is BetterASKAP 1.0Test: tConvolve MPI - DegriddingNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996510K20K30K40K50KSE +/- 33.40, N = 3SE +/- 0.00, N = 3SE +/- 455.47, N = 1315401.643732.646390.51. (CXX) g++ options: -O3 -fstrict-aliasing -fopenmp

Numpy Benchmark

This is a test to obtain the general Numpy performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgScore, More Is BetterNumpy BenchmarkNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F2004006008001000SE +/- 1.78, N = 3SE +/- 0.78, N = 3SE +/- 3.50, N = 3699.30723.81962.28
OpenBenchmarking.orgScore Per Watt, More Is BetterNumpy BenchmarkEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e481216208.87310.29116.277
MinAvgMaxEPYC 9575F38.193.5100.9EPYC 996538.081.690.0NVIDIA GH200 144G HBM3e38.743.047.7OpenBenchmarking.orgWatts, Fewer Is BetterNumpy BenchmarkCPU Power Consumption Monitor204060801001. EPYC 9575F: Approximate power consumption of 9070 Joules per run.2. EPYC 9965: Approximate power consumption of 10468 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 5184 Joules per run.

Blender

Blender is an open-source 3D creation and modeling software project. This test is of Blender's Cycles performance with various sample files. GPU computing via NVIDIA OptiX and NVIDIA CUDA is currently supported as well as HIP for AMD Radeon GPUs and Intel oneAPI for Intel Graphics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 4.1Blend File: Barbershop - Compute: CPU-OnlyEPYC 9575FEPYC 9965306090120150SE +/- 0.10, N = 3SE +/- 0.08, N = 3138.3674.24
MinAvgMaxEPYC 9575F38.3378.8400.2EPYC 996538.2366.0403.4OpenBenchmarking.orgWatts, Fewer Is BetterBlender 4.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 54175 Joules per run.2. EPYC 9965: Approximate power consumption of 29277 Joules per run.

Timed Godot Game Engine Compilation

This test times how long it takes to compile the Godot Game Engine. Godot is a popular, open-source, cross-platform 2D/3D game engine and is built using the SCons build system and targeting the X11 platform. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Godot Game Engine Compilation 4.0Time To CompileNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F306090120150SE +/- 0.28, N = 3SE +/- 0.19, N = 3SE +/- 0.37, N = 3124.0091.0583.44
MinAvgMaxEPYC 9575F38.5249.1401.2EPYC 996539.2177.2410.2NVIDIA GH200 144G HBM3e39.3141.2272.9OpenBenchmarking.orgWatts, Fewer Is BetterTimed Godot Game Engine Compilation 4.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 22000 Joules per run.2. EPYC 9965: Approximate power consumption of 17067 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 18401 Joules per run.

OpenRadioss

OpenRadioss is an open-source AGPL-licensed finite element solver for dynamic event analysis OpenRadioss is based on Altair Radioss and open-sourced in 2022. This open-source finite element solver is benchmarked with various example models available from https://www.openradioss.org/models/ and https://github.com/OpenRadioss/ModelExchange/tree/main/Examples. This test is currently using a reference OpenRadioss binary build offered via GitHub. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterOpenRadioss 2023.09.15Model: INIVOL and Fluid Structure Interaction Drop ContainerEPYC 9575FEPYC 996520406080100SE +/- 0.36, N = 3SE +/- 0.04, N = 389.6677.76
MinAvgMaxEPYC 9575F38.8369.6400.0EPYC 996538.9344.2406.7OpenBenchmarking.orgWatts, Fewer Is BetterOpenRadioss 2023.09.15CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 36093 Joules per run.2. EPYC 9965: Approximate power consumption of 31209 Joules per run.

TensorFlow

This is a benchmark of the TensorFlow deep learning framework using the TensorFlow reference benchmarks (tensorflow/benchmarks with tf_cnn_benchmarks.py). Note with the Phoronix Test Suite there is also pts/tensorflow-lite for benchmarking the TensorFlow Lite binaries if desired for complementary metrics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: ResNet-50EPYC 9965EPYC 9575F50100150200250SE +/- 0.20, N = 3SE +/- 0.13, N = 3201.12234.68
OpenBenchmarking.orgimages/sec Per Watt, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: ResNet-50EPYC 9965EPYC 9575F0.17390.34780.52170.69560.86950.7020.773
MinAvgMaxEPYC 9575F38.3303.7322.5EPYC 996538.0286.5307.5OpenBenchmarking.orgWatts, Fewer Is BetterTensorFlow 2.16.1CPU Power Consumption Monitor801602403204001. EPYC 9575F: Approximate power consumption of 38674 Joules per run.2. EPYC 9965: Approximate power consumption of 42686 Joules per run.

Device: CPU - Batch Size: 256 - Model: ResNet-50

NVIDIA GH200 144G HBM3e: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'absl'

QMCPACK

QMCPACK is a modern high-performance open-source Quantum Monte Carlo (QMC) simulation code making use of MPI for this benchmark of the H20 example code. QMCPACK is an open-source production level many-body ab initio Quantum Monte Carlo code for computing the electronic structure of atoms, molecules, and solids. QMCPACK is supported by the U.S. Department of Energy. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTotal Execution Time - Seconds, Fewer Is BetterQMCPACK 3.17.1Input: Li2_STO_aeEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e20406080100SE +/- 0.32, N = 3SE +/- 0.39, N = 3SE +/- 0.31, N = 394.5972.9369.61-march=native-march=native-mcpu=native1. (CXX) g++ options: -fopenmp -foffload=disable -finline-limit=1000 -fstrict-aliasing -funroll-all-loops -ffast-math -O3 -lm -ldl
MinAvgMaxEPYC 9575F38.7376.4399.8EPYC 996538.3360.0391.3NVIDIA GH200 144G HBM3e41.5268.3298.1OpenBenchmarking.orgWatts, Fewer Is BetterQMCPACK 3.17.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 29110 Joules per run.2. EPYC 9965: Approximate power consumption of 36600 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 20299 Joules per run.

Zstd Compression

This test measures the time needed to compress/decompress a sample file (silesia.tar) using Zstd (Zstandard) compression with options for different compression levels / settings. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.5.4Compression Level: 19, Long Mode - Decompression SpeedNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F400800120016002000SE +/- 5.67, N = 3SE +/- 9.80, N = 3SE +/- 9.59, N = 31305.91469.01970.81. (CC) gcc options: -O3 -pthread -lz -llzma
OpenBenchmarking.orgMB/s Per Watt, More Is BetterZstd Compression 1.5.4Compression Level: 19, Long Mode - Decompression SpeedEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e71421283516.9519.4129.48
MinAvgMaxEPYC 9575F38.2101.5128.9EPYC 996537.986.7100.8NVIDIA GH200 144G HBM3e39.244.349.0OpenBenchmarking.orgWatts, Fewer Is BetterZstd Compression 1.5.4CPU Power Consumption Monitor40801201602001. EPYC 9575F: Approximate power consumption of 7207 Joules per run.2. EPYC 9965: Approximate power consumption of 7050 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 3810 Joules per run.

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.5.4Compression Level: 19, Long Mode - Compression SpeedNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F3691215SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.06, N = 38.649.2911.801. (CC) gcc options: -O3 -pthread -lz -llzma

OSPRay Studio

Intel OSPRay Studio is an open-source, interactive visualization and ray-tracing software package. OSPRay Studio makes use of Intel OSPRay, a portable ray-tracing engine for high-performance, high-fidelity visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOSPRay Studio 1.0Camera: 1 - Resolution: 4K - Samples Per Pixel: 16 - Renderer: Path Tracer - Acceleration: CPUEPYC 9575FEPYC 99653K6K9K12K15KSE +/- 1.76, N = 3SE +/- 7.81, N = 3161748530
MinAvgMaxEPYC 996538.3335.2391.7EPYC 9575F38.5333.6401.2OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay Studio 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 30725 Joules per run.2. EPYC 9575F: Approximate power consumption of 20241 Joules per run.

OSPRay

Intel OSPRay is a portable ray-tracing engine for high-performance, high-fidelity scientific visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgItems Per Second, More Is BetterOSPRay 3.1Benchmark: gravity_spheres_volume/dim_512/pathtracer/real_timeEPYC 9575FEPYC 99651428425670SE +/- 0.01, N = 3SE +/- 0.02, N = 335.6564.80
OpenBenchmarking.orgItems Per Second Per Watt, More Is BetterOSPRay 3.1Benchmark: gravity_spheres_volume/dim_512/pathtracer/real_timeEPYC 9575FEPYC 99650.04120.08240.12360.16480.2060.0970.183
MinAvgMaxEPYC 9575F38.5367.4400.9EPYC 996538.1354.6403.8OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay 3.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 29635 Joules per run.2. EPYC 9965: Approximate power consumption of 24116 Joules per run.

Zstd Compression

This test measures the time needed to compress/decompress a sample file (silesia.tar) using Zstd (Zstandard) compression with options for different compression levels / settings. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.5.4Compression Level: 19 - Decompression SpeedNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F400800120016002000SE +/- 1.62, N = 3SE +/- 3.17, N = 3SE +/- 4.06, N = 31266.61520.52081.21. (CC) gcc options: -O3 -pthread -lz -llzma
OpenBenchmarking.orgMB/s Per Watt, More Is BetterZstd Compression 1.5.4Compression Level: 19 - Decompression SpeedEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e61218243016.5418.1927.05
MinAvgMaxEPYC 9575F38.2114.4155.8EPYC 996538.091.9112.9NVIDIA GH200 144G HBM3e39.146.854.5OpenBenchmarking.orgWatts, Fewer Is BetterZstd Compression 1.5.4CPU Power Consumption Monitor40801201602001. EPYC 9575F: Approximate power consumption of 7895 Joules per run.2. EPYC 9965: Approximate power consumption of 6281 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 3714 Joules per run.

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.5.4Compression Level: 19 - Compression SpeedNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F612182430SE +/- 0.07, N = 3SE +/- 0.10, N = 3SE +/- 0.12, N = 315.219.425.51. (CC) gcc options: -O3 -pthread -lz -llzma

libavif avifenc

This is a test of the AOMedia libavif library testing the encoding of a JPEG image to AV1 Image Format (AVIF). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is Betterlibavif avifenc 1.0Encoder Speed: 0NVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F20406080100SE +/- 0.08, N = 3SE +/- 0.08, N = 3SE +/- 0.03, N = 3108.2153.2740.851. (CXX) g++ options: -O3 -fPIC -lm
MinAvgMaxEPYC 9575F38.2201.2399.6EPYC 996538.0147.9311.6NVIDIA GH200 144G HBM3e38.868.6236.5OpenBenchmarking.orgWatts, Fewer Is Betterlibavif avifenc 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 9054 Joules per run.2. EPYC 9965: Approximate power consumption of 8479 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 7757 Joules per run.

Memcached

Memcached is a high performance, distributed memory object caching system. This Memcached test profiles makes use of memtier_benchmark for excuting this CPU/memory-focused server benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOps/sec, More Is BetterMemcached 1.6.19Set To Get Ratio: 1:100NVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F3M6M9M12M15MSE +/- 22880.28, N = 3SE +/- 70542.57, N = 3SE +/- 30489.72, N = 33049705.549290017.4613808558.421. (CXX) g++ options: -O2 -levent_openssl -levent -lcrypto -lssl -lpthread -lz -lpcre
OpenBenchmarking.orgOps/sec Per Watt, More Is BetterMemcached 1.6.19Set To Get Ratio: 1:100NVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F9K18K27K36K45K27521.0735232.5939869.27
MinAvgMaxEPYC 9575F38.3346.3402.0EPYC 996537.9263.7306.3NVIDIA GH200 144G HBM3e39.5110.8124.9OpenBenchmarking.orgWatts, Fewer Is BetterMemcached 1.6.19CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 24244 Joules per run.2. EPYC 9965: Approximate power consumption of 18545 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 7831 Joules per run.

