AMD EPYC Zen 4C power efficiency benchmarks by Michael Larabel for a future article.
EPYC 8534PN - Zen 4C Processor: AMD EPYC 8534PN 64-Core @ 2.00GHz (64 Cores / 128 Threads), Motherboard: AMD Cinnabar (RCB1009C BIOS), Chipset: AMD Device 14a4, Memory: 6 x 32 GB DRAM-4800MT/s Samsung M321R4GA0BB0-CQKMG, Disk: 1000GB INTEL SSDPE2KX010T8, Graphics: llvmpipe, Network: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, OpenGL: 4.5 Mesa 23.2.1-1ubuntu3.1 (LLVM 15.0.7 256 bits), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
Kernel Notes: Transparent Huge Pages: madviseCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/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-build-config=bootstrap-lto-lean --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 -vProcessor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xaa00212Java Notes: OpenJDK Runtime Environment (build 11.0.21+9-post-Ubuntu-0ubuntu123.10)Python Notes: Python 3.11.6Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of Safe RET + 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 + srbds: Not affected + tsx_async_abort: Not affected
EPYC 8324PN - Zen 4C Changed Processor to AMD EPYC 8534PN 32-Core @ 2.05GHz (32 Cores / 64 Threads) .
EPYC 8324P - Zen 4C Changed Processor to AMD EPYC 8324P 32-Core @ 2.65GHz (32 Cores / 64 Threads) .
EPYC 8324P - Zen 4C, 155W Processor: AMD EPYC 8324P 32-Core @ 2.65GHz (32 Cores / 64 Threads), Motherboard: AMD Cinnabar (RCB1009C BIOS), Chipset: AMD Device 14a4, Memory: 6 x 32 GB DRAM-4800MT/s Samsung M321R4GA0BB0-CQKMG, Disk: 1000GB INTEL SSDPE2KX010T8, Graphics: ASPEEDNetwork: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
EPYC 7601 - Zen 1 Processor: AMD EPYC 7601 32-Core @ 2.20GHz (32 Cores / 64 Threads)Motherboard: TYAN B8026T70AE24HR (V1.02.B10 BIOS)Chipset: AMD 17hMemory: 8 x 16 GB 2667MT/s Samsung M393A2K40BB2-CTDDisk: 1000GB INTEL SSDPE2KX010T8 + 280GB INTEL SSDPE21D280GAGraphics: ASPEED, Monitor: VE228Network: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, OpenGL: 4.5 Mesa 23.2.1-1ubuntu3.1 (LLVM 15.0.7 256 bits), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1080
Kernel Notes: Transparent Huge Pages: madviseCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/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-build-config=bootstrap-lto-lean --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 -vProcessor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0x800126eJava Notes: OpenJDK Runtime Environment (build 11.0.21+9-post-Ubuntu-0ubuntu123.10)Python Notes: Python 3.11.6Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Mitigation of untrained return thunk; SMT vulnerable + spec_rstack_overflow: Mitigation of Safe RET + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional STIBP: disabled RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected
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.org FPS, More Is Better OpenVINO 2023.2.dev Model: Vehicle Detection FP16-INT8 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 600 1200 1800 2400 3000 SE +/- 0.56, N = 3 SE +/- 1.22, N = 3 SE +/- 0.21, N = 3 SE +/- 0.65, N = 3 SE +/- 0.10, N = 3 2816.09 1527.32 1860.65 1710.42 296.42 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
OpenBenchmarking.org FPS, More Is Better OpenVINO 2023.2.dev Model: Face Detection Retail FP16-INT8 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 2K 4K 6K 8K 10K SE +/- 9.00, N = 3 SE +/- 1.85, N = 3 SE +/- 1.73, N = 3 SE +/- 1.37, N = 3 SE +/- 0.47, N = 3 8310.71 4405.62 5418.92 4977.13 878.71 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
OpenBenchmarking.org FPS, More Is Better OpenVINO 2023.2.dev Model: Machine Translation EN To DE FP16 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 60 120 180 240 300 SE +/- 0.54, N = 3 SE +/- 0.05, N = 3 SE +/- 0.19, N = 3 SE +/- 0.04, N = 3 SE +/- 0.16, N = 3 295.61 163.09 194.71 180.32 38.63 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
Llama.cpp Llama.cpp is a port of Facebook's LLaMA model in C/C++ developed by Georgi Gerganov. Llama.cpp allows the inference of LLaMA and other supported models in C/C++. For CPU inference Llama.cpp supports AVX2/AVX-512, ARM NEON, and other modern ISAs along with features like OpenBLAS usage. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Tokens Per Second, More Is Better Llama.cpp b1808 Model: llama-2-70b-chat.Q5_0.gguf EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.7718 1.5436 2.3154 3.0872 3.859 SE +/- 0.01, N = 3 SE +/- 0.00, N = 3 SE +/- 0.00, N = 3 SE +/- 0.00, N = 3 3.11 3.43 2.93 0.46 1. (CXX) g++ options: -std=c++11 -fPIC -O3 -pthread -march=native -mtune=native -lopenblas Tokens Per Second Per Watt
OpenBenchmarking.org Tokens Per Second Per Watt, More Is Better Llama.cpp b1808 Model: llama-2-70b-chat.Q5_0.gguf EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.007 0.014 0.021 0.028 0.035 0.031 0.027 0.028 0.003 CPU Power Consumption
Min Avg Max EPYC 8324PN - Zen 4C 12.8 101.1 109.8 EPYC 8324P - Zen 4C 13.0 126.4 141.3 EPYC 8324P - Zen 4C, 155W 12.2 104.0 121.2 EPYC 7601 - Zen 1 64.0 134.9 177.0 OpenBenchmarking.org Watts, Fewer Is Better Llama.cpp b1808 CPU Power Consumption Monitor 50 100 150 200 250
Result
OpenBenchmarking.org Tokens Per Second, More Is Better Llama.cpp b1808 Model: llama-2-13b.Q4_0.gguf EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4 8 12 16 20 SE +/- 0.05, N = 3 SE +/- 0.02, N = 3 SE +/- 0.04, N = 3 SE +/- 0.03, N = 12 17.06 17.93 17.09 2.49 1. (CXX) g++ options: -std=c++11 -fPIC -O3 -pthread -march=native -mtune=native -lopenblas Tokens Per Second Per Watt
OpenBenchmarking.org Tokens Per Second Per Watt, More Is Better Llama.cpp b1808 Model: llama-2-13b.Q4_0.gguf EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0419 0.0838 0.1257 0.1676 0.2095 0.186 0.174 0.179 0.019 CPU Power Consumption
Min Avg Max EPYC 8324PN - Zen 4C 12.6 91.5 104.2 EPYC 8324P - Zen 4C 12.9 103.1 130.5 EPYC 8324P - Zen 4C, 155W 13.2 95.6 116.7 EPYC 7601 - Zen 1 64.3 132.7 166.7 OpenBenchmarking.org Watts, Fewer Is Better Llama.cpp b1808 CPU Power Consumption Monitor 50 100 150 200 250
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.org GFInst/s, More Is Better miniBUDE 20210901 Implementation: OpenMP - Input Deck: BM2 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 500 1000 1500 2000 2500 SE +/- 0.34, N = 3 SE +/- 0.10, N = 3 SE +/- 0.47, N = 3 SE +/- 0.17, N = 3 SE +/- 0.44, N = 3 2489.20 1276.47 1532.27 1425.32 355.99 1. (CC) gcc options: -std=c99 -Ofast -ffast-math -fopenmp -march=native -lm
Result