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Face Detection FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F6001200180024003000SE +/- 2.14, N = 3SE +/- 1.17, N = 3SE +/- 0.13, N = 32708.94414.32225.86MIN: 2034.06 / MAX: 5676.5MIN: 173.52 / MAX: 443.021. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 9575F38.4348.4377.7EPYC 996538.5337.5378.1NVIDIA GH200 144G HBM3e38.8160.7212.5OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 22879 Joules per run.2. EPYC 9965: Approximate power consumption of 23062 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 11572 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Face Detection FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996550100150200250SE +/- 0.00, N = 3SE +/- 0.09, N = 3SE +/- 0.69, N = 36.56141.43231.141. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Noise Suppression Poconet-Like FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F20406080100SE +/- 0.21, N = 3SE +/- 0.04, N = 3SE +/- 0.00, N = 390.2010.888.38MIN: 85.88 / MAX: 141.56MIN: 6.04 / MAX: 54.331. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 996539.1359.7403.1EPYC 9575F38.5345.0376.0NVIDIA GH200 144G HBM3e41.0187.5212.8OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 24579 Joules per run.2. EPYC 9575F: Approximate power consumption of 22542 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 12312 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Noise Suppression Poconet-Like FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652K4K6K8K10KSE +/- 0.46, N = 3SE +/- 4.59, N = 3SE +/- 13.78, N = 3199.487401.3110251.071. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Person Detection FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F120240360480600SE +/- 1.50, N = 3SE +/- 1.29, N = 3SE +/- 0.07, N = 3549.25149.8149.09MIN: 332.02 / MAX: 748.7MIN: 54.22 / MAX: 226.991. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 9575F38.7371.7400.8EPYC 996538.5358.1413.8NVIDIA GH200 144G HBM3e40.096.6136.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 24160 Joules per run.2. EPYC 9965: Approximate power consumption of 23872 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 6440 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Person Detection FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F140280420560700SE +/- 0.09, N = 3SE +/- 5.55, N = 3SE +/- 0.93, N = 332.71640.09650.801. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OSPRay Studio

Intel OSPRay Studio is an open-source, interactive visualization and ray-tracing software package. OSPRay Studio makes use of Intel OSPRay, a portable ray-tracing engine for high-performance, high-fidelity visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOSPRay Studio 1.0Camera: 3 - Resolution: 4K - Samples Per Pixel: 1 - Renderer: Path Tracer - Acceleration: CPUEPYC 9575FEPYC 996530060090012001500SE +/- 0.33, N = 3SE +/- 0.33, N = 31203632
MinAvgMaxEPYC 9575F38.3339.6400.8EPYC 996538.3312.8391.1OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay Studio 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 22979 Joules per run.2. EPYC 9965: Approximate power consumption of 19912 Joules per run.

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Machine Translation EN To DE FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F4080120160200SE +/- 0.04, N = 3SE +/- 0.12, N = 3SE +/- 0.02, N = 3161.9878.0741.94MIN: 141.94 / MAX: 399.19MIN: 33.39 / MAX: 161.89MIN: 20.28 / MAX: 62.171. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 9575F38.4372.0401.3EPYC 996538.8364.4410.0NVIDIA GH200 144G HBM3e41.2160.2178.5OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 24059 Joules per run.2. EPYC 9965: Approximate power consumption of 24053 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 10625 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Machine Translation EN To DE FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996530060090012001500SE +/- 0.03, N = 3SE +/- 0.37, N = 3SE +/- 1.87, N = 3110.98761.691227.231. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Person Vehicle Bike Detection FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F816243240SE +/- 0.04, N = 3SE +/- 0.02, N = 3SE +/- 0.01, N = 332.898.635.52MIN: 19.5 / MAX: 53.28MIN: 4.74 / MAX: 37.97MIN: 3.58 / MAX: 12.721. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 996538.7350.2386.6EPYC 9575F38.6297.8322.4NVIDIA GH200 144G HBM3e41.2148.9161.7OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 23230 Joules per run.2. EPYC 9575F: Approximate power consumption of 19256 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 9776 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Person Vehicle Bike Detection FP16 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652K4K6K8K10KSE +/- 0.71, N = 3SE +/- 12.01, N = 3SE +/- 20.79, N = 3547.015717.1911033.611. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Road Segmentation ADAS FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F80160240320400SE +/- 0.04, N = 3SE +/- 0.05, N = 3SE +/- 0.01, N = 3351.9027.3414.41MIN: 342.58 / MAX: 423.031. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 996538.8333.6367.7EPYC 9575F38.5317.4342.3NVIDIA GH200 144G HBM3e41.7147.9175.5OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 21797 Joules per run.2. EPYC 9575F: Approximate power consumption of 20420 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 9909 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Road Segmentation ADAS FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99658001600240032004000SE +/- 0.02, N = 3SE +/- 1.81, N = 3SE +/- 6.65, N = 351.072198.483497.171. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Face Detection Retail FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F816243240SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 334.664.972.85MIN: 33.13 / MAX: 44.01MIN: 1.9 / MAX: 35.751. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 996539.2346.2374.4EPYC 9575F38.4335.2360.8NVIDIA GH200 144G HBM3e41.3176.7194.0OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 22616 Joules per run.2. EPYC 9575F: Approximate power consumption of 21566 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 11545 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Face Detection Retail FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99658K16K24K32K40KSE +/- 0.10, N = 3SE +/- 45.20, N = 3SE +/- 31.19, N = 3519.0221512.0737263.741. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F0.82581.65162.47743.30324.129SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 33.670.730.25MIN: 1.08 / MAX: 40.71MIN: 0.14 / MAX: 57.971. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 996538.7328.4358.0EPYC 9575F38.7308.2330.5NVIDIA GH200 144G HBM3e41.0118.2127.1OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 21344 Joules per run.2. EPYC 9575F: Approximate power consumption of 19825 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 7685 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996540K80K120K160K200KSE +/- 6.97, N = 3SE +/- 689.22, N = 3SE +/- 2422.71, N = 34896.13173795.43208220.831. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Vehicle Detection FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F306090120150SE +/- 0.08, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3117.257.524.24MIN: 114.47 / MAX: 135.851. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 996538.8373.2403.5EPYC 9575F38.4333.2357.4NVIDIA GH200 144G HBM3e40.2154.0182.4OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 24135 Joules per run.2. EPYC 9575F: Approximate power consumption of 21326 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 10113 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Vehicle Detection FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99653K6K9K12K15KSE +/- 0.10, N = 3SE +/- 12.36, N = 3SE +/- 17.14, N = 3153.477427.2712591.541. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OSPRay Studio

Intel OSPRay Studio is an open-source, interactive visualization and ray-tracing software package. OSPRay Studio makes use of Intel OSPRay, a portable ray-tracing engine for high-performance, high-fidelity visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOSPRay Studio 1.0Camera: 1 - Resolution: 4K - Samples Per Pixel: 1 - Renderer: Path Tracer - Acceleration: CPUEPYC 9575FEPYC 99652004006008001000SE +/- 2.19, N = 3SE +/- 0.33, N = 31015535
MinAvgMaxEPYC 9575F38.5337.6401.2EPYC 996538.2311.5391.4OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay Studio 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 22394 Joules per run.2. EPYC 9965: Approximate power consumption of 19418 Joules per run.

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Weld Porosity Detection FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F918273645SE +/- 0.07, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 338.638.384.60MIN: 23.06 / MAX: 413.24MIN: 2.46 / MAX: 47.2MIN: 2.36 / MAX: 11.161. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl
MinAvgMaxEPYC 996538.7333.5358.9EPYC 9575F38.8330.6355.4NVIDIA GH200 144G HBM3e39.8122.4132.5OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2024.0CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 21453 Joules per run.2. EPYC 9575F: Approximate power consumption of 21159 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 7998 Joules per run.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Weld Porosity Detection FP16-INT8 - Device: CPUNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99655K10K15K20K25KSE +/- 0.86, N = 3SE +/- 5.04, N = 3SE +/- 6.08, N = 3465.8213533.2022159.641. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

GraphicsMagick

This is a test of GraphicsMagick with its OpenMP implementation that performs various imaging tests on a sample high resolution (currently 15400 x 6940) JPEG image. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.43Operation: SharpenEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 9965100200300400500SE +/- 0.33, N = 3SE +/- 0.00, N = 3SE +/- 0.88, N = 3301428445-lfreetype -lbz2-lfreetype -lbz21. (CC) gcc options: -fopenmp -O2 -ltiff -ljbig -lsharpyuv -lwebp -lwebpmux -ljpeg -lXext -lSM -lICE -lX11 -lxml2 -lzstd -llzma -lz -lm -lpthread -lgomp
OpenBenchmarking.orgIterations Per Minute Per Watt, More Is BetterGraphicsMagick 1.3.43Operation: SharpenEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e0.360.721.081.441.80.9041.5361.600
MinAvgMaxEPYC 9575F38.7333.1356.2EPYC 996538.4289.7312.1NVIDIA GH200 144G HBM3e42.8267.5300.5OpenBenchmarking.orgWatts, Fewer Is BetterGraphicsMagick 1.3.43CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 21319 Joules per run.2. EPYC 9965: Approximate power consumption of 18447 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 17299 Joules per run.

Speedb

Speedb is a next-generation key value storage engine that is RocksDB compatible and aiming for stability, efficiency, and performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterSpeedb 2.7Test: Random ReadEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 9965200M400M600M800M1000MSE +/- 755684.49, N = 3SE +/- 126199.44, N = 3SE +/- 370648.73, N = 3526671792569021111969890739-lpthread-lpthread1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti
OpenBenchmarking.orgOp/s Per Watt, More Is BetterSpeedb 2.7Test: Random ReadEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 9965600K1200K1800K2400K3000K1520207.332132224.782712772.81
MinAvgMaxEPYC 996538.2357.5383.6EPYC 9575F38.3346.4371.3NVIDIA GH200 144G HBM3e42.0266.9286.5OpenBenchmarking.orgWatts, Fewer Is BetterSpeedb 2.7CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 22882 Joules per run.2. EPYC 9575F: Approximate power consumption of 22173 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 17257 Joules per run.

GraphicsMagick

This is a test of GraphicsMagick with its OpenMP implementation that performs various imaging tests on a sample high resolution (currently 15400 x 6940) JPEG image. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.43Operation: EnhancedEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 9965110220330440550SE +/- 0.67, N = 3SE +/- 0.00, N = 3SE +/- 1.53, N = 3336369495-lfreetype -lbz2-lfreetype -lbz21. (CC) gcc options: -fopenmp -O2 -ltiff -ljbig -lsharpyuv -lwebp -lwebpmux -ljpeg -lXext -lSM -lICE -lX11 -lxml2 -lzstd -llzma -lz -lm -lpthread -lgomp
OpenBenchmarking.orgIterations Per Minute Per Watt, More Is BetterGraphicsMagick 1.3.43Operation: EnhancedEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 99650.41990.83981.25971.67962.09950.9861.5421.866
MinAvgMaxEPYC 9575F38.3340.8367.5EPYC 996538.0265.3285.1NVIDIA GH200 144G HBM3e40.7239.3262.5OpenBenchmarking.orgWatts, Fewer Is BetterGraphicsMagick 1.3.43CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 21695 Joules per run.2. EPYC 9965: Approximate power consumption of 17068 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 15552 Joules per run.

OpenBenchmarking.orgIterations Per Minute, More Is BetterGraphicsMagick 1.3.43Operation: Noise-GaussianEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e90180270360450SE +/- 1.00, N = 3SE +/- 1.00, N = 3SE +/- 0.00, N = 3291296397-lfreetype -lbz2-lfreetype -lbz21. (CC) gcc options: -fopenmp -O2 -ltiff -ljbig -lsharpyuv -lwebp -lwebpmux -ljpeg -lXext -lSM -lICE -lX11 -lxml2 -lzstd -llzma -lz -lm -lpthread -lgomp
OpenBenchmarking.orgIterations Per Minute Per Watt, More Is BetterGraphicsMagick 1.3.43Operation: Noise-GaussianEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e0.53661.07321.60982.14642.6831.2951.7792.385
MinAvgMaxEPYC 9575F38.4224.7243.8NVIDIA GH200 144G HBM3e41.9166.4283.3EPYC 996538.3166.4181.2OpenBenchmarking.orgWatts, Fewer Is BetterGraphicsMagick 1.3.43CPU Power Consumption Monitor701402102803501. EPYC 9575F: Approximate power consumption of 14456 Joules per run.2. NVIDIA GH200 144G HBM3e: Approximate power consumption of 10708 Joules per run.3. EPYC 9965: Approximate power consumption of 10595 Joules per run.

Liquid-DSP

LiquidSDR's Liquid-DSP is a software-defined radio (SDR) digital signal processing library. This test profile runs a multi-threaded benchmark of this SDR/DSP library focused on embedded platform usage. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgsamples/s, More Is BetterLiquid-DSP 1.6Threads: 1 - Buffer Length: 256 - Filter Length: 512NVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F6M12M18M24M30MSE +/- 33.33, N = 3SE +/- 244710.59, N = 12SE +/- 666.67, N = 3345873320527750282746671. (CC) gcc options: -O3 -pthread -lm -lc -lliquid
OpenBenchmarking.orgsamples/s Per Watt, More Is BetterLiquid-DSP 1.6Threads: 1 - Buffer Length: 256 - Filter Length: 512NVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F70K140K210K280K350K82973.28266440.14325399.16
MinAvgMaxEPYC 9575F38.186.994.0EPYC 996537.977.082.5NVIDIA GH200 144G HBM3e38.841.742.9OpenBenchmarking.orgWatts, Fewer Is BetterLiquid-DSP 1.6CPU Power Consumption Monitor204060801001. EPYC 9575F: Approximate power consumption of 2954 Joules per run.2. EPYC 9965: Approximate power consumption of 2620 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 1445 Joules per run.

RocksDB

This is a benchmark of Meta/Facebook's RocksDB as an embeddable persistent key-value store for fast storage based on Google's LevelDB. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 9.0Test: Random ReadEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 9965200M400M600M800M1000MSE +/- 1563552.18, N = 3SE +/- 143829.35, N = 3SE +/- 1700402.40, N = 3505733525568339463958354064-lpthread-lpthread1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 9.0Test: Random ReadEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 9965600K1200K1800K2400K3000K1462898.912103254.002704939.66
MinAvgMaxEPYC 996538.0354.3380.0EPYC 9575F38.3345.7371.1NVIDIA GH200 144G HBM3e40.1270.2291.0OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 9.0CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 22675 Joules per run.2. EPYC 9575F: Approximate power consumption of 22125 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 17564 Joules per run.

NAMD

NAMD is a parallel molecular dynamics code designed for high-performance simulation of large biomolecular systems. NAMD was developed by the Theoretical and Computational Biophysics Group in the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgns/day, More Is BetterNAMD 3.0b6Input: STMV with 1,066,628 AtomsEPYC 9575FEPYC 99651.13222.26443.39664.52885.661SE +/- 0.00773, N = 3SE +/- 0.01244, N = 33.803135.03185
OpenBenchmarking.orgns/day Per Watt, More Is BetterNAMD 3.0b6Input: STMV with 1,066,628 AtomsEPYC 9575FEPYC 99650.00830.01660.02490.03320.04150.0140.037
MinAvgMaxEPYC 9575F38.1276.0398.0EPYC 996538.0135.3374.0OpenBenchmarking.orgWatts, Fewer Is BetterNAMD 3.0b6CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 9200 Joules per run.2. EPYC 9965: Approximate power consumption of 12580 Joules per run.

OSPRay Studio

Intel OSPRay Studio is an open-source, interactive visualization and ray-tracing software package. OSPRay Studio makes use of Intel OSPRay, a portable ray-tracing engine for high-performance, high-fidelity visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOSPRay Studio 1.0Camera: 3 - Resolution: 4K - Samples Per Pixel: 32 - Renderer: Path Tracer - Acceleration: CPUEPYC 9575FEPYC 99659K18K27K36K45KSE +/- 11.14, N = 3SE +/- 14.95, N = 34132420149
MinAvgMaxEPYC 996538.4324.3392.9EPYC 9575F38.5319.0401.2OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay Studio 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 24542 Joules per run.2. EPYC 9575F: Approximate power consumption of 16374 Joules per run.

OpenBenchmarking.orgms, Fewer Is BetterOSPRay Studio 1.0Camera: 3 - Resolution: 4K - Samples Per Pixel: 16 - Renderer: Path Tracer - Acceleration: CPUEPYC 9575FEPYC 99654K8K12K16K20KSE +/- 28.92, N = 3SE +/- 10.12, N = 31919610067
MinAvgMaxEPYC 9575F38.3341.8401.2EPYC 996538.3302.0391.2OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay Studio 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 23814 Joules per run.2. EPYC 9965: Approximate power consumption of 16810 Joules per run.

Helsing

Helsing is an open-source POSIX vampire number generator. This test profile measures the time it takes to generate vampire numbers between varying numbers of digits. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterHelsing 1.0-betaDigit Range: 14 digitEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 99651428425670SE +/- 0.23, N = 3SE +/- 0.02, N = 3SE +/- 0.08, N = 361.8356.9739.771. (CC) gcc options: -O2 -pthread
MinAvgMaxEPYC 9575F38.4346.7373.1EPYC 996538.6342.9387.2NVIDIA GH200 144G HBM3e40.6266.7291.3OpenBenchmarking.orgWatts, Fewer Is BetterHelsing 1.0-betaCPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 22769 Joules per run.2. EPYC 9965: Approximate power consumption of 14971 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 16357 Joules per run.

OSPRay Studio

Intel OSPRay Studio is an open-source, interactive visualization and ray-tracing software package. OSPRay Studio makes use of Intel OSPRay, a portable ray-tracing engine for high-performance, high-fidelity visualizations. OSPRay builds off Intel's Embree and Intel SPMD Program Compiler (ISPC) components as part of the oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOSPRay Studio 1.0Camera: 1 - Resolution: 4K - Samples Per Pixel: 32 - Renderer: Path Tracer - Acceleration: CPUEPYC 9575FEPYC 99658K16K24K32K40KSE +/- 25.41, N = 3SE +/- 21.39, N = 33527417091
MinAvgMaxEPYC 996538.3316.1391.9EPYC 9575F38.5307.8401.2OpenBenchmarking.orgWatts, Fewer Is BetterOSPRay Studio 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 20969 Joules per run.2. EPYC 9575F: Approximate power consumption of 13750 Joules per run.

QuantLib

QuantLib is an open-source library/framework around quantitative finance for modeling, trading and risk management scenarios. QuantLib is written in C++ with Boost and its built-in benchmark used reports the QuantLib Benchmark Index benchmark score. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMFLOPS, More Is BetterQuantLib 1.32Configuration: Multi-ThreadedNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965120K240K360K480K600KSE +/- 316.93, N = 3SE +/- 661.19, N = 3SE +/- 485.45, N = 3261141.5297665.1566481.41. (CXX) g++ options: -O3 -march=native -fPIE -pie
OpenBenchmarking.orgMFLOPS Per Watt, More Is BetterQuantLib 1.32Configuration: Multi-ThreadedEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 996530060090012001500897.441209.801606.22
MinAvgMaxEPYC 996537.9352.7387.8EPYC 9575F38.5331.7399.4NVIDIA GH200 144G HBM3e41.0215.9290.4OpenBenchmarking.orgWatts, Fewer Is BetterQuantLib 1.32CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 25275 Joules per run.2. EPYC 9575F: Approximate power consumption of 15479 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 6692 Joules per run.

miniBUDE

MiniBUDE is a mini application for the the core computation of the Bristol University Docking Engine (BUDE). This test profile currently makes use of the OpenMP implementation of miniBUDE for CPU benchmarking. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgBillion Interactions/s, More Is BetterminiBUDE 20210901Implementation: OpenMP - Input Deck: BM2NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996590180270360450SE +/- 0.02, N = 3SE +/- 0.62, N = 3SE +/- 1.89, N = 454.66244.07428.05-mcpu=native-march=native-march=native1. (CC) gcc options: -std=c99 -Ofast -ffast-math -fopenmp -lm
OpenBenchmarking.orgBillion Interactions/s Per Watt, More Is BetterminiBUDE 20210901Implementation: OpenMP - Input Deck: BM2NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99650.32580.65160.97741.30321.6290.2550.7631.448
MinAvgMaxEPYC 9575F38.3319.7382.8EPYC 996537.8295.7388.6NVIDIA GH200 144G HBM3e41.1214.0244.9OpenBenchmarking.orgWatts, Fewer Is BetterminiBUDE 20210901CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 8100 Joules per run.2. EPYC 9965: Approximate power consumption of 4805 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 21253 Joules per run.

OpenBenchmarking.orgGFInst/s, More Is BetterminiBUDE 20210901Implementation: OpenMP - Input Deck: BM2NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652K4K6K8K10KSE +/- 0.49, N = 3SE +/- 15.38, N = 3SE +/- 47.25, N = 41366.406101.7910701.18-march=native-march=native1. (CC) gcc options: -std=c99 -Ofast -ffast-math -fopenmp -lm

NAMD

NAMD is a parallel molecular dynamics code designed for high-performance simulation of large biomolecular systems. NAMD was developed by the Theoretical and Computational Biophysics Group in the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgns/day, More Is BetterNAMD 3.0b6Input: ATPase with 327,506 AtomsEPYC 9575FEPYC 9965510152025SE +/- 0.01, N = 3SE +/- 0.02, N = 313.3419.90
OpenBenchmarking.orgns/day Per Watt, More Is BetterNAMD 3.0b6Input: ATPase with 327,506 AtomsEPYC 9575FEPYC 99650.04030.08060.12090.16120.20150.0560.179
MinAvgMaxEPYC 9575F38.2236.1397.6EPYC 996537.8111.2383.0OpenBenchmarking.orgWatts, Fewer Is BetterNAMD 3.0b6CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 5273 Joules per run.2. EPYC 9965: Approximate power consumption of 6855 Joules per run.

WebP Image Encode

This is a test of Google's libwebp with the cwebp image encode utility and using a sample 6000x4000 pixel JPEG image as the input. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMP/s, More Is BetterWebP Image Encode 1.2.4Encode Settings: Quality 100, Lossless, Highest CompressionNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F0.18680.37360.56040.74720.934SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 30.540.610.83-lpng16 -ljpeg -ltiff1. (CC) gcc options: -fvisibility=hidden -O2 -lm
OpenBenchmarking.orgMP/s Per Watt, More Is BetterWebP Image Encode 1.2.4Encode Settings: Quality 100, Lossless, Highest CompressionEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e0.00270.00540.00810.01080.01350.0070.0090.012
MinAvgMaxEPYC 9575F38.295.0114.3EPYC 996537.883.093.8NVIDIA GH200 144G HBM3e38.743.746.0OpenBenchmarking.orgWatts, Fewer Is BetterWebP Image Encode 1.2.4CPU Power Consumption Monitor40801201602001. EPYC 9575F: Approximate power consumption of 3134 Joules per run.2. EPYC 9965: Approximate power consumption of 3599 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 2155 Joules per run.

Blender

Blender is an open-source 3D creation and modeling software project. This test is of Blender's Cycles performance with various sample files. GPU computing via NVIDIA OptiX and NVIDIA CUDA is currently supported as well as HIP for AMD Radeon GPUs and Intel oneAPI for Intel Graphics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 4.1Blend File: Pabellon Barcelona - Compute: CPU-OnlyEPYC 9575FEPYC 99651020304050SE +/- 0.03, N = 3SE +/- 0.08, N = 345.7525.67
MinAvgMaxEPYC 9575F38.3356.3394.4EPYC 996538.3319.9381.1OpenBenchmarking.orgWatts, Fewer Is BetterBlender 4.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 17814 Joules per run.2. EPYC 9965: Approximate power consumption of 9704 Joules per run.