OpenBenchmarking.org Billion Interactions/s, More Is Better miniBUDE 20210901 Implementation: OpenMP - Input Deck: BM2 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 20 40 60 80 100 SE +/- 0.01, N = 3 SE +/- 0.00, N = 3 SE +/- 0.02, N = 3 SE +/- 0.01, N = 3 SE +/- 0.02, N = 3 99.57 51.06 61.29 57.01 14.24 1. (CC) gcc options: -std=c99 -Ofast -ffast-math -fopenmp -march=native -lm Billion Interactions/s Per Watt
OpenBenchmarking.org Billion Interactions/s Per Watt, More Is Better miniBUDE 20210901 Implementation: OpenMP - Input Deck: BM2 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1859 0.3718 0.5577 0.7436 0.9295 0.826 0.645 0.570 0.606 0.084 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.8 120.5 131.6 EPYC 8324PN - Zen 4C 12.1 79.2 88.9 EPYC 8324P - Zen 4C 12.4 107.5 121.8 EPYC 8324P - Zen 4C, 155W 13.8 94.1 106.1 EPYC 7601 - Zen 1 65.5 169.5 175.3 OpenBenchmarking.org Watts, Fewer Is Better miniBUDE 20210901 CPU Power Consumption Monitor 50 100 150 200 250
Result
OpenBenchmarking.org Billion Interactions/s, More Is Better miniBUDE 20210901 Implementation: OpenMP - Input Deck: BM1 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 20 40 60 80 100 SE +/- 0.02, N = 6 SE +/- 0.06, N = 4 SE +/- 0.01, N = 4 SE +/- 0.03, N = 4 SE +/- 0.02, N = 3 99.65 51.12 61.15 57.10 14.33 1. (CC) gcc options: -std=c99 -Ofast -ffast-math -fopenmp -march=native -lm Billion Interactions/s Per Watt
OpenBenchmarking.org Billion Interactions/s Per Watt, More Is Better miniBUDE 20210901 Implementation: OpenMP - Input Deck: BM1 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.2446 0.4892 0.7338 0.9784 1.223 1.087 0.746 0.684 0.718 0.087 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.4 91.7 132.3 EPYC 8324PN - Zen 4C 13.1 68.5 88.4 EPYC 8324P - Zen 4C 12.7 89.5 122.9 EPYC 8324P - Zen 4C, 155W 14.1 79.5 107.9 EPYC 7601 - Zen 1 66.1 164.9 176.2 OpenBenchmarking.org Watts, Fewer Is Better miniBUDE 20210901 CPU Power Consumption Monitor 50 100 150 200 250
OpenBenchmarking.org GFInst/s, More Is Better miniBUDE 20210901 Implementation: OpenMP - Input Deck: BM1 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 500 1000 1500 2000 2500 SE +/- 0.47, N = 6 SE +/- 1.51, N = 4 SE +/- 0.21, N = 4 SE +/- 0.86, N = 4 SE +/- 0.43, N = 3 2491.17 1278.03 1528.85 1427.41 358.25 1. (CC) gcc options: -std=c99 -Ofast -ffast-math -fopenmp -march=native -lm
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 13 26 39 52 65 SE +/- 0.1049, N = 3 SE +/- 0.0604, N = 3 SE +/- 0.0215, N = 3 SE +/- 0.0667, N = 3 SE +/- 0.0850, N = 3 56.8419 32.1289 38.8202 35.7929 8.5097
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.
Result
OpenBenchmarking.org byte/s, More Is Better OpenSSL 3.1 Algorithm: ChaCha20-Poly1305 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40000M 80000M 120000M 160000M 200000M SE +/- 52537274.39, N = 3 SE +/- 36167959.31, N = 3 SE +/- 5327895.48, N = 3 SE +/- 3407926.56, N = 3 SE +/- 21793844.67, N = 3 196012573853 101065023123 113105892877 112344083843 30341046637 1. (CC) gcc options: -pthread -m64 -O3 -ldl byte/s Per Watt
OpenBenchmarking.org byte/s Per Watt, More Is Better OpenSSL 3.1 Algorithm: ChaCha20-Poly1305 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 300M 600M 900M 1200M 1500M 1450412054.66 1181271903.83 947613075.92 1067969605.53 177417166.50 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.2 135.1 142.3 EPYC 8324PN - Zen 4C 12.7 85.6 95.5 EPYC 8324P - Zen 4C 13.3 119.4 130.2 EPYC 8324P - Zen 4C, 155W 13.9 105.2 110.7 EPYC 7601 - Zen 1 65.9 171.0 177.9 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 3.1 CPU Power Consumption Monitor 50 100 150 200 250
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.org FPS, More Is Better OpenVINO 2023.2.dev Model: Road Segmentation ADAS FP16-INT8 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 200 400 600 800 1000 SE +/- 3.62, N = 3 SE +/- 0.24, N = 3 SE +/- 0.09, N = 3 SE +/- 0.14, N = 3 SE +/- 0.05, N = 3 934.59 520.24 633.21 579.34 149.01 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
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.
Result
OpenBenchmarking.org Items Per Second, More Is Better OSPRay 2.12 Benchmark: gravity_spheres_volume/dim_512/scivis/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4 8 12 16 20 SE +/- 0.00073, N = 3 SE +/- 0.01310, N = 3 SE +/- 0.01463, N = 3 SE +/- 0.02613, N = 3 SE +/- 0.01198, N = 3 14.61310 7.66532 8.83622 8.45634 2.45276 Items Per Second Per Watt
OpenBenchmarking.org Items Per Second Per Watt, More Is Better OSPRay 2.12 Benchmark: gravity_spheres_volume/dim_512/scivis/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0293 0.0586 0.0879 0.1172 0.1465 0.130 0.101 0.083 0.095 0.016 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.8 112.7 134.3 EPYC 8324PN - Zen 4C 12.4 75.5 89.1 EPYC 8324P - Zen 4C 11.4 106.4 122.1 EPYC 8324P - Zen 4C, 155W 13.7 89.2 107.3 EPYC 7601 - Zen 1 65.1 155.2 177.3 OpenBenchmarking.org Watts, Fewer Is Better OSPRay 2.12 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org byte/s, More Is Better OpenSSL 3.1 Algorithm: ChaCha20 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 60000M 120000M 180000M 240000M 300000M SE +/- 8141159.25, N = 3 SE +/- 4106091.67, N = 3 SE +/- 3614035.56, N = 3 SE +/- 2748202.21, N = 3 SE +/- 18792715.18, N = 3 284696293373 147602077650 159447675517 159431008620 47935580060 1. (CC) gcc options: -pthread -m64 -O3 -ldl byte/s Per Watt
OpenBenchmarking.org byte/s Per Watt, More Is Better OpenSSL 3.1 Algorithm: ChaCha20 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 400M 800M 1200M 1600M 2000M 2095501066.50 1694474523.22 1370606684.98 1474962096.07 280245644.36 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.3 135.9 143.4 EPYC 8324PN - Zen 4C 12.5 87.1 90.7 EPYC 8324P - Zen 4C 13.3 116.3 131.9 EPYC 8324P - Zen 4C, 155W 13.7 108.1 113.9 EPYC 7601 - Zen 1 65.3 171.0 177.6 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 3.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org images/sec, More Is Better TensorFlow 2.12 Device: CPU - Batch Size: 16 - Model: ResNet-50 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 12 24 36 48 60 SE +/- 0.03, N = 3 SE +/- 0.09, N = 3 SE +/- 0.08, N = 3 SE +/- 0.04, N = 3 SE +/- 0.07, N = 3 48.56 47.77 51.70 51.03 8.77 images/sec Per Watt
OpenBenchmarking.org images/sec Per Watt, More Is Better TensorFlow 2.12 Device: CPU - Batch Size: 16 - Model: ResNet-50 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1314 0.2628 0.3942 0.5256 0.657 0.519 0.584 0.553 0.551 0.068 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.0 93.6 109.8 EPYC 8324PN - Zen 4C 6.7 81.9 97.1 EPYC 8324P - Zen 4C 13.0 93.5 112.6 EPYC 8324P - Zen 4C, 155W 13.2 92.6 109.9 EPYC 7601 - Zen 1 64.2 129.4 142.8 OpenBenchmarking.org Watts, Fewer Is Better TensorFlow 2.12 CPU Power Consumption Monitor 40 80 120 160 200
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.