NAS Parallel Benchmarks

NPB, NAS Parallel Benchmarks, is a benchmark developed by NASA for high-end computer systems. This test profile currently uses the MPI version of NPB. This test profile offers selecting the different NPB tests/problems and varying problem sizes. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTotal Mop/s, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: SP.CNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996530K60K90K120K150KSE +/- 60.46, N = 3SE +/- 376.98, N = 5SE +/- 1227.63, N = 424153.15148567.25152747.871. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.6
OpenBenchmarking.orgTotal Mop/s Per Watt, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: SP.CNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965120240360480600105.21518.54566.79
MinAvgMaxEPYC 9575F38.2286.5399.7EPYC 996538.1269.5413.1NVIDIA GH200 144G HBM3e41.3229.6257.1OpenBenchmarking.orgWatts, Fewer Is BetterNAS Parallel Benchmarks 3.4CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 4183 Joules per run.2. EPYC 9965: Approximate power consumption of 4447 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 15151 Joules per run.

Liquid-DSP

LiquidSDR's Liquid-DSP is a software-defined radio (SDR) digital signal processing library. This test profile runs a multi-threaded benchmark of this SDR/DSP library focused on embedded platform usage. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgsamples/s, More Is BetterLiquid-DSP 1.6Threads: 256 - Buffer Length: 256 - Filter Length: 512NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965700M1400M2100M2800M3500MSE +/- 165227.12, N = 3SE +/- 1319511.69, N = 3SE +/- 1602428.71, N = 3253630000181793333331816333331. (CC) gcc options: -O3 -pthread -lm -lc -lliquid
OpenBenchmarking.orgsamples/s Per Watt, More Is BetterLiquid-DSP 1.6Threads: 256 - Buffer Length: 256 - Filter Length: 512NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652M4M6M8M10M1400334.375494414.938815828.44
MinAvgMaxEPYC 996538.5360.9416.7EPYC 9575F38.4330.9371.6NVIDIA GH200 144G HBM3e42.5181.1204.9OpenBenchmarking.orgWatts, Fewer Is BetterLiquid-DSP 1.6CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 12511 Joules per run.2. EPYC 9575F: Approximate power consumption of 11360 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 6460 Joules per run.

John The Ripper

This is a benchmark of John The Ripper, which is a password cracker. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 2023.03.14Test: WPA PSKNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965400K800K1200K1600K2000KSE +/- 14.15, N = 3SE +/- 630.47, N = 3SE +/- 2333.33, N = 3761308577021774667-m64 -lgmp -lbz2-m64 -lgmp -lbz21. (CC) gcc options: -lssl -lcrypto -fopenmp -lm -lrt -lz -ldl -lcrypt
OpenBenchmarking.orgReal C/S Per Watt, More Is BetterJohn The Ripper 2023.03.14Test: WPA PSKNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996510002000300040005000525.732539.284459.57
MinAvgMaxEPYC 996538.3397.9449.6EPYC 9575F38.7337.8381.2NVIDIA GH200 144G HBM3e41.0144.8161.9OpenBenchmarking.orgWatts, Fewer Is BetterJohn The Ripper 2023.03.14CPU Power Consumption Monitor1202403604806001. EPYC 9965: Approximate power consumption of 13530 Joules per run.2. EPYC 9575F: Approximate power consumption of 11372 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 5261 Joules per run.

Liquid-DSP

LiquidSDR's Liquid-DSP is a software-defined radio (SDR) digital signal processing library. This test profile runs a multi-threaded benchmark of this SDR/DSP library focused on embedded platform usage. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgsamples/s, More Is BetterLiquid-DSP 1.6Threads: 256 - Buffer Length: 256 - Filter Length: 57NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652000M4000M6000M8000M10000MSE +/- 1675642.50, N = 3SE +/- 9980647.94, N = 3SE +/- 21733844.57, N = 32669733333544260000081858000001. (CC) gcc options: -O3 -pthread -lm -lc -lliquid
OpenBenchmarking.orgsamples/s Per Watt, More Is BetterLiquid-DSP 1.6Threads: 256 - Buffer Length: 256 - Filter Length: 57NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99655M10M15M20M25M11492978.8615772371.3123573051.76
MinAvgMaxEPYC 996538.8347.3399.8EPYC 9575F38.4345.1391.3NVIDIA GH200 144G HBM3e42.4232.3269.6OpenBenchmarking.orgWatts, Fewer Is BetterLiquid-DSP 1.6CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 11807 Joules per run.2. EPYC 9575F: Approximate power consumption of 11732 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 8208 Joules per run.

OpenBenchmarking.orgsamples/s, More Is BetterLiquid-DSP 1.6Threads: 128 - Buffer Length: 256 - Filter Length: 512NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965500M1000M1500M2000M2500MSE +/- 488683.04, N = 3SE +/- 1785434.90, N = 3SE +/- 1855622.09, N = 3248263333182293333322817000001. (CC) gcc options: -O3 -pthread -lm -lc -lliquid
OpenBenchmarking.orgsamples/s Per Watt, More Is BetterLiquid-DSP 1.6Threads: 128 - Buffer Length: 256 - Filter Length: 512NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99651.4M2.8M4.2M5.6M7M1396900.825541653.376644407.92
MinAvgMaxEPYC 996538.8343.4397.5EPYC 9575F38.4329.0377.4NVIDIA GH200 144G HBM3e38.7177.7203.0OpenBenchmarking.orgWatts, Fewer Is BetterLiquid-DSP 1.6CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 11790 Joules per run.2. EPYC 9575F: Approximate power consumption of 11294 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 6220 Joules per run.

OpenBenchmarking.orgsamples/s, More Is BetterLiquid-DSP 1.6Threads: 256 - Buffer Length: 256 - Filter Length: 32NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652000M4000M6000M8000M10000MSE +/- 3002221.40, N = 3SE +/- 6072982.06, N = 3SE +/- 10497036.62, N = 34502400000526586666780642666671. (CC) gcc options: -O3 -pthread -lm -lc -lliquid
OpenBenchmarking.orgsamples/s Per Watt, More Is BetterLiquid-DSP 1.6Threads: 256 - Buffer Length: 256 - Filter Length: 32EPYC 9575FNVIDIA GH200 144G HBM3eEPYC 99655M10M15M20M25M14781256.8817535441.6623123930.33
MinAvgMaxEPYC 9575F38.4356.3401.0EPYC 996538.4348.7397.1NVIDIA GH200 144G HBM3e41.2256.8292.1OpenBenchmarking.orgWatts, Fewer Is BetterLiquid-DSP 1.6CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 11994 Joules per run.2. EPYC 9965: Approximate power consumption of 11857 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 9072 Joules per run.

John The Ripper

This is a benchmark of John The Ripper, which is a password cracker. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 2023.03.14Test: BlowfishNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996580K160K240K320K400KSE +/- 19.86, N = 3SE +/- 41.32, N = 3SE +/- 201.40, N = 374756198954386082-m64 -lgmp -lbz2-m64 -lgmp -lbz21. (CC) gcc options: -lssl -lcrypto -fopenmp -lm -lrt -lz -ldl -lcrypt
OpenBenchmarking.orgReal C/S Per Watt, More Is BetterJohn The Ripper 2023.03.14Test: BlowfishNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652004006008001000508.72562.771128.50
MinAvgMaxXeon 6980P - DDR5-640050.4446.6539.6Xeon 6980P50.8445.3539.8EPYC 9575F38.3354.4400.5EPYC 996538.3342.8386.5NVIDIA GH200 144G HBM3e42.2146.9165.2OpenBenchmarking.orgWatts, Fewer Is BetterJohn The Ripper 2023.03.14CPU Power Consumption Monitor1402804205607001. EPYC 9575F: Approximate power consumption of 12168 Joules per run.2. EPYC 9965: Approximate power consumption of 11654 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 5094 Joules per run.

OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 2023.03.14Test: bcryptNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996580K160K240K320K400KSE +/- 31.26, N = 3SE +/- 77.93, N = 3SE +/- 8.33, N = 374749199078386474-m64 -lgmp -lbz2-m64 -lgmp -lbz21. (CC) gcc options: -lssl -lcrypto -fopenmp -lm -lrt -lz -ldl -lcrypt
OpenBenchmarking.orgReal C/S Per Watt, More Is BetterJohn The Ripper 2023.03.14Test: bcryptNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652004006008001000517.44561.711127.52

Liquid-DSP

LiquidSDR's Liquid-DSP is a software-defined radio (SDR) digital signal processing library. This test profile runs a multi-threaded benchmark of this SDR/DSP library focused on embedded platform usage. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgsamples/s, More Is BetterLiquid-DSP 1.6Threads: 1 - Buffer Length: 256 - Filter Length: 32EPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e13M26M39M52M65MSE +/- 154211.98, N = 3SE +/- 63828.76, N = 3SE +/- 32671.77, N = 33989600053801667607903331. (CC) gcc options: -O3 -pthread -lm -lc -lliquid
OpenBenchmarking.orgsamples/s Per Watt, More Is BetterLiquid-DSP 1.6Threads: 1 - Buffer Length: 256 - Filter Length: 32EPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e300K600K900K1200K1500K525714.32637828.741381459.61
MinAvgMaxEPYC 9575F38.484.491.2EPYC 996538.175.983.0NVIDIA GH200 144G HBM3e39.744.046.2OpenBenchmarking.orgWatts, Fewer Is BetterLiquid-DSP 1.6CPU Power Consumption Monitor204060801001. EPYC 9575F: Approximate power consumption of 2868 Joules per run.2. EPYC 9965: Approximate power consumption of 2606 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 1525 Joules per run.

Blender

Blender is an open-source 3D creation and modeling software project. This test is of Blender's Cycles performance with various sample files. GPU computing via NVIDIA OptiX and NVIDIA CUDA is currently supported as well as HIP for AMD Radeon GPUs and Intel oneAPI for Intel Graphics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 4.1Blend File: Classroom - Compute: CPU-OnlyEPYC 9575FEPYC 9965918273645SE +/- 0.06, N = 3SE +/- 0.04, N = 338.2120.31
MinAvgMaxEPYC 9575F38.3349.8389.2EPYC 996538.0322.0389.8OpenBenchmarking.orgWatts, Fewer Is BetterBlender 4.1CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 14927 Joules per run.2. EPYC 9965: Approximate power consumption of 7943 Joules per run.

7-Zip Compression

This is a test of 7-Zip compression/decompression with its integrated benchmark feature. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMIPS, More Is Better7-Zip Compression 22.01Test: Decompression RatingNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965200K400K600K800K1000KSE +/- 414.91, N = 3SE +/- 344.02, N = 3SE +/- 150.31, N = 342464352703710018581. (CXX) g++ options: -lpthread -ldl -O2 -fPIC
OpenBenchmarking.orgMIPS Per Watt, More Is Better7-Zip Compression 22.01Test: Decompression RatingEPYC 9575FNVIDIA GH200 144G HBM3eEPYC 996570014002100280035001634.382612.523214.79
MinAvgMaxEPYC 9575F38.4322.5400.8EPYC 996538.9311.6408.2NVIDIA GH200 144G HBM3e42.0162.5246.0OpenBenchmarking.orgWatts, Fewer Is Better7-Zip Compression 22.01CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 9244 Joules per run.2. EPYC 9965: Approximate power consumption of 14439 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 4009 Joules per run.

OpenBenchmarking.orgMIPS, More Is Better7-Zip Compression 22.01Test: Compression RatingNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965200K400K600K800K1000KSE +/- 652.77, N = 3SE +/- 4011.15, N = 3SE +/- 5680.94, N = 343660065524110071501. (CXX) g++ options: -lpthread -ldl -O2 -fPIC

ASTC Encoder

ASTC Encoder (astcenc) is for the Adaptive Scalable Texture Compression (ASTC) format commonly used with OpenGL, OpenGL ES, and Vulkan graphics APIs. This test profile does a coding test of both compression/decompression. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMT/s, More Is BetterASTC Encoder 4.7Preset: ExhaustiveNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99653691215SE +/- 0.0036, N = 3SE +/- 0.0055, N = 3SE +/- 0.0016, N = 44.06606.203511.53121. (CXX) g++ options: -O3 -flto -pthread
OpenBenchmarking.orgMT/s Per Watt, More Is BetterASTC Encoder 4.7Preset: ExhaustiveNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99650.01040.02080.03120.04160.0520.0180.0200.046
MinAvgMaxEPYC 9575F38.4312.4391.4EPYC 996538.5253.0377.8NVIDIA GH200 144G HBM3e42.1223.2267.3OpenBenchmarking.orgWatts, Fewer Is BetterASTC Encoder 4.7CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 9371 Joules per run.2. EPYC 9965: Approximate power consumption of 5187 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 9747 Joules per run.

OpenRadioss

OpenRadioss is an open-source AGPL-licensed finite element solver for dynamic event analysis OpenRadioss is based on Altair Radioss and open-sourced in 2022. This open-source finite element solver is benchmarked with various example models available from https://www.openradioss.org/models/ and https://github.com/OpenRadioss/ModelExchange/tree/main/Examples. This test is currently using a reference OpenRadioss binary build offered via GitHub. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterOpenRadioss 2023.09.15Model: Cell Phone Drop TestEPYC 9965EPYC 9575F510152025SE +/- 0.06, N = 3SE +/- 0.09, N = 322.0317.64
MinAvgMaxEPYC 9575F38.5285.2399.1EPYC 996538.6232.2389.0OpenBenchmarking.orgWatts, Fewer Is BetterOpenRadioss 2023.09.15CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 7320 Joules per run.2. EPYC 9965: Approximate power consumption of 9209 Joules per run.

ASTC Encoder

ASTC Encoder (astcenc) is for the Adaptive Scalable Texture Compression (ASTC) format commonly used with OpenGL, OpenGL ES, and Vulkan graphics APIs. This test profile does a coding test of both compression/decompression. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMT/s, More Is BetterASTC Encoder 4.7Preset: Very ThoroughNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965510152025SE +/- 0.0008, N = 3SE +/- 0.0104, N = 3SE +/- 0.0042, N = 36.588510.119718.77131. (CXX) g++ options: -O3 -flto -pthread
OpenBenchmarking.orgMT/s Per Watt, More Is BetterASTC Encoder 4.7Preset: Very ThoroughNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99650.01690.03380.05070.06760.08450.0300.0320.075
MinAvgMaxEPYC 9575F38.4314.2390.3EPYC 996538.1249.9374.6NVIDIA GH200 144G HBM3e39.8218.7261.2OpenBenchmarking.orgWatts, Fewer Is BetterASTC Encoder 4.7CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 9634 Joules per run.2. EPYC 9965: Approximate power consumption of 5166 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 9697 Joules per run.

Algebraic Multi-Grid Benchmark

AMG is a parallel algebraic multigrid solver for linear systems arising from problems on unstructured grids. The driver provided with AMG builds linear systems for various 3-dimensional problems. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFigure Of Merit, More Is BetterAlgebraic Multi-Grid Benchmark 1.2NVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F700M1400M2100M2800M3500MSE +/- 3694072.61, N = 3SE +/- 2247127.13, N = 3SE +/- 7735064.17, N = 42256454667297634500031976112501. (CC) gcc options: -lparcsr_ls -lparcsr_mv -lseq_mv -lIJ_mv -lkrylov -lHYPRE_utilities -lm -fopenmp -lmpi
OpenBenchmarking.orgFigure Of Merit Per Watt, More Is BetterAlgebraic Multi-Grid Benchmark 1.2EPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e3M6M9M12M15M10018207.5210869334.8315435851.48
MinAvgMaxEPYC 996537.6297.1369.7EPYC 9575F38.3294.2401.4NVIDIA GH200 144G HBM3e39.8146.2260.2OpenBenchmarking.orgWatts, Fewer Is BetterAlgebraic Multi-Grid Benchmark 1.2CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 14161 Joules per run.2. EPYC 9575F: Approximate power consumption of 5001 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 3411 Joules per run.

Primesieve

Primesieve generates prime numbers using a highly optimized sieve of Eratosthenes implementation. Primesieve primarily benchmarks the CPU's L1/L2 cache performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterPrimesieve 12.1Length: 1e13NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965714212835SE +/- 0.01, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 431.4625.4015.981. (CXX) g++ options: -O3
MinAvgMaxEPYC 9575F38.3321.0393.1EPYC 996538.5301.5387.0NVIDIA GH200 144G HBM3e40.9233.0271.7OpenBenchmarking.orgWatts, Fewer Is BetterPrimesieve 12.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 9524 Joules per run.2. EPYC 9965: Approximate power consumption of 6030 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 8389 Joules per run.

NAS Parallel Benchmarks

NPB, NAS Parallel Benchmarks, is a benchmark developed by NASA for high-end computer systems. This test profile currently uses the MPI version of NPB. This test profile offers selecting the different NPB tests/problems and varying problem sizes. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTotal Mop/s, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: LU.CNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996570K140K210K280K350KSE +/- 235.16, N = 3SE +/- 1780.21, N = 6SE +/- 3237.26, N = 544257.15303092.49313412.101. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.6
OpenBenchmarking.orgTotal Mop/s Per Watt, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: LU.CNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996530060090012001500204.411157.051309.07
MinAvgMaxEPYC 9575F38.5262.0399.8EPYC 996538.5239.4392.4NVIDIA GH200 144G HBM3e41.1216.5254.9OpenBenchmarking.orgWatts, Fewer Is BetterNAS Parallel Benchmarks 3.4CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 3012 Joules per run.2. EPYC 9965: Approximate power consumption of 3112 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 11331 Joules per run.

srsRAN Project

srsRAN Project is a complete ORAN-native 5G RAN solution created by Software Radio Systems (SRS). The srsRAN Project radio suite was formerly known as srsLTE and can be used for building your own software-defined radio (SDR) 4G/5G mobile network. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMbps, More Is BettersrsRAN Project 23.10.1-20240325Test: PUSCH Processor Benchmark, Throughput TotalNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99652K4K6K8K10KSE +/- 0.06, N = 3SE +/- 0.56, N = 3SE +/- 0.19, N = 31848.24804.78940.9MIN: 1042.6 / MAX: 1848.3-march=native -mavx2 -mavx -msse4.1 -mfma -mavx512f -mavx512cd -mavx512bw -mavx512dq - MIN: 3396.5 / MAX: 4805.8-march=native -mavx2 -mavx -msse4.1 -mfma -mavx512f -mavx512cd -mavx512bw -mavx512dq - MIN: 7191.5 / MAX: 8941.31. (CXX) g++ options: -O3 -fno-trapping-math -fno-math-errno -ldl
OpenBenchmarking.orgMbps Per Watt, More Is BettersrsRAN Project 23.10.1-20240325Test: PUSCH Processor Benchmark, Throughput TotalNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996571421283516.7217.9328.62
MinAvgMaxEPYC 996539.0312.4368.2EPYC 9575F38.6268.0361.7NVIDIA GH200 144G HBM3e41.3110.5134.1OpenBenchmarking.orgWatts, Fewer Is BettersrsRAN Project 23.10.1-20240325CPU Power Consumption Monitor1002003004005001. EPYC 9965: Approximate power consumption of 10204 Joules per run.2. EPYC 9575F: Approximate power consumption of 6075 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 3758 Joules per run.

GROMACS

The GROMACS (GROningen MAchine for Chemical Simulations) molecular dynamics package testing with the water_GMX50 data. This test profile allows selecting between CPU and GPU-based GROMACS builds. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgNs Per Day, More Is BetterGROMACS 2024Implementation: MPI CPU - Input: water_GMX50_bareNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965612182430SE +/- 0.005, N = 3SE +/- 0.053, N = 3SE +/- 0.050, N = 36.00214.65123.142-O3 -lm-O3 -lm1. (CXX) g++ options:
OpenBenchmarking.orgNs Per Day Per Watt, More Is BetterGROMACS 2024Implementation: MPI CPU - Input: water_GMX50_bareNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99650.0180.0360.0540.0720.090.0280.0460.080
MinAvgMaxEPYC 9575F38.3315.2400.1EPYC 996537.7288.8417.2NVIDIA GH200 144G HBM3e40.8212.1289.0OpenBenchmarking.orgWatts, Fewer Is BetterGROMACS 2024CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 6829 Joules per run.2. EPYC 9965: Approximate power consumption of 8182 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 8341 Joules per run.

CloverLeaf

CloverLeaf is a Lagrangian-Eulerian hydrodynamics benchmark. This test profile currently makes use of CloverLeaf's OpenMP version. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterCloverLeaf 1.3Input: clover_bm64_shortNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F714212835SE +/- 0.06, N = 3SE +/- 0.09, N = 3SE +/- 0.00, N = 328.4124.1121.121. (F9X) gfortran options: -O3 -march=native -funroll-loops -fopenmp
MinAvgMaxEPYC 9575F38.2332.1399.4EPYC 996538.1286.3339.7NVIDIA GH200 144G HBM3e41.4142.4183.4OpenBenchmarking.orgWatts, Fewer Is BetterCloverLeaf 1.3CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 8525 Joules per run.2. EPYC 9965: Approximate power consumption of 8207 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 4747 Joules per run.

NAS Parallel Benchmarks

NPB, NAS Parallel Benchmarks, is a benchmark developed by NASA for high-end computer systems. This test profile currently uses the MPI version of NPB. This test profile offers selecting the different NPB tests/problems and varying problem sizes. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTotal Mop/s, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: IS.DNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996516003200480064008000SE +/- 5.02, N = 3SE +/- 39.21, N = 6SE +/- 113.41, N = 151931.217043.187503.011. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.6
OpenBenchmarking.orgTotal Mop/s Per Watt, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: IS.DNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996591827364514.0630.3140.83
MinAvgMaxEPYC 9575F38.2232.3399.8EPYC 996537.8183.8364.4NVIDIA GH200 144G HBM3e42.5137.4235.8OpenBenchmarking.orgWatts, Fewer Is BetterNAS Parallel Benchmarks 3.4CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 2556 Joules per run.2. EPYC 9965: Approximate power consumption of 2169 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 3434 Joules per run.

srsRAN Project

srsRAN Project is a complete ORAN-native 5G RAN solution created by Software Radio Systems (SRS). The srsRAN Project radio suite was formerly known as srsLTE and can be used for building your own software-defined radio (SDR) 4G/5G mobile network. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMbps, More Is BettersrsRAN Project 23.10.1-20240325Test: PUSCH Processor Benchmark, Throughput ThreadNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F50100150200250SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 575.0129.6224.5MIN: 53-march=native -mavx2 -mavx -msse4.1 -mfma -mavx512f -mavx512cd -mavx512bw -mavx512dq - MIN: 105.8-march=native -mavx2 -mavx -msse4.1 -mfma -mavx512f -mavx512cd -mavx512bw -mavx512dq - MIN: 158.71. (CXX) g++ options: -O3 -fno-trapping-math -fno-math-errno -ldl
OpenBenchmarking.orgMbps Per Watt, More Is BettersrsRAN Project 23.10.1-20240325Test: PUSCH Processor Benchmark, Throughput ThreadEPYC 9965NVIDIA GH200 144G HBM3eEPYC 9575F0.64421.28841.93262.57683.2211.7511.7972.863
MinAvgMaxEPYC 9575F38.378.494.6EPYC 996538.374.083.7NVIDIA GH200 144G HBM3e39.141.743.5OpenBenchmarking.orgWatts, Fewer Is BettersrsRAN Project 23.10.1-20240325CPU Power Consumption Monitor204060801001. EPYC 9575F: Approximate power consumption of 1145 Joules per run.2. EPYC 9965: Approximate power consumption of 1579 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 1516 Joules per run.

GPAW

GPAW is a density-functional theory (DFT) Python code based on the projector-augmented wave (PAW) method and the atomic simulation environment (ASE). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterGPAW 23.6Input: Carbon NanotubeEPYC 9575FEPYC 9965714212835SE +/- 0.21, N = 3SE +/- 0.02, N = 329.8727.561. (CC) gcc options: -shared -lxc -lblas -lmpi
MinAvgMaxEPYC 9575F38.3351.8400.0EPYC 996538.1324.6416.9OpenBenchmarking.orgWatts, Fewer Is BetterGPAW 23.6CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 12195 Joules per run.2. EPYC 9965: Approximate power consumption of 11792 Joules per run.