Result
OpenBenchmarking.org Items Per Second, More Is Better OSPRay 2.12 Benchmark: gravity_spheres_volume/dim_512/ao/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4 8 12 16 20 SE +/- 0.01491, N = 3 SE +/- 0.02953, N = 3 SE +/- 0.02105, N = 3 SE +/- 0.02664, N = 3 SE +/- 0.00308, N = 3 15.05960 7.97910 9.14257 8.78898 2.59961 Items Per Second Per Watt
OpenBenchmarking.org Items Per Second Per Watt, More Is Better OSPRay 2.12 Benchmark: gravity_spheres_volume/dim_512/ao/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0302 0.0604 0.0906 0.1208 0.151 0.134 0.106 0.086 0.099 0.017 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 112.5 134.5 EPYC 8324PN - Zen 4C 13.1 75.3 89.0 EPYC 8324P - Zen 4C 13.5 105.7 121.6 EPYC 8324P - Zen 4C, 155W 13.7 88.9 106.3 EPYC 7601 - Zen 1 65.9 154.5 177.4 OpenBenchmarking.org Watts, Fewer Is Better OSPRay 2.12 CPU Power Consumption Monitor 50 100 150 200 250
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.org FPS, More Is Better OpenVINO 2023.2.dev Model: Handwritten English Recognition FP16-INT8 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 200 400 600 800 1000 SE +/- 0.42, N = 3 SE +/- 0.39, N = 3 SE +/- 0.64, N = 3 SE +/- 5.36, N = 8 SE +/- 0.74, N = 3 1086.68 555.71 688.60 621.82 188.86 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
OpenBenchmarking.org FPS, More Is Better OpenVINO 2023.2.dev Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 15K 30K 45K 60K 75K SE +/- 93.27, N = 3 SE +/- 33.66, N = 3 SE +/- 23.70, N = 3 SE +/- 37.85, N = 3 SE +/- 8.89, N = 3 71960.41 43375.55 52709.19 49194.55 13039.70 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
OpenBenchmarking.org FPS, More Is Better OpenVINO 2023.2.dev Model: Person Vehicle Bike Detection FP16 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 500 1000 1500 2000 2500 SE +/- 2.30, N = 3 SE +/- 1.32, N = 3 SE +/- 1.86, N = 3 SE +/- 1.13, N = 3 SE +/- 1.37, N = 3 2547.73 1424.53 1701.18 1563.36 464.48 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
ACES DGEMM This is a multi-threaded DGEMM benchmark. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org GFLOP/s, More Is Better ACES DGEMM 1.0 Sustained Floating-Point Rate EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 5 10 15 20 25 SE +/- 0.111892, N = 5 SE +/- 0.073791, N = 3 SE +/- 0.076181, N = 15 SE +/- 0.135732, N = 3 SE +/- 0.031226, N = 3 21.138819 11.517272 11.645616 11.434849 3.886776 1. (CC) gcc options: -O3 -march=native -fopenmp GFLOP/s Per Watt
OpenBenchmarking.org GFLOP/s Per Watt, More Is Better ACES DGEMM 1.0 Sustained Floating-Point Rate EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.054 0.108 0.162 0.216 0.27 0.240 0.170 0.133 0.140 0.025 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.9 88.1 130.7 EPYC 8324PN - Zen 4C 12.6 67.9 89.1 EPYC 8324P - Zen 4C 12.3 87.3 115.0 EPYC 8324P - Zen 4C, 155W 13.5 81.5 109.9 EPYC 7601 - Zen 1 64.6 154.9 170.2 OpenBenchmarking.org Watts, Fewer Is Better ACES DGEMM 1.0 CPU Power Consumption Monitor 50 100 150 200 250
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 300 600 900 1200 1500 SE +/- 0.39, N = 3 SE +/- 0.29, N = 3 SE +/- 1.14, N = 3 SE +/- 0.55, N = 3 SE +/- 2.58, N = 3 1287.88 674.93 818.31 752.59 246.64
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.
Result
OpenBenchmarking.org byte/s, More Is Better OpenSSL 3.1 Algorithm: AES-128-GCM EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 110000M 220000M 330000M 440000M 550000M SE +/- 350336826.28, N = 3 SE +/- 43613214.37, N = 3 SE +/- 338919345.53, N = 3 SE +/- 1389668244.12, N = 3 SE +/- 228892930.90, N = 3 493434496250 251415558370 308207308437 281191878567 97707034820 1. (CC) gcc options: -pthread -m64 -O3 -ldl byte/s Per Watt
OpenBenchmarking.org byte/s Per Watt, More Is Better OpenSSL 3.1 Algorithm: AES-128-GCM EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 800M 1600M 2400M 3200M 4000M 3626240465.70 2927424070.20 2612982097.09 2743728420.23 559684188.57 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.3 136.1 141.2 EPYC 8324PN - Zen 4C 12.9 85.9 94.2 EPYC 8324P - Zen 4C 13.3 118.0 131.8 EPYC 8324P - Zen 4C, 155W 13.6 102.5 112.1 EPYC 7601 - Zen 1 65.9 174.6 178.2 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 3.1 CPU Power Consumption Monitor 50 100 150 200 250
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.org FPS, More Is Better OpenVINO 2023.2.dev Model: Person Detection FP16 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 SE +/- 0.77, N = 3 SE +/- 0.65, N = 3 SE +/- 0.51, N = 3 SE +/- 0.22, N = 3 SE +/- 0.18, N = 3 202.29 131.24 150.74 144.27 42.43 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 130 260 390 520 650 SE +/- 0.38, N = 3 SE +/- 0.28, N = 3 SE +/- 1.61, N = 3 SE +/- 1.23, N = 3 SE +/- 1.39, N = 3 613.23 320.86 383.70 357.30 128.67
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.
Result
OpenBenchmarking.org byte/s, More Is Better OpenSSL 3.1 Algorithm: AES-256-GCM EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 90000M 180000M 270000M 360000M 450000M SE +/- 171818503.35, N = 3 SE +/- 288672373.67, N = 3 SE +/- 72395520.64, N = 3 SE +/- 104993783.37, N = 3 SE +/- 220160896.72, N = 3 424400038207 216435660207 265303113457 242830526873 89976610817 1. (CC) gcc options: -pthread -m64 -O3 -ldl byte/s Per Watt
OpenBenchmarking.org byte/s Per Watt, More Is Better OpenSSL 3.1 Algorithm: AES-256-GCM EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 700M 1400M 2100M 2800M 3500M 3123938397.23 2539685896.32 2258891761.22 2396310835.77 516298912.20 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.4 135.9 141.5 EPYC 8324PN - Zen 4C 13.0 85.2 93.8 EPYC 8324P - Zen 4C 13.3 117.4 132.0 EPYC 8324P - Zen 4C, 155W 13.7 101.3 115.1 EPYC 7601 - Zen 1 65.3 174.3 178.2 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 3.1 CPU Power Consumption Monitor 50 100 150 200 250
OpenBenchmarking.org sign/s, More Is Better OpenSSL 3.1 Algorithm: RSA4096 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 5K 10K 15K 20K 25K SE +/- 4.13, N = 3 SE +/- 3.00, N = 3 SE +/- 2.49, N = 3 SE +/- 3.08, N = 3 SE +/- 12.09, N = 3 21064.1 11909.2 15040.7 13565.8 4497.1 1. (CC) gcc options: -pthread -m64 -O3 -ldl
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 400 800 1200 1600 2000 SE +/- 0.40, N = 3 SE +/- 0.37, N = 3 SE +/- 0.25, N = 3 SE +/- 0.46, N = 3 SE +/- 6.61, N = 3 556.15 493.99 410.47 443.43 1840.95 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.6 103.7 123.1 EPYC 8324PN - Zen 4C 12.8 69.8 81.3 EPYC 8324P - Zen 4C 13.0 89.2 114.7 EPYC 8324P - Zen 4C, 155W 13.1 81.2 97.7 EPYC 7601 - Zen 1 64.6 142.0 172.7 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org H/s, More Is Better Xmrig 6.21 Variant: GhostRider - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 900 1800 2700 3600 4500 SE +/- 6.81, N = 3 SE +/- 16.58, N = 3 SE +/- 7.57, N = 3 SE +/- 19.57, N = 3 SE +/- 12.26, N = 3 4374.0 3938.5 4043.2 4031.7 1042.7 1. (CXX) g++ options: -fexceptions -fno-rtti -maes -O3 -Ofast -static-libgcc -static-libstdc++ -rdynamic -lssl -lcrypto -luv -lpthread -lrt -ldl -lhwloc H/s Per Watt
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: GhostRider - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 11 22 33 44 55 35.481 50.340 41.050 40.527 7.453 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.9 123.3 127.0 EPYC 8324PN - Zen 4C 13.0 78.2 88.5 EPYC 8324P - Zen 4C 12.7 98.5 101.3 EPYC 8324P - Zen 4C, 155W 13.5 99.5 102.3 EPYC 7601 - Zen 1 65.8 139.9 146.3 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 40 80 120 160 200
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.
Result
OpenBenchmarking.org Items Per Second, More Is Better OSPRay 2.12 Benchmark: gravity_spheres_volume/dim_512/pathtracer/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4 8 12 16 20 SE +/- 0.00619, N = 3 SE +/- 0.00484, N = 3 SE +/- 0.00075, N = 3 SE +/- 0.00699, N = 3 SE +/- 0.00520, N = 3 16.89540 8.89361 10.49390 9.87373 4.14655 Items Per Second Per Watt
OpenBenchmarking.org Items Per Second Per Watt, More Is Better OSPRay 2.12 Benchmark: gravity_spheres_volume/dim_512/pathtracer/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.034 0.068 0.102 0.136 0.17 0.151 0.119 0.107 0.113 0.026 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.2 111.9 133.3 EPYC 8324PN - Zen 4C 12.6 74.9 86.9 EPYC 8324P - Zen 4C 13.0 98.1 122.1 EPYC 8324P - Zen 4C, 155W 13.7 87.1 106.8 EPYC 7601 - Zen 1 64.9 158.4 177.4 OpenBenchmarking.org Watts, Fewer Is Better OSPRay 2.12 CPU Power Consumption Monitor 50 100 150 200 250
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.
easyWave The easyWave software allows simulating tsunami generation and propagation in the context of early warning systems. EasyWave supports making use of OpenMP for CPU multi-threading and there are also GPU ports available but not currently incorporated as part of this test profile. The easyWave tsunami generation software is run with one of the example/reference input files for measuring the CPU execution time. Learn more via the OpenBenchmarking.org test page.
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.
Result
OpenBenchmarking.org H/s, More Is Better Xmrig 6.21 Variant: Wownero - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 8K 16K 24K 32K 40K SE +/- 47.09, N = 3 SE +/- 107.46, N = 3 SE +/- 52.06, N = 3 SE +/- 6.52, N = 3 SE +/- 120.26, N = 3 35370.0 21769.2 25644.3 24096.2 9331.6 1. (CXX) g++ options: -fexceptions -fno-rtti -maes -O3 -Ofast -static-libgcc -static-libstdc++ -rdynamic -lssl -lcrypto -luv -lpthread -lrt -ldl -lhwloc H/s Per Watt
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: Wownero - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 60 120 180 240 300 265.91 247.53 219.80 233.10 57.58 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.3 133.0 154.3 EPYC 8324PN - Zen 4C 13.1 87.9 95.9 EPYC 8324P - Zen 4C 13.3 116.4 133.1 EPYC 8324P - Zen 4C, 155W 12.8 103.4 113.7 EPYC 7601 - Zen 1 65.7 162.1 172.1 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
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.
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.
Intel Open Image Denoise Result
OpenBenchmarking.org Images / Sec, More Is Better Intel Open Image Denoise 2.1 Run: RT.ldr_alb_nrm.3840x2160 - Device: CPU-Only EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.3848 0.7696 1.1544 1.5392 1.924 SE +/- 0.00, N = 3 SE +/- 0.00, N = 3 SE +/- 0.00, N = 3 SE +/- 0.00, N = 3 SE +/- 0.00, N = 3 1.71 0.93 1.11 1.03 0.48 Images / Sec Per Watt
OpenBenchmarking.org Images / Sec Per Watt, More Is Better Intel Open Image Denoise 2.1 Run: RT.ldr_alb_nrm.3840x2160 - Device: CPU-Only EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0038 0.0076 0.0114 0.0152 0.019 0.017 0.013 0.012 0.012 0.003 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.9 101.2 124.6 EPYC 8324PN - Zen 4C 11.7 72.9 83.5 EPYC 8324P - Zen 4C 13.2 96.0 121.1 EPYC 8324P - Zen 4C, 155W 13.7 85.6 101.3 EPYC 7601 - Zen 1 65.5 165.0 174.6 OpenBenchmarking.org Watts, Fewer Is Better Intel Open Image Denoise 2.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better Embree 4.3 Binary: Pathtracer ISPC - Model: Asian Dragon EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 20 40 60 80 100 SE +/- 0.10, N = 5 SE +/- 0.03, N = 3 SE +/- 0.01, N = 3 SE +/- 0.03, N = 3 SE +/- 0.06, N = 3 74.93 39.60 45.51 43.13 21.84 MIN: 73.96 / MAX: 76.92 MIN: 39.34 / MAX: 40.89 MIN: 45.23 / MAX: 46.5 MIN: 42.84 / MAX: 43.66 MIN: 21.59 / MAX: 22.16 Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better Embree 4.3 Binary: Pathtracer ISPC - Model: Asian Dragon EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1694 0.3388 0.5082 0.6776 0.847 0.753 0.522 0.434 0.498 0.135 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.4 99.5 140.7 EPYC 8324PN - Zen 4C 12.8 75.9 95.2 EPYC 8324P - Zen 4C 12.7 105.0 134.4 EPYC 8324P - Zen 4C, 155W 14.0 86.6 107.3 EPYC 7601 - Zen 1 65.7 162.4 177.6 OpenBenchmarking.org Watts, Fewer Is Better Embree 4.3 CPU Power Consumption Monitor 50 100 150 200 250
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 60 120 180 240 300 SE +/- 0.08, N = 3 SE +/- 0.12, N = 3 SE +/- 0.11, N = 3 SE +/- 0.11, N = 3 SE +/- 0.71, N = 3 281.49 152.21 181.23 169.56 84.43
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: BERT-Large, NLP Question Answering - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 9 18 27 36 45 SE +/- 0.04, N = 3 SE +/- 0.02, N = 3 SE +/- 0.04, N = 3 SE +/- 0.02, N = 3 SE +/- 0.13, N = 3 40.64 21.91 25.84 24.47 12.22
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 14 28 42 56 70 SE +/- 0.01, N = 3 SE +/- 0.01, N = 3 SE +/- 0.03, N = 3 SE +/- 0.02, N = 3 SE +/- 0.70, N = 3 24.82 23.69 19.53 21.24 64.80 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.2 79.6 118.6 EPYC 8324PN - Zen 4C 12.6 64.6 82.1 EPYC 8324P - Zen 4C 12.1 81.3 112.8 EPYC 8324P - Zen 4C, 155W 13.4 73.7 97.1 EPYC 7601 - Zen 1 64.8 139.0 177.3 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
easyWave The easyWave software allows simulating tsunami generation and propagation in the context of early warning systems. EasyWave supports making use of OpenMP for CPU multi-threading and there are also GPU ports available but not currently incorporated as part of this test profile. The easyWave tsunami generation software is run with one of the example/reference input files for measuring the CPU execution time. Learn more via the OpenBenchmarking.org test page.