Input: Carbon Nanotube

NVIDIA GH200 144G HBM3e: The test quit with a non-zero exit status. E: ImportError: numpy.core.multiarray failed to import

Pennant

Pennant is an application focused on hydrodynamics on general unstructured meshes in 2D. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgHydro Cycle Time - Seconds, Fewer Is BetterPennant 1.0.1Test: sedovbigNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965246810SE +/- 0.012489, N = 6SE +/- 0.034501, N = 6SE +/- 0.046313, N = 156.4303046.2211083.8069651. (CXX) g++ options: -fopenmp -lmpi_cxx -lmpi
MinAvgMaxEPYC 9575F38.2248.3399.7EPYC 996537.9197.4381.4NVIDIA GH200 144G HBM3e41.5150.9252.9OpenBenchmarking.orgWatts, Fewer Is BetterPennant 1.0.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 2731 Joules per run.2. EPYC 9965: Approximate power consumption of 2198 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 1811 Joules per run.

Coremark

This is a test of EEMBC CoreMark processor benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgIterations/Sec, More Is BetterCoremark 1.0CoreMark Size 666 - Iterations Per SecondNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965900K1800K2700K3600K4500KSE +/- 24238.24, N = 3SE +/- 4584.67, N = 3SE +/- 2487.95, N = 32396679.003884226.134247085.811. (CC) gcc options: -O2 -lrt" -lrt
OpenBenchmarking.orgIterations/Sec Per Watt, More Is BetterCoremark 1.0CoreMark Size 666 - Iterations Per SecondEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e4K8K12K16K20K12356.0816273.7520070.84
MinAvgMaxEPYC 9575F38.0314.4399.6EPYC 996537.6261.0347.0NVIDIA GH200 144G HBM3e38.8119.4159.8OpenBenchmarking.orgWatts, Fewer Is BetterCoremark 1.0CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 8173 Joules per run.2. EPYC 9965: Approximate power consumption of 6524 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 3105 Joules per run.

OpenFOAM

OpenFOAM is the leading free, open-source software for computational fluid dynamics (CFD). This test profile currently uses the drivaerFastback test case for analyzing automotive aerodynamics or alternatively the older motorBike input. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterOpenFOAM 10Input: drivaerFastback, Small Mesh Size - Mesh TimeNVIDIA GH200 144G HBM3e4812162016.741. (CXX) g++ options: -std=c++14 -O3 -mcpu=native -ftemplate-depth-100 -fPIC -fuse-ld=bfd -Xlinker --add-needed --no-as-needed -lfoamToVTK -ldynamicMesh -llagrangian -lgenericPatchFields -lfileFormats -lOpenFOAM -ldl -lm

SVT-AV1

This is a benchmark of the SVT-AV1 open-source video encoder/decoder. SVT-AV1 was originally developed by Intel as part of their Open Visual Cloud / Scalable Video Technology (SVT). Development of SVT-AV1 has since moved to the Alliance for Open Media as part of upstream AV1 development. SVT-AV1 is a CPU-based multi-threaded video encoder for the AV1 video format with a sample YUV video file. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 4 - Input: Bosphorus 4KEPYC 9965EPYC 9575F3691215SE +/- 0.03, N = 3SE +/- 0.13, N = 410.0212.461. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 4 - Input: Bosphorus 4KEPYC 9575FEPYC 99650.01670.03340.05010.06680.08350.0660.074
MinAvgMaxEPYC 9575F38.3187.3334.2EPYC 996537.8134.8221.3OpenBenchmarking.orgWatts, Fewer Is BetterSVT-AV1 2.0CPU Power Consumption Monitor801602403204001. EPYC 9575F: Approximate power consumption of 3372 Joules per run.2. EPYC 9965: Approximate power consumption of 3011 Joules per run.

OpenFOAM

OpenFOAM is the leading free, open-source software for computational fluid dynamics (CFD). This test profile currently uses the drivaerFastback test case for analyzing automotive aerodynamics or alternatively the older motorBike input. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterOpenFOAM 10Input: drivaerFastback, Small Mesh Size - Execution TimeNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 996591827364537.3725.6025.34-mcpu=native -lfoamToVTK -ldynamicMesh -lfileFormats-m64 -lfiniteVolume -lmeshTools -lparallel -lregionModels-m64 -lfiniteVolume -lmeshTools -lparallel -lregionModels1. (CXX) g++ options: -std=c++14 -O3 -ftemplate-depth-100 -fPIC -fuse-ld=bfd -Xlinker --add-needed --no-as-needed -llagrangian -lgenericPatchFields -lOpenFOAM -ldl -lm
MinAvgMaxEPYC 9575F38.8346.4399.6EPYC 996537.7322.4406.1NVIDIA GH200 144G HBM3e41.1200.0287.7OpenBenchmarking.orgWatts, Fewer Is BetterOpenFOAM 10CPU Power Consumption Monitor110220330440550

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: SlowEPYC 9575FEPYC 99651122334455SE +/- 0.00, N = 3SE +/- 0.05, N = 428.8250.96
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: SlowEPYC 9575FEPYC 99650.04120.08240.12360.16480.2060.0900.183
MinAvgMaxEPYC 9575F38.3319.5384.7EPYC 996538.1278.8375.6OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 7988 Joules per run.2. EPYC 9965: Approximate power consumption of 4461 Joules per run.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: MediumEPYC 9575FEPYC 99651326395265SE +/- 0.02, N = 3SE +/- 0.03, N = 532.6055.69
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: MediumEPYC 9575FEPYC 99650.04570.09140.13710.18280.22850.1040.203
MinAvgMaxEPYC 9575F38.5312.9383.4EPYC 996539.0274.3374.9OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 6885 Joules per run.2. EPYC 9965: Approximate power consumption of 4114 Joules per run.

NAS Parallel Benchmarks

NPB, NAS Parallel Benchmarks, is a benchmark developed by NASA for high-end computer systems. This test profile currently uses the MPI version of NPB. This test profile offers selecting the different NPB tests/problems and varying problem sizes. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTotal Mop/s, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: CG.CNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F14K28K42K56K70KSE +/- 17.96, N = 6SE +/- 985.07, N = 15SE +/- 728.80, N = 1527521.8751319.4665470.231. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.6
OpenBenchmarking.orgTotal Mop/s Per Watt, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: CG.CNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F80160240320400175.40280.56381.14
MinAvgMaxEPYC 996537.8182.9399.7EPYC 9575F38.1171.8399.7NVIDIA GH200 144G HBM3e40.4156.9270.9OpenBenchmarking.orgWatts, Fewer Is BetterNAS Parallel Benchmarks 3.4CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 1683 Joules per run.2. EPYC 9575F: Approximate power consumption of 1225 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 1700 Joules per run.

Blender

Blender is an open-source 3D creation and modeling software project. This test is of Blender's Cycles performance with various sample files. GPU computing via NVIDIA OptiX and NVIDIA CUDA is currently supported as well as HIP for AMD Radeon GPUs and Intel oneAPI for Intel Graphics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 4.1Blend File: Junkshop - Compute: CPU-OnlyEPYC 9575FEPYC 9965510152025SE +/- 0.04, N = 3SE +/- 0.06, N = 419.8411.26
MinAvgMaxEPYC 9575F38.5323.6400.8EPYC 996538.3278.1409.7OpenBenchmarking.orgWatts, Fewer Is BetterBlender 4.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 7767 Joules per run.2. EPYC 9965: Approximate power consumption of 4380 Joules per run.

PyBench

This test profile reports the total time of the different average timed test results from PyBench. PyBench reports average test times for different functions such as BuiltinFunctionCalls and NestedForLoops, with this total result providing a rough estimate as to Python's average performance on a given system. This test profile runs PyBench each time for 20 rounds. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMilliseconds, Fewer Is BetterPyBench 2018-02-16Total For Average Test TimesNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F160320480640800SE +/- 1.53, N = 3SE +/- 2.21, N = 4SE +/- 1.58, N = 4724646482
MinAvgMaxEPYC 9575F38.181.497.6EPYC 996537.873.883.6NVIDIA GH200 144G HBM3e38.941.743.0OpenBenchmarking.orgWatts, Fewer Is BetterPyBench 2018-02-16CPU Power Consumption Monitor204060801001. EPYC 9575F: Approximate power consumption of 1261 Joules per run.2. EPYC 9965: Approximate power consumption of 1458 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 904 Joules per run.

Blender

Blender is an open-source 3D creation and modeling software project. This test is of Blender's Cycles performance with various sample files. GPU computing via NVIDIA OptiX and NVIDIA CUDA is currently supported as well as HIP for AMD Radeon GPUs and Intel oneAPI for Intel Graphics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 4.1Blend File: Fishy Cat - Compute: CPU-OnlyEPYC 9575FEPYC 9965510152025SE +/- 0.04, N = 3SE +/- 0.03, N = 419.1211.02
MinAvgMaxEPYC 9575F38.4322.0398.2EPYC 996537.8271.4392.6OpenBenchmarking.orgWatts, Fewer Is BetterBlender 4.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 7513 Joules per run.2. EPYC 9965: Approximate power consumption of 4207 Joules per run.

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: SlowEPYC 9575FEPYC 99651530456075SE +/- 0.02, N = 4SE +/- 0.16, N = 540.3268.611. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: SlowEPYC 9575FEPYC 99650.06350.1270.19050.2540.31750.1410.282
MinAvgMaxEPYC 9575F38.4285.5368.4EPYC 996538.0243.3354.3OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 5425 Joules per run.2. EPYC 9965: Approximate power consumption of 3114 Joules per run.

Blender

Blender is an open-source 3D creation and modeling software project. This test is of Blender's Cycles performance with various sample files. GPU computing via NVIDIA OptiX and NVIDIA CUDA is currently supported as well as HIP for AMD Radeon GPUs and Intel oneAPI for Intel Graphics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 4.1Blend File: BMW27 - Compute: CPU-OnlyEPYC 9575FEPYC 996548121620SE +/- 0.06, N = 4SE +/- 0.02, N = 514.638.20
MinAvgMaxEPYC 9575F38.3298.8388.7EPYC 996538.3251.7384.4OpenBenchmarking.orgWatts, Fewer Is BetterBlender 4.1CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 5677 Joules per run.2. EPYC 9965: Approximate power consumption of 3222 Joules per run.

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: MediumEPYC 9575FEPYC 99651530456075SE +/- 0.02, N = 4SE +/- 0.17, N = 541.3068.821. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: MediumEPYC 9575FEPYC 99650.06350.1270.19050.2540.31750.1460.282
MinAvgMaxEPYC 9575F38.4283.8372.6EPYC 996538.1244.3356.2OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 5179 Joules per run.2. EPYC 9965: Approximate power consumption of 3127 Joules per run.

SVT-AV1

This is a benchmark of the SVT-AV1 open-source video encoder/decoder. SVT-AV1 was originally developed by Intel as part of their Open Visual Cloud / Scalable Video Technology (SVT). Development of SVT-AV1 has since moved to the Alliance for Open Media as part of upstream AV1 development. SVT-AV1 is a CPU-based multi-threaded video encoder for the AV1 video format with a sample YUV video file. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 13 - Input: Bosphorus 4KEPYC 9965EPYC 9575F60120180240300SE +/- 5.92, N = 15SE +/- 7.85, N = 15244.58283.701. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 13 - Input: Bosphorus 4KEPYC 9965EPYC 9575F0.55761.11521.67282.23042.7882.4452.478
MinAvgMaxEPYC 9575F38.1114.5297.7EPYC 996537.9100.0236.1OpenBenchmarking.orgWatts, Fewer Is BetterSVT-AV1 2.0CPU Power Consumption Monitor701402102803501. EPYC 9575F: Approximate power consumption of 809 Joules per run.2. EPYC 9965: Approximate power consumption of 780 Joules per run.

Intel Open Image Denoise

Open Image Denoise is a denoising library for ray-tracing and part of the Intel oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgImages / Sec, More Is BetterIntel Open Image Denoise 2.2Run: RTLightmap.hdr.4096x4096 - Device: CPU-OnlyEPYC 9575FEPYC 99650.72451.4492.17352.8983.6225SE +/- 0.00, N = 4SE +/- 0.00, N = 52.403.22
OpenBenchmarking.orgImages / Sec Per Watt, More Is BetterIntel Open Image Denoise 2.2Run: RTLightmap.hdr.4096x4096 - Device: CPU-OnlyEPYC 9575FEPYC 99650.00290.00580.00870.01160.01450.0080.013
MinAvgMaxEPYC 9575F38.4292.0383.0EPYC 996538.0250.7355.3OpenBenchmarking.orgWatts, Fewer Is BetterIntel Open Image Denoise 2.2CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 4965 Joules per run.2. EPYC 9965: Approximate power consumption of 3460 Joules per run.

ASTC Encoder

ASTC Encoder (astcenc) is for the Adaptive Scalable Texture Compression (ASTC) format commonly used with OpenGL, OpenGL ES, and Vulkan graphics APIs. This test profile does a coding test of both compression/decompression. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMT/s, More Is BetterASTC Encoder 4.7Preset: ThoroughNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 9965306090120150SE +/- 0.06, N = 4SE +/- 0.06, N = 5SE +/- 0.07, N = 646.4172.38128.761. (CXX) g++ options: -O3 -flto -pthread
OpenBenchmarking.orgMT/s Per Watt, More Is BetterASTC Encoder 4.7Preset: ThoroughNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99650.18180.36360.54540.72720.9090.2860.3290.808
MinAvgMaxEPYC 9575F38.2219.9386.4NVIDIA GH200 144G HBM3e39.1162.3263.1EPYC 996538.2159.3364.0OpenBenchmarking.orgWatts, Fewer Is BetterASTC Encoder 4.7CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 3035 Joules per run.2. NVIDIA GH200 144G HBM3e: Approximate power consumption of 2962 Joules per run.3. EPYC 9965: Approximate power consumption of 1885 Joules per run.

Embree

Intel Embree is a collection of high-performance ray-tracing kernels for execution on CPUs (and GPUs via SYCL) and supporting instruction sets such as SSE, AVX, AVX2, and AVX-512. Embree also supports making use of the Intel SPMD Program Compiler (ISPC). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterEmbree 4.3Binary: Pathtracer ISPC - Model: Asian Dragon ObjEPYC 9575FEPYC 996550100150200250SE +/- 0.06, N = 4SE +/- 0.14, N = 5117.90230.97MIN: 116.73 / MAX: 119.59MIN: 225.48 / MAX: 241.14
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterEmbree 4.3Binary: Pathtracer ISPC - Model: Asian Dragon ObjEPYC 9575FEPYC 99650.36790.73581.10371.47161.83950.5801.635
MinAvgMaxEPYC 9575F38.2203.4400.4EPYC 996538.2141.3395.1OpenBenchmarking.orgWatts, Fewer Is BetterEmbree 4.3CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 3203 Joules per run.2. EPYC 9965: Approximate power consumption of 2063 Joules per run.

Pennant

Pennant is an application focused on hydrodynamics on general unstructured meshes in 2D. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgHydro Cycle Time - Seconds, Fewer Is BetterPennant 1.0.1Test: leblancbigNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99651.00662.01323.01984.02645.033SE +/- 0.007724, N = 7SE +/- 0.077441, N = 15SE +/- 0.017533, N = 74.4737793.2388662.3795131. (CXX) g++ options: -fopenmp -lmpi_cxx -lmpi
MinAvgMaxEPYC 9575F38.2195.5397.8EPYC 996538.0172.4384.3NVIDIA GH200 144G HBM3e41.7140.4265.6OpenBenchmarking.orgWatts, Fewer Is BetterPennant 1.0.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 1551 Joules per run.2. EPYC 9965: Approximate power consumption of 1700 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 1404 Joules per run.

SVT-AV1

This is a benchmark of the SVT-AV1 open-source video encoder/decoder. SVT-AV1 was originally developed by Intel as part of their Open Visual Cloud / Scalable Video Technology (SVT). Development of SVT-AV1 has since moved to the Alliance for Open Media as part of upstream AV1 development. SVT-AV1 is a CPU-based multi-threaded video encoder for the AV1 video format with a sample YUV video file. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 8 - Input: Bosphorus 4KEPYC 9965EPYC 9575F306090120150SE +/- 1.06, N = 6SE +/- 0.58, N = 7104.64121.051. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 8 - Input: Bosphorus 4KEPYC 9575FEPYC 99650.18560.37120.55680.74240.9280.7900.825
MinAvgMaxEPYC 9575F38.1153.2299.1EPYC 996537.9126.8237.7OpenBenchmarking.orgWatts, Fewer Is BetterSVT-AV1 2.0CPU Power Consumption Monitor801602403204001. EPYC 9575F: Approximate power consumption of 1576 Joules per run.2. EPYC 9965: Approximate power consumption of 1437 Joules per run.

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Very FastEPYC 9965EPYC 9575F20406080100SE +/- 0.11, N = 5SE +/- 0.12, N = 670.4882.89
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Very FastEPYC 9575FEPYC 99650.07670.15340.23010.30680.38350.3190.341
MinAvgMaxEPYC 9575F38.3259.7393.5EPYC 996538.7206.8302.8OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 2944 Joules per run.2. EPYC 9965: Approximate power consumption of 2605 Joules per run.

miniFE

MiniFE Finite Element is an application for unstructured implicit finite element codes. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgCG Mflops, More Is BetterminiFE 2.2Problem Size: SmallNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F15K30K45K60K75KSE +/- 346.24, N = 6SE +/- 21.96, N = 6SE +/- 50.38, N = 646786.365294.569114.41. (CXX) g++ options: -O3 -fopenmp -lmpi_cxx -lmpi
OpenBenchmarking.orgCG Mflops Per Watt, More Is BetterminiFE 2.2Problem Size: SmallEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e100200300400500372.32414.89453.14
MinAvgMaxEPYC 9575F38.4185.6399.9EPYC 996537.8157.4341.3NVIDIA GH200 144G HBM3e42.4103.2197.0OpenBenchmarking.orgWatts, Fewer Is BetterminiFE 2.2CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 1949 Joules per run.2. EPYC 9965: Approximate power consumption of 1810 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 1239 Joules per run.

TensorFlow

This is a benchmark of the TensorFlow deep learning framework using the TensorFlow reference benchmarks (tensorflow/benchmarks with tf_cnn_benchmarks.py). Note with the Phoronix Test Suite there is also pts/tensorflow-lite for benchmarking the TensorFlow Lite binaries if desired for complementary metrics. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: AlexNetEPYC 9575FEPYC 9965400800120016002000SE +/- 2.59, N = 4SE +/- 17.23, N = 42050.632085.02
OpenBenchmarking.orgimages/sec Per Watt, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: AlexNetEPYC 9575FEPYC 996536912158.73610.778
MinAvgMaxEPYC 9575F38.5234.7310.5EPYC 996538.3193.4260.2OpenBenchmarking.orgWatts, Fewer Is BetterTensorFlow 2.16.1CPU Power Consumption Monitor801602403204001. EPYC 9575F: Approximate power consumption of 4577 Joules per run.2. EPYC 9965: Approximate power consumption of 3821 Joules per run.

Device: CPU - Batch Size: 256 - Model: AlexNet

NVIDIA GH200 144G HBM3e: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'absl'

Intel Open Image Denoise

Open Image Denoise is a denoising library for ray-tracing and part of the Intel oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgImages / Sec, More Is BetterIntel Open Image Denoise 2.2Run: RT.hdr_alb_nrm.3840x2160 - Device: CPU-OnlyEPYC 9575FEPYC 9965246810SE +/- 0.00, N = 7SE +/- 0.00, N = 75.046.73
OpenBenchmarking.orgImages / Sec Per Watt, More Is BetterIntel Open Image Denoise 2.2Run: RT.hdr_alb_nrm.3840x2160 - Device: CPU-OnlyEPYC 9575FEPYC 99650.00770.01540.02310.03080.03850.0210.034
MinAvgMaxEPYC 9575F38.3240.2389.0EPYC 996538.0199.8360.2OpenBenchmarking.orgWatts, Fewer Is BetterIntel Open Image Denoise 2.2CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 2505 Joules per run.2. EPYC 9965: Approximate power consumption of 1799 Joules per run.

NAS Parallel Benchmarks

NPB, NAS Parallel Benchmarks, is a benchmark developed by NASA for high-end computer systems. This test profile currently uses the MPI version of NPB. This test profile offers selecting the different NPB tests/problems and varying problem sizes. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTotal Mop/s, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: MG.CNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F30K60K90K120K150KSE +/- 28.42, N = 8SE +/- 2066.13, N = 15SE +/- 1509.34, N = 1563701.22112588.99159641.781. (F9X) gfortran options: -O3 -march=native -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -levent_core -levent_pthreads -lm -lz 2. Open MPI 4.1.6
OpenBenchmarking.orgTotal Mop/s Per Watt, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: MG.CNVIDIA GH200 144G HBM3eEPYC 9965EPYC 9575F30060090012001500529.28791.791247.11
MinAvgMaxEPYC 996537.9142.2373.9EPYC 9575F38.1128.0396.0NVIDIA GH200 144G HBM3e41.3120.4274.2OpenBenchmarking.orgWatts, Fewer Is BetterNAS Parallel Benchmarks 3.4CPU Power Consumption Monitor1102203304405501. EPYC 9965: Approximate power consumption of 1138 Joules per run.2. EPYC 9575F: Approximate power consumption of 751 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 978 Joules per run.

Intel Open Image Denoise

Open Image Denoise is a denoising library for ray-tracing and part of the Intel oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgImages / Sec, More Is BetterIntel Open Image Denoise 2.2Run: RT.ldr_alb_nrm.3840x2160 - Device: CPU-OnlyEPYC 9575FEPYC 9965246810SE +/- 0.00, N = 7SE +/- 0.01, N = 75.066.74
OpenBenchmarking.orgImages / Sec Per Watt, More Is BetterIntel Open Image Denoise 2.2Run: RT.ldr_alb_nrm.3840x2160 - Device: CPU-OnlyEPYC 9575FEPYC 99650.00770.01540.02310.03080.03850.0210.034
MinAvgMaxEPYC 9575F38.3237.7390.0EPYC 996538.2200.3356.3OpenBenchmarking.orgWatts, Fewer Is BetterIntel Open Image Denoise 2.2CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 2411 Joules per run.2. EPYC 9965: Approximate power consumption of 1803 Joules per run.

Embree

Intel Embree is a collection of high-performance ray-tracing kernels for execution on CPUs (and GPUs via SYCL) and supporting instruction sets such as SSE, AVX, AVX2, and AVX-512. Embree also supports making use of the Intel SPMD Program Compiler (ISPC). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterEmbree 4.3Binary: Pathtracer ISPC - Model: CrownEPYC 9575FEPYC 996550100150200250SE +/- 0.10, N = 6SE +/- 0.14, N = 8111.29210.40MIN: 109.32 / MAX: 113.75MIN: 203.25 / MAX: 221.34
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterEmbree 4.3Binary: Pathtracer ISPC - Model: CrownEPYC 9575FEPYC 99650.23090.46180.69270.92361.15450.4291.026
MinAvgMaxEPYC 9575F38.4259.5401.1EPYC 996538.0205.1399.1OpenBenchmarking.orgWatts, Fewer Is BetterEmbree 4.3CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 2855 Joules per run.2. EPYC 9965: Approximate power consumption of 1641 Joules per run.

OpenBenchmarking.orgFrames Per Second, More Is BetterEmbree 4.3Binary: Pathtracer ISPC - Model: Asian DragonEPYC 9575FEPYC 996560120180240300SE +/- 0.08, N = 7SE +/- 0.09, N = 9137.41267.70MIN: 135.82 / MAX: 139.45MIN: 262.05 / MAX: 280.59
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterEmbree 4.3Binary: Pathtracer ISPC - Model: Asian DragonEPYC 9575FEPYC 99650.33620.67241.00861.34481.6810.5731.494
MinAvgMaxEPYC 9575F38.3239.9400.8EPYC 996538.0179.2397.6OpenBenchmarking.orgWatts, Fewer Is BetterEmbree 4.3CPU Power Consumption Monitor1102203304405501. EPYC 9575F: Approximate power consumption of 2331 Joules per run.2. EPYC 9965: Approximate power consumption of 1294 Joules per run.