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better Embree 4.3 Binary: Pathtracer ISPC - Model: Crown EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 13 26 39 52 65 SE +/- 0.11, N = 3 SE +/- 0.07, N = 3 SE +/- 0.12, N = 3 SE +/- 0.08, N = 3 SE +/- 0.04, N = 3 59.05 31.08 36.89 34.24 18.44 MIN: 57.98 / MAX: 61.17 MIN: 30.63 / MAX: 31.81 MIN: 36.22 / MAX: 37.68 MIN: 33.72 / MAX: 35.12 MIN: 18.15 / MAX: 18.78 Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better Embree 4.3 Binary: Pathtracer ISPC - Model: Crown EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1827 0.3654 0.5481 0.7308 0.9135 0.812 0.398 0.361 0.373 0.112 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 72.7 142.1 EPYC 8324PN - Zen 4C 12.9 78.1 96.0 EPYC 8324P - Zen 4C 12.5 102.3 133.5 EPYC 8324P - Zen 4C, 155W 13.7 91.8 110.4 EPYC 7601 - Zen 1 65.9 165.2 178.1 OpenBenchmarking.org Watts, Fewer Is Better Embree 4.3 CPU Power Consumption Monitor 50 100 150 200 250
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 90 180 270 360 450 SE +/- 0.04, N = 3 SE +/- 0.03, N = 3 SE +/- 0.10, N = 3 SE +/- 0.08, N = 3 SE +/- 0.41, N = 3 429.33 229.54 266.96 254.30 138.03
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 7 14 21 28 35 SE +/- 0.03, N = 3 SE +/- 0.03, N = 3 SE +/- 0.04, N = 3 SE +/- 0.02, N = 3 SE +/- 0.02, N = 3 32.15 17.67 21.00 19.75 10.34
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: ResNet-50, Baseline - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 90 180 270 360 450 SE +/- 0.37, N = 3 SE +/- 0.16, N = 3 SE +/- 0.34, N = 3 SE +/- 0.06, N = 3 SE +/- 0.48, N = 3 429.36 229.91 266.99 254.13 138.53
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 7 14 21 28 35 SE +/- 0.03, N = 3 SE +/- 0.01, N = 3 SE +/- 0.07, N = 3 SE +/- 0.05, N = 3 SE +/- 0.08, N = 3 32.16 17.76 21.02 19.64 10.45
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: ResNet-50, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 700 1400 2100 2800 3500 SE +/- 2.41, N = 3 SE +/- 2.26, N = 3 SE +/- 7.34, N = 3 SE +/- 2.41, N = 3 SE +/- 0.21, N = 3 3450.43 1906.39 2193.09 2099.32 152.96
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 30 60 90 120 150 SE +/- 0.03, N = 3 SE +/- 0.04, N = 3 SE +/- 0.18, N = 3 SE +/- 0.15, N = 3 SE +/- 1.37, N = 3 52.12 49.80 41.64 44.72 124.09 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 11.8 92.3 120.3 EPYC 8324PN - Zen 4C 12.7 64.9 80.5 EPYC 8324P - Zen 4C 12.8 81.7 112.7 EPYC 8324P - Zen 4C, 155W 13.0 74.1 96.2 EPYC 7601 - Zen 1 65.2 138.7 176.8 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Op/s, More Is Better RocksDB 8.0 Test: Random Read EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50M 100M 150M 200M 250M SE +/- 187915.38, N = 3 SE +/- 242981.96, N = 3 SE +/- 126210.06, N = 3 SE +/- 294601.47, N = 3 SE +/- 709803.02, N = 3 248969387 131206830 161678876 147721769 84616408 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread Op/s Per Watt
OpenBenchmarking.org Op/s Per Watt, More Is Better RocksDB 8.0 Test: Random Read EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 500K 1000K 1500K 2000K 2500K 2230797.09 1742913.89 1613507.12 1652219.58 499648.82 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.8 111.6 129.8 EPYC 8324PN - Zen 4C 12.7 75.3 80.3 EPYC 8324P - Zen 4C 12.6 100.2 110.8 EPYC 8324P - Zen 4C, 155W 13.6 89.4 96.6 EPYC 7601 - Zen 1 66.2 169.4 178.6 OpenBenchmarking.org Watts, Fewer Is Better RocksDB 8.0 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Op/s, More Is Better Speedb 2.7 Test: Random Read EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50M 100M 150M 200M 250M SE +/- 139678.30, N = 3 SE +/- 227689.47, N = 3 SE +/- 141623.12, N = 3 SE +/- 1751764.05, N = 3 SE +/- 986396.38, N = 3 248842616 132248192 163771762 147991555 85990445 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread Op/s Per Watt
OpenBenchmarking.org Op/s Per Watt, More Is Better Speedb 2.7 Test: Random Read EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 500K 1000K 1500K 2000K 2500K 2534080.33 1749308.71 1624547.01 1658329.33 505924.53 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.1 98.2 128.2 EPYC 8324PN - Zen 4C 12.8 75.6 86.1 EPYC 8324P - Zen 4C 13.1 100.8 111.0 EPYC 8324P - Zen 4C, 155W 13.7 89.2 97.6 EPYC 7601 - Zen 1 65.6 170.0 178.3 OpenBenchmarking.org Watts, Fewer Is Better Speedb 2.7 CPU Power Consumption Monitor 50 100 150 200 250
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.
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.
Result
OpenBenchmarking.org Items Per Second, More Is Better OSPRay 2.12 Benchmark: particle_volume/scivis/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4 8 12 16 20 SE +/- 0.01627, N = 3 SE +/- 0.00550, N = 3 SE +/- 0.00419, N = 3 SE +/- 0.01425, N = 3 SE +/- 0.00416, N = 3 14.85490 8.14358 8.77043 8.75896 5.22927 Items Per Second Per Watt
OpenBenchmarking.org Items Per Second Per Watt, More Is Better OSPRay 2.12 Benchmark: particle_volume/scivis/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.032 0.064 0.096 0.128 0.16 0.142 0.111 0.093 0.100 0.034 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.6 104.4 125.1 EPYC 8324PN - Zen 4C 12.6 73.4 85.4 EPYC 8324P - Zen 4C 12.8 94.0 110.8 EPYC 8324P - Zen 4C, 155W 13.8 87.3 109.1 EPYC 7601 - Zen 1 60.0 155.5 175.7 OpenBenchmarking.org Watts, Fewer Is Better OSPRay 2.12 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org verify/s, More Is Better OpenSSL 3.1 Algorithm: RSA4096 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 200K 400K 600K 800K 1000K SE +/- 136.41, N = 3 SE +/- 22.57, N = 3 SE +/- 61.89, N = 3 SE +/- 30.98, N = 3 SE +/- 714.27, N = 3 828697.4 430171.6 457516.3 457514.1 292445.3 1. (CC) gcc options: -pthread -m64 -O3 -ldl verify/s Per Watt
OpenBenchmarking.org verify/s Per Watt, More Is Better OpenSSL 3.1 Algorithm: RSA4096 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 1500 3000 4500 6000 7500 7110.60 5622.23 4477.07 5076.37 1719.96 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.6 116.5 129.4 EPYC 8324PN - Zen 4C 12.2 76.5 83.6 EPYC 8324P - Zen 4C 13.4 102.2 116.0 EPYC 8324P - Zen 4C, 155W 13.7 90.1 102.2 EPYC 7601 - Zen 1 65.7 170.0 178.2 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 3.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Items Per Second, More Is Better OSPRay 2.12 Benchmark: particle_volume/ao/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4 8 12 16 20 SE +/- 0.02133, N = 3 SE +/- 0.01797, N = 3 SE +/- 0.00420, N = 3 SE +/- 0.02372, N = 3 SE +/- 0.00450, N = 3 14.90860 8.13231 8.79773 8.75784 5.28576 Items Per Second Per Watt
OpenBenchmarking.org Items Per Second Per Watt, More Is Better OSPRay 2.12 Benchmark: particle_volume/ao/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0286 0.0572 0.0858 0.1144 0.143 0.127 0.105 0.083 0.091 0.031 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 117.1 126.2 EPYC 8324PN - Zen 4C 12.5 77.2 86.1 EPYC 8324P - Zen 4C 13.2 106.5 111.2 EPYC 8324P - Zen 4C, 155W 13.7 95.9 110.1 EPYC 7601 - Zen 1 65.2 170.3 175.8 OpenBenchmarking.org Watts, Fewer Is Better OSPRay 2.12 CPU Power Consumption Monitor 50 100 150 200 250
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: CV Detection, YOLOv5s COCO - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 SE +/- 0.20, N = 3 SE +/- 0.03, N = 3 SE +/- 0.11, N = 3 SE +/- 0.10, N = 3 SE +/- 0.28, N = 3 191.07 103.44 122.37 114.62 68.16
OpenBenchmarking.org items/sec, More Is Better Neural Magic DeepSparse 1.6 Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 SE +/- 0.11, N = 3 SE +/- 0.03, N = 3 SE +/- 0.18, N = 3 SE +/- 0.07, N = 3 SE +/- 0.44, N = 3 192.06 104.13 122.93 115.42 68.95
Chaos Group V-RAY This is a test of Chaos Group's V-RAY benchmark. V-RAY is a commercial renderer that can integrate with various creator software products like SketchUp and 3ds Max. The V-RAY benchmark is standalone and supports CPU and NVIDIA CUDA/RTX based rendering. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org vsamples, More Is Better Chaos Group V-RAY 5.02 Mode: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 12K 24K 36K 48K 60K SE +/- 123.39, N = 3 SE +/- 326.21, N = 3 SE +/- 277.30, N = 3 SE +/- 150.82, N = 3 SE +/- 105.66, N = 3 54204 28177 33510 31536 19914 vsamples Per Watt
OpenBenchmarking.org vsamples Per Watt, More Is Better Chaos Group V-RAY 5.02 Mode: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 110 220 330 440 550 487.44 377.52 351.76 361.93 136.76 Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 111.2 137.4 EPYC 8324PN - Zen 4C 12.2 74.6 91.5 EPYC 8324P - Zen 4C 12.9 95.3 122.0 EPYC 8324P - Zen 4C, 155W 13.5 87.1 106.2 EPYC 7601 - Zen 1 64.1 145.6 179.2 OpenBenchmarking.org Watts, Fewer Is Better Chaos Group V-RAY 5.02 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org byte/s, More Is Better OpenSSL 3.1 Algorithm: SHA512 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 5000M 10000M 15000M 20000M 25000M SE +/- 1693446.92, N = 3 SE +/- 3220312.96, N = 3 SE +/- 1511880.16, N = 3 SE +/- 794607.82, N = 3 SE +/- 7664614.59, N = 3 22697266207 11644867590 13803340747 12993940737 8355425337 1. (CC) gcc options: -pthread -m64 -O3 -ldl byte/s Per Watt
OpenBenchmarking.org byte/s Per Watt, More Is Better OpenSSL 3.1 Algorithm: SHA512 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40M 80M 120M 160M 200M 175273393.29 142108901.03 122734575.99 132126423.40 47835855.20 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.3 129.5 138.1 EPYC 8324PN - Zen 4C 13.0 81.9 88.0 EPYC 8324P - Zen 4C 12.8 112.5 125.9 EPYC 8324P - Zen 4C, 155W 13.6 98.3 106.9 EPYC 7601 - Zen 1 65.5 174.7 178.2 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 3.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better SVT-AV1 1.8 Encoder Mode: Preset 12 - Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 SE +/- 2.10, N = 5 SE +/- 1.17, N = 5 SE +/- 1.24, N = 13 SE +/- 1.73, N = 6 SE +/- 0.47, N = 3 190.89 167.16 182.26 175.75 70.77 1. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better SVT-AV1 1.8 Encoder Mode: Preset 12 - Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.8829 1.7658 2.6487 3.5316 4.4145 3.924 3.739 3.665 3.666 0.726 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.7 48.6 109.6 EPYC 8324PN - Zen 4C 12.5 44.5 82.7 EPYC 8324P - Zen 4C 12.0 49.7 112.3 EPYC 8324P - Zen 4C, 155W 13.2 47.9 100.4 EPYC 7601 - Zen 1 65.4 97.5 157.6 OpenBenchmarking.org Watts, Fewer Is Better SVT-AV1 1.8 CPU Power Consumption Monitor 40 80 120 160 200
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.org MIPS, More Is Better 7-Zip Compression 22.01 Test: Compression Rating EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 70K 140K 210K 280K 350K SE +/- 303.12, N = 3 SE +/- 39.63, N = 3 SE +/- 329.87, N = 3 SE +/- 172.23, N = 3 SE +/- 1639.95, N = 3 332922 212458 240083 228957 123746 1. (CXX) g++ options: -lpthread -ldl -O2 -fPIC
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.
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.
Result
OpenBenchmarking.org byte/s, More Is Better OpenSSL 3.1 Algorithm: SHA256 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 15000M 30000M 45000M 60000M 75000M SE +/- 41929433.75, N = 3 SE +/- 16148314.98, N = 3 SE +/- 11419799.65, N = 3 SE +/- 12904999.14, N = 3 SE +/- 58715423.07, N = 3 69910430987 35795826073 43033082363 40200353400 27015393267 1. (CC) gcc options: -pthread -m64 -O3 -ldl byte/s Per Watt
OpenBenchmarking.org byte/s Per Watt, More Is Better OpenSSL 3.1 Algorithm: SHA256 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 120M 240M 360M 480M 600M 546243197.78 436238930.85 387067179.89 416875314.97 155415474.32 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.1 128.0 136.2 EPYC 8324PN - Zen 4C 12.8 82.1 90.4 EPYC 8324P - Zen 4C 13.0 111.2 118.1 EPYC 8324P - Zen 4C, 155W 13.5 96.4 102.1 EPYC 7601 - Zen 1 65.3 173.8 178.2 OpenBenchmarking.org Watts, Fewer Is Better OpenSSL 3.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
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.
Result
OpenBenchmarking.org MIPS, More Is Better 7-Zip Compression 22.01 Test: Decompression Rating EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 70K 140K 210K 280K 350K SE +/- 107.49, N = 3 SE +/- 170.94, N = 3 SE +/- 111.02, N = 3 SE +/- 181.34, N = 3 SE +/- 602.42, N = 3 343946 179072 211187 200034 133490 1. (CXX) g++ options: -lpthread -ldl -O2 -fPIC MIPS Per Watt
OpenBenchmarking.org MIPS Per Watt, More Is Better 7-Zip Compression 22.01 Test: Decompression Rating EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 700 1400 2100 2800 3500 3105.30 2439.42 2150.10 2334.06 846.16 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.2 110.8 145.7 EPYC 8324PN - Zen 4C 12.3 73.4 90.8 EPYC 8324P - Zen 4C 12.8 98.2 130.5 EPYC 8324P - Zen 4C, 155W 13.6 85.7 113.3 EPYC 7601 - Zen 1 65.9 157.8 178.5 OpenBenchmarking.org Watts, Fewer Is Better 7-Zip Compression 22.01 CPU Power Consumption Monitor 50 100 150 200 250
nginx This is a benchmark of the lightweight Nginx HTTP(S) web-server. This Nginx web server benchmark test profile makes use of the wrk program for facilitating the HTTP requests over a fixed period time with a configurable number of concurrent clients/connections. HTTPS with a self-signed OpenSSL certificate is used by this test for local benchmarking. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Requests Per Second, More Is Better nginx 1.23.2 Connections: 1000 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50K 100K 150K 200K 250K SE +/- 302.03, N = 3 SE +/- 199.18, N = 3 SE +/- 343.16, N = 3 SE +/- 49.01, N = 3 SE +/- 357.56, N = 3 232146.41 139259.26 166799.14 151923.62 90952.15 1. (CC) gcc options: -lluajit-5.1 -lm -lssl -lcrypto -lpthread -ldl -std=c99 -O2 Requests Per Second Per Watt
OpenBenchmarking.org Requests Per Second Per Watt, More Is Better nginx 1.23.2 Connections: 1000 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 400 800 1200 1600 2000 1785.71 1663.27 1365.65 1532.61 540.03 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.4 130.0 141.7 EPYC 8324PN - Zen 4C 12.8 83.7 92.6 EPYC 8324P - Zen 4C 12.3 122.1 134.6 EPYC 8324P - Zen 4C, 155W 13.5 99.1 113.1 EPYC 7601 - Zen 1 63.9 168.4 177.8 OpenBenchmarking.org Watts, Fewer Is Better nginx 1.23.2 CPU Power Consumption Monitor 50 100 150 200 250
IndigoBench This is a test of Indigo Renderer's IndigoBench benchmark. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org M samples/s, More Is Better IndigoBench 4.4 Acceleration: CPU - Scene: Supercar EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 5 10 15 20 25 SE +/- 0.020, N = 3 SE +/- 0.010, N = 3 SE +/- 0.058, N = 3 SE +/- 0.005, N = 3 SE +/- 0.024, N = 3 21.767 11.249 13.573 12.572 8.586 M samples/s Per Watt
OpenBenchmarking.org M samples/s Per Watt, More Is Better IndigoBench 4.4 Acceleration: CPU - Scene: Supercar EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0437 0.0874 0.1311 0.1748 0.2185 0.194 0.137 0.122 0.130 0.051 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.4 112.4 140.3 EPYC 8324PN - Zen 4C 12.4 82.0 88.8 EPYC 8324P - Zen 4C 12.6 111.0 122.4 EPYC 8324P - Zen 4C, 155W 13.7 96.6 108.8 EPYC 7601 - Zen 1 64.9 166.9 178.8 OpenBenchmarking.org Watts, Fewer Is Better IndigoBench 4.4 CPU Power Consumption Monitor 50 100 150 200 250
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.