LAMMPS Molecular Dynamics Simulator

LAMMPS is a classical molecular dynamics code, and an acronym for Large-scale Atomic/Molecular Massively Parallel Simulator. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgns/day, More Is BetterLAMMPS Molecular Dynamics Simulator 23Jun2022Model: Rhodopsin ProteinEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e1326395265SE +/- 0.16, N = 11SE +/- 0.48, N = 15SE +/- 0.13, N = 1142.8555.0259.841. (CXX) g++ options: -O3 -lm -ldl
OpenBenchmarking.orgns/day Per Watt, More Is BetterLAMMPS Molecular Dynamics Simulator 23Jun2022Model: Rhodopsin ProteinEPYC 9965EPYC 9575FNVIDIA GH200 144G HBM3e0.21760.43520.65280.87041.0880.4220.4620.967
MinAvgMaxEPYC 996520.5130.4330.9EPYC 9575F38.192.8302.5NVIDIA GH200 144G HBM3e39.961.9280.4OpenBenchmarking.orgWatts, Fewer Is BetterLAMMPS Molecular Dynamics Simulator 23Jun2022CPU Power Consumption Monitor801602403204001. EPYC 9965: Approximate power consumption of 1052 Joules per run.2. EPYC 9575F: Approximate power consumption of 557 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 360 Joules per run.

SVT-AV1

This is a benchmark of the SVT-AV1 open-source video encoder/decoder. SVT-AV1 was originally developed by Intel as part of their Open Visual Cloud / Scalable Video Technology (SVT). Development of SVT-AV1 has since moved to the Alliance for Open Media as part of upstream AV1 development. SVT-AV1 is a CPU-based multi-threaded video encoder for the AV1 video format with a sample YUV video file. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 12 - Input: Bosphorus 4KEPYC 9965EPYC 9575F60120180240300SE +/- 1.44, N = 9SE +/- 1.47, N = 9248.70294.411. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 12 - Input: Bosphorus 4KEPYC 9965EPYC 9575F0.58121.16241.74362.32482.9062.5242.583
MinAvgMaxEPYC 9575F38.2114.0322.8EPYC 996537.998.5241.1OpenBenchmarking.orgWatts, Fewer Is BetterSVT-AV1 2.0CPU Power Consumption Monitor801602403204001. EPYC 9575F: Approximate power consumption of 798 Joules per run.2. EPYC 9965: Approximate power consumption of 745 Joules per run.

Primesieve

Primesieve generates prime numbers using a highly optimized sieve of Eratosthenes implementation. Primesieve primarily benchmarks the CPU's L1/L2 cache performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterPrimesieve 12.1Length: 1e12NVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99650.60011.20021.80032.40043.0005SE +/- 0.003, N = 10SE +/- 0.003, N = 11SE +/- 0.003, N = 122.6672.1231.3961. (CXX) g++ options: -O3
MinAvgMaxEPYC 9575F38.5146.3375.7EPYC 996538.2119.3378.8NVIDIA GH200 144G HBM3e41.8111.4254.2OpenBenchmarking.orgWatts, Fewer Is BetterPrimesieve 12.1CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 891 Joules per run.2. EPYC 9965: Approximate power consumption of 636 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 769 Joules per run.

Parallel BZIP2 Compression

This test measures the time needed to compress a file (FreeBSD-13.0-RELEASE-amd64-memstick.img) using Parallel BZIP2 compression. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterParallel BZIP2 Compression 1.1.13FreeBSD-13.0-RELEASE-amd64-memstick.img CompressionNVIDIA GH200 144G HBM3eEPYC 9575FEPYC 99650.37860.75721.13581.51441.893SE +/- 0.004252, N = 15SE +/- 0.009364, N = 15SE +/- 0.009424, N = 151.6827971.2634080.8912381. (CXX) g++ options: -O2 -pthread -lbz2 -lpthread
MinAvgMaxEPYC 9575F38.2104.3361.5EPYC 996538.080.5240.4NVIDIA GH200 144G HBM3e40.068.7207.5OpenBenchmarking.orgWatts, Fewer Is BetterParallel BZIP2 Compression 1.1.13CPU Power Consumption Monitor1002003004005001. EPYC 9575F: Approximate power consumption of 556 Joules per run.2. EPYC 9965: Approximate power consumption of 397 Joules per run.3. NVIDIA GH200 144G HBM3e: Approximate power consumption of 399 Joules per run.

Meta Performance Per Watts

OpenBenchmarking.orgPerformance Per Watts, More Is BetterMeta Performance Per WattsPerformance Per WattsNVIDIA GH200 144G HBM3e50K100K150K200K250K227401.82

CPU Power Consumption Monitor

OpenBenchmarking.orgWattsCPU Power Consumption MonitorPhoronix Test Suite System MonitoringEPYC 9575FEPYC 9965NVIDIA GH200 144G HBM3e80160240320400Min: 19.17 / Avg: 313.09 / Max: 403.08Min: 19.04 / Avg: 275.17 / Max: 461.59Min: 38.42 / Avg: 170.19 / Max: 300.47

148 Results Shown

OpenVKL
RELION
NWChem
High Performance Conjugate Gradient
OpenSSL:
  RSA4096:
    verify/s
    sign/s
Speedb
BRL-CAD
SecureMark
Xmrig
Stockfish
Timed Linux Kernel Compilation
Xcompact3d Incompact3d
LuxCoreRender
OpenSSL
LuxCoreRender
OpenSSL:
  AES-128-GCM
  SHA256
  AES-256-GCM
  ChaCha20
  ChaCha20-Poly1305
OSPRay
TensorFlow
Timed Node.js Compilation
ClickHouse:
  100M Rows Hits Dataset, Third Run
  100M Rows Hits Dataset, Second Run
  100M Rows Hits Dataset, First Run / Cold Cache
LAMMPS Molecular Dynamics Simulator
Timed Gem5 Compilation
OSPRay
Timed LLVM Compilation
FFmpeg
OpenRadioss
OpenFOAM:
  drivaerFastback, Medium Mesh Size - Execution Time
  drivaerFastback, Medium Mesh Size - Mesh Time
ASKAP:
  tConvolve MPI - Gridding
  tConvolve MPI - Degridding
Numpy Benchmark
Blender
Timed Godot Game Engine Compilation
OpenRadioss
TensorFlow
QMCPACK
Zstd Compression:
  19, Long Mode - Decompression Speed
  19, Long Mode - Compression Speed
OSPRay Studio
OSPRay
Zstd Compression:
  19 - Decompression Speed
  19 - Compression Speed
libavif avifenc
Memcached
OpenVINO:
  Face Detection FP16-INT8 - CPU:
    ms
    FPS
  Noise Suppression Poconet-Like FP16 - CPU:
    ms
    FPS
  Person Detection FP16 - CPU:
    ms
    FPS
OSPRay Studio
OpenVINO:
  Machine Translation EN To DE FP16 - CPU:
    ms
    FPS
  Person Vehicle Bike Detection FP16 - CPU:
    ms
    FPS
  Road Segmentation ADAS FP16-INT8 - CPU:
    ms
    FPS
  Face Detection Retail FP16-INT8 - CPU:
    ms
    FPS
  Age Gender Recognition Retail 0013 FP16-INT8 - CPU:
    ms
    FPS
  Vehicle Detection FP16-INT8 - CPU:
    ms
    FPS
OSPRay Studio
OpenVINO:
  Weld Porosity Detection FP16-INT8 - CPU:
    ms
    FPS
GraphicsMagick
Speedb
GraphicsMagick:
  Enhanced
  Noise-Gaussian
Liquid-DSP
RocksDB
NAMD
OSPRay Studio:
  3 - 4K - 32 - Path Tracer - CPU
  3 - 4K - 16 - Path Tracer - CPU
Helsing
OSPRay Studio
QuantLib
miniBUDE:
  OpenMP - BM2:
    Billion Interactions/s
    GFInst/s
NAMD
WebP Image Encode
Blender
NAS Parallel Benchmarks
Liquid-DSP
John The Ripper
Liquid-DSP:
  256 - 256 - 57
  128 - 256 - 512
  256 - 256 - 32
John The Ripper:
  Blowfish
  bcrypt
Liquid-DSP
Blender
7-Zip Compression:
  Decompression Rating
  Compression Rating
ASTC Encoder
OpenRadioss
ASTC Encoder
Algebraic Multi-Grid Benchmark
Primesieve
NAS Parallel Benchmarks
srsRAN Project
GROMACS
CloverLeaf
NAS Parallel Benchmarks
srsRAN Project
GPAW
Pennant
Coremark
OpenFOAM
SVT-AV1
OpenFOAM
uvg266:
  Bosphorus 4K - Slow
  Bosphorus 4K - Medium
NAS Parallel Benchmarks
Blender
PyBench
Blender
Kvazaar
Blender
Kvazaar
SVT-AV1
Intel Open Image Denoise
ASTC Encoder
Embree
Pennant
SVT-AV1
uvg266
miniFE
TensorFlow
Intel Open Image Denoise
NAS Parallel Benchmarks
Intel Open Image Denoise
Embree:
  Pathtracer ISPC - Crown
  Pathtracer ISPC - Asian Dragon
LAMMPS Molecular Dynamics Simulator
SVT-AV1
Primesieve
Parallel BZIP2 Compression
Meta Performance Per Watts:
  Performance Per Watts
  Phoronix Test Suite System Monitoring

EPYC 9965

Processor: AMD EPYC 9965 192-Core @ 2.25GHz (192 Cores / 384 Threads), Motherboard: AMD VOLCANO (RVOT1000D BIOS), Chipset: AMD Device 153a, Memory: 12 x 64GB DDR5-6000MT/s Samsung M321R8GA0PB1-CCPKC, Disk: 2 x 1920GB KIOXIA KCD8XPUG1T92, Graphics: ASPEED, Network: Broadcom NetXtreme BCM5720 PCIe

OS: Ubuntu 24.04, Kernel: 6.10.0-phx (x86_64), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xb101018
Java Notes: OpenJDK Runtime Environment (build 21.0.3-ea+7-Ubuntu-1build1)
Python Notes: Python 3.12.2
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS; IBPB: conditional; STIBP: always-on; RSB filling; PBRSB-eIBRS: Not affected; BHI: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 4 October 2024 14:32 by user phoronix.

EPYC 9575F

Processor: AMD EPYC 9575F 64-Core @ 3.30GHz (64 Cores / 128 Threads), Motherboard: AMD VOLCANO (RVOT1000D BIOS), Chipset: AMD Device 153a, Memory: 12 x 64GB DDR5-6000MT/s Samsung M321R8GA0PB1-CCPKC, Disk: 2 x 1920GB KIOXIA KCD8XPUG1T92, Graphics: ASPEED, Network: Broadcom NetXtreme BCM5720 PCIe

OS: Ubuntu 24.04, Kernel: 6.10.0-phx (x86_64), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-OiuXZC/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xb002110
Java Notes: OpenJDK Runtime Environment (build 21.0.3-ea+7-Ubuntu-1build1)
Python Notes: Python 3.12.2
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS; IBPB: conditional; STIBP: always-on; RSB filling; PBRSB-eIBRS: Not affected; BHI: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 2 October 2024 15:16 by user phoronix.

NVIDIA GH200 144G HBM3e

Processor: ARMv8 Neoverse-V2 @ 3.47GHz (72 Cores), Motherboard: Pegatron JIMBO P4352 (00022432 BIOS), Memory: 1 x 480GB LPDDR5-6400MT/s NVIDIA 699-2G530-0236-RC1, Disk: 1000GB CT1000T700SSD3, Graphics: NVIDIA GH200 144G HBM3e 143GB, Network: 2 x Intel X550

OS: Ubuntu 24.04, Kernel: 6.8.0-47-generic-64k (aarch64), Display Driver: NVIDIA, OpenCL: OpenCL 3.0 CUDA 12.6.65, Compiler: GCC 13.2.0 + CUDA 11.8, File-System: ext4, Screen Resolution: 1920x1200

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-dIwDw0/gcc-13-13.2.0/debian/tmp-nvptx/usr --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto --without-cuda-driver -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Python Notes: Python 3.12.6
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 28 October 2024 17:33 by user x.