IndigoBench This is a test of Indigo Renderer's IndigoBench benchmark. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org M samples/s, More Is Better IndigoBench 4.4 Acceleration: CPU - Scene: Bedroom EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 3 6 9 12 15 SE +/- 0.011, N = 3 SE +/- 0.010, N = 3 SE +/- 0.021, N = 3 SE +/- 0.013, N = 3 SE +/- 0.010, N = 3 9.973 5.127 6.177 5.720 3.985 M samples/s Per Watt
OpenBenchmarking.org M samples/s Per Watt, More Is Better IndigoBench 4.4 Acceleration: CPU - Scene: Bedroom EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0221 0.0442 0.0663 0.0884 0.1105 0.098 0.061 0.055 0.058 0.024 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 7.0 101.7 142.9 EPYC 8324PN - Zen 4C 12.4 83.6 91.8 EPYC 8324P - Zen 4C 13.3 112.5 129.6 EPYC 8324P - Zen 4C, 155W 13.8 98.9 111.2 EPYC 7601 - Zen 1 65.4 165.8 179.2 OpenBenchmarking.org Watts, Fewer Is Better IndigoBench 4.4 CPU Power Consumption Monitor 50 100 150 200 250
nginx This is a benchmark of the lightweight Nginx HTTP(S) web-server. This Nginx web server benchmark test profile makes use of the wrk program for facilitating the HTTP requests over a fixed period time with a configurable number of concurrent clients/connections. HTTPS with a self-signed OpenSSL certificate is used by this test for local benchmarking. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Requests Per Second, More Is Better nginx 1.23.2 Connections: 500 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50K 100K 150K 200K 250K SE +/- 217.83, N = 3 SE +/- 213.01, N = 3 SE +/- 549.43, N = 3 SE +/- 1108.21, N = 3 SE +/- 911.83, N = 3 229161.91 136660.96 164234.98 149458.34 94936.99 1. (CC) gcc options: -lluajit-5.1 -lm -lssl -lcrypto -lpthread -ldl -std=c99 -O2 Requests Per Second Per Watt
OpenBenchmarking.org Requests Per Second Per Watt, More Is Better nginx 1.23.2 Connections: 500 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 400 800 1200 1600 2000 1802.11 1656.12 1355.86 1536.38 561.58 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.6 127.2 140.0 EPYC 8324PN - Zen 4C 11.7 82.5 91.8 EPYC 8324P - Zen 4C 11.6 121.1 132.4 EPYC 8324P - Zen 4C, 155W 12.9 97.3 110.2 EPYC 7601 - Zen 1 60.9 169.1 178.8 OpenBenchmarking.org Watts, Fewer Is Better nginx 1.23.2 CPU Power Consumption Monitor 50 100 150 200 250
Redis 7.0.12 + memtier_benchmark Memtier_benchmark is a NoSQL Redis/Memcache traffic generation plus benchmarking tool developed by Redis Labs. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Ops/sec, More Is Better Redis 7.0.12 + memtier_benchmark 2.0 Protocol: Redis - Clients: 100 - Set To Get Ratio: 1:5 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 600K 1200K 1800K 2400K 3000K SE +/- 10691.43, N = 3 SE +/- 18767.16, N = 3 SE +/- 10431.24, N = 3 SE +/- 15002.29, N = 3 SE +/- 11184.62, N = 3 2513019.32 2500035.09 2795965.05 2720321.30 1159654.33 1. (CXX) g++ options: -O2 -levent_openssl -levent -lcrypto -lssl -lpthread -lz -lpcre Ops/sec Per Watt
OpenBenchmarking.org Ops/sec Per Watt, More Is Better Redis 7.0.12 + memtier_benchmark 2.0 Protocol: Redis - Clients: 100 - Set To Get Ratio: 1:5 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 9K 18K 27K 36K 45K 27624.53 39908.91 32252.28 36173.12 7148.89 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.3 91.0 105.2 EPYC 8324PN - Zen 4C 12.3 62.6 75.6 EPYC 8324P - Zen 4C 12.0 86.7 99.3 EPYC 8324P - Zen 4C, 155W 13.4 75.2 93.2 EPYC 7601 - Zen 1 66.5 162.2 178.4 OpenBenchmarking.org Watts, Fewer Is Better Redis 7.0.12 + memtier_benchmark 2.0 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better SVT-AV1 1.8 Encoder Mode: Preset 8 - Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 14 28 42 56 70 SE +/- 0.39, N = 4 SE +/- 0.05, N = 4 SE +/- 0.13, N = 4 SE +/- 0.14, N = 4 SE +/- 0.28, N = 5 64.34 51.54 58.00 55.72 26.75 1. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better SVT-AV1 1.8 Encoder Mode: Preset 8 - Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.2171 0.4342 0.6513 0.8684 1.0855 0.965 0.882 0.826 0.855 0.215 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.2 66.7 108.8 EPYC 8324PN - Zen 4C 12.5 58.5 84.2 EPYC 8324P - Zen 4C 12.4 70.2 112.0 EPYC 8324P - Zen 4C, 155W 13.2 65.2 97.5 EPYC 7601 - Zen 1 63.7 124.3 172.5 OpenBenchmarking.org Watts, Fewer Is Better SVT-AV1 1.8 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org MFLOPS, More Is Better QuantLib 1.32 Configuration: Multi-Threaded EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 30K 60K 90K 120K 150K SE +/- 29.72, N = 3 SE +/- 122.23, N = 3 SE +/- 60.04, N = 3 SE +/- 100.68, N = 3 SE +/- 165.76, N = 3 154830.9 80175.3 98717.2 90654.1 64469.4 1. (CXX) g++ options: -O3 -march=native -fPIE -pie MFLOPS Per Watt
OpenBenchmarking.org MFLOPS Per Watt, More Is Better QuantLib 1.32 Configuration: Multi-Threaded EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 300 600 900 1200 1500 1273.59 1007.58 945.62 967.97 384.04 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.5 121.6 136.3 EPYC 8324PN - Zen 4C 12.5 79.6 95.3 EPYC 8324P - Zen 4C 12.8 104.4 122.9 EPYC 8324P - Zen 4C, 155W 13.3 93.7 108.8 EPYC 7601 - Zen 1 64.3 167.9 179.0 OpenBenchmarking.org Watts, Fewer Is Better QuantLib 1.32 CPU Power Consumption Monitor 50 100 150 200 250
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.
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.
Result
OpenBenchmarking.org days/ns, Fewer Is Better NAMD 2.14 ATPase Simulation - 327,506 Atoms EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.2189 0.4378 0.6567 0.8756 1.0945 SE +/- 0.00128, N = 3 SE +/- 0.00094, N = 3 SE +/- 0.00067, N = 3 SE +/- 0.00161, N = 3 SE +/- 0.00038, N = 3 0.43200 0.82790 0.67728 0.73558 0.97282 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.7 118.1 142.8 EPYC 8324PN - Zen 4C 12.1 80.9 90.2 EPYC 8324P - Zen 4C 13.5 106.4 125.6 EPYC 8324P - Zen 4C, 155W 13.9 94.8 107.7 EPYC 7601 - Zen 1 65.7 165.3 178.5 OpenBenchmarking.org Watts, Fewer Is Better NAMD 2.14 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org FPS, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Video On Demand EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 11 22 33 44 55 SE +/- 0.03, N = 3 SE +/- 0.05, N = 3 SE +/- 0.02, N = 3 SE +/- 0.04, N = 3 SE +/- 0.03, N = 3 47.08 44.99 45.07 45.06 21.00 1. (CXX) g++ options: -O3 -rdynamic -lpthread -lrt -ldl -lnuma FPS Per Watt
OpenBenchmarking.org FPS Per Watt, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Video On Demand EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.173 0.346 0.519 0.692 0.865 0.765 0.769 0.753 0.744 0.241 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.9 61.5 79.2 EPYC 8324PN - Zen 4C 12.7 58.5 74.5 EPYC 8324P - Zen 4C 12.6 59.8 78.4 EPYC 8324P - Zen 4C, 155W 13.2 60.5 78.7 EPYC 7601 - Zen 1 37.8 87.0 127.4 OpenBenchmarking.org Watts, Fewer Is Better FFmpeg 6.1 CPU Power Consumption Monitor 40 80 120 160 200
Result
OpenBenchmarking.org FPS, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Upload EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 6 12 18 24 30 SE +/- 0.03, N = 3 SE +/- 0.02, N = 3 SE +/- 0.03, N = 3 SE +/- 0.02, N = 3 SE +/- 0.01, N = 3 23.31 22.24 22.24 22.25 10.40 1. (CXX) g++ options: -O3 -rdynamic -lpthread -lrt -ldl -lnuma FPS Per Watt
OpenBenchmarking.org FPS Per Watt, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Upload EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0911 0.1822 0.2733 0.3644 0.4555 0.404 0.405 0.396 0.391 0.123 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.2 57.7 89.3 EPYC 8324PN - Zen 4C 12.7 55.0 76.7 EPYC 8324P - Zen 4C 12.7 56.2 86.2 EPYC 8324P - Zen 4C, 155W 13.3 57.0 89.0 EPYC 7601 - Zen 1 63.0 84.5 141.4 OpenBenchmarking.org Watts, Fewer Is Better FFmpeg 6.1 CPU Power Consumption Monitor 40 80 120 160 200
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.
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.
Result
OpenBenchmarking.org FPS, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Platform EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 11 22 33 44 55 SE +/- 0.04, N = 3 SE +/- 0.01, N = 3 SE +/- 0.04, N = 3 SE +/- 0.03, N = 3 SE +/- 0.01, N = 3 47.03 45.03 45.01 45.12 21.02 1. (CXX) g++ options: -O3 -rdynamic -lpthread -lrt -ldl -lnuma FPS Per Watt
OpenBenchmarking.org FPS Per Watt, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Platform EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.173 0.346 0.519 0.692 0.865 0.766 0.769 0.753 0.745 0.241 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 11.3 61.4 77.5 EPYC 8324PN - Zen 4C 11.9 58.6 75.3 EPYC 8324P - Zen 4C 12.5 59.8 75.2 EPYC 8324P - Zen 4C, 155W 13.3 60.5 78.2 EPYC 7601 - Zen 1 62.5 87.1 125.3 OpenBenchmarking.org Watts, Fewer Is Better FFmpeg 6.1 CPU Power Consumption Monitor 40 80 120 160 200
Redis 7.0.12 + memtier_benchmark Memtier_benchmark is a NoSQL Redis/Memcache traffic generation plus benchmarking tool developed by Redis Labs. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Ops/sec, More Is Better Redis 7.0.12 + memtier_benchmark 2.0 Protocol: Redis - Clients: 100 - Set To Get Ratio: 1:10 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 600K 1200K 1800K 2400K 3000K SE +/- 20726.56, N = 3 SE +/- 11338.77, N = 3 SE +/- 3585.70, N = 3 SE +/- 16343.62, N = 3 SE +/- 10214.62, N = 3 2568868.49 2562876.16 2842511.46 2789711.34 1277853.47 1. (CXX) g++ options: -O2 -levent_openssl -levent -lcrypto -lssl -lpthread -lz -lpcre Ops/sec Per Watt
OpenBenchmarking.org Ops/sec Per Watt, More Is Better Redis 7.0.12 + memtier_benchmark 2.0 Protocol: Redis - Clients: 100 - Set To Get Ratio: 1:10 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 9K 18K 27K 36K 45K 28227.25 41473.46 32673.79 37066.25 7946.04 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.5 91.0 106.9 EPYC 8324PN - Zen 4C 11.3 61.8 76.7 EPYC 8324P - Zen 4C 11.2 87.0 99.1 EPYC 8324P - Zen 4C, 155W 13.6 75.3 93.2 EPYC 7601 - Zen 1 64.4 160.8 178.4 OpenBenchmarking.org Watts, Fewer Is Better Redis 7.0.12 + memtier_benchmark 2.0 CPU Power Consumption Monitor 50 100 150 200 250
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.
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 SE +/- 0.01, N = 3 SE +/- 0.04, N = 3 SE +/- 0.03, N = 3 SE +/- 0.04, N = 3 SE +/- 1.56, N = 3 113.45 104.90 88.15 94.20 189.32 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.8 99.4 123.8 EPYC 8324PN - Zen 4C 12.6 67.5 83.1 EPYC 8324P - Zen 4C 12.9 87.7 109.6 EPYC 8324P - Zen 4C, 155W 12.7 78.5 91.9 EPYC 7601 - Zen 1 65.4 148.0 176.5 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better uvg266 0.4.1 Video Input: Bosphorus 4K - Video Preset: Very Fast EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 11 22 33 44 55 SE +/- 0.04, N = 4 SE +/- 0.03, N = 3 SE +/- 0.02, N = 4 SE +/- 0.02, N = 4 SE +/- 0.08, N = 3 47.65 33.82 39.25 37.33 22.26 Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better uvg266 0.4.1 Video Input: Bosphorus 4K - Video Preset: Very Fast EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1217 0.2434 0.3651 0.4868 0.6085 0.541 0.494 0.439 0.464 0.146 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 88.2 122.3 EPYC 8324PN - Zen 4C 12.8 68.5 88.7 EPYC 8324P - Zen 4C 13.0 89.3 117.0 EPYC 8324P - Zen 4C, 155W 13.7 80.5 105.1 EPYC 7601 - Zen 1 64.3 152.6 171.9 OpenBenchmarking.org Watts, Fewer Is Better uvg266 0.4.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better SVT-AV1 1.8 Encoder Mode: Preset 4 - Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 2 4 6 8 10 SE +/- 0.026, N = 3 SE +/- 0.003, N = 3 SE +/- 0.020, N = 3 SE +/- 0.004, N = 3 SE +/- 0.004, N = 3 6.556 5.551 5.911 5.816 3.065 1. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better SVT-AV1 1.8 Encoder Mode: Preset 4 - Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0205 0.041 0.0615 0.082 0.1025 0.090 0.091 0.080 0.082 0.024 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.6 72.9 105.4 EPYC 8324PN - Zen 4C 12.6 61.3 86.3 EPYC 8324P - Zen 4C 12.7 74.3 120.0 EPYC 8324P - Zen 4C, 155W 13.3 70.5 101.6 EPYC 7601 - Zen 1 63.4 125.1 171.7 OpenBenchmarking.org Watts, Fewer Is Better SVT-AV1 1.8 CPU Power Consumption Monitor 50 100 150 200 250
PyTorch Result
OpenBenchmarking.org batches/sec, More Is Better PyTorch 2.1 Device: CPU - Batch Size: 1 - Model: ResNet-50 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 12 24 36 48 60 SE +/- 0.10, N = 3 SE +/- 0.07, N = 3 SE +/- 0.20, N = 3 SE +/- 0.07, N = 3 SE +/- 0.09, N = 3 45.23 52.90 52.72 52.91 24.89 MIN: 16.58 / MAX: 46.26 MIN: 50.6 / MAX: 53.56 MIN: 50.17 / MAX: 53.72 MIN: 50.65 / MAX: 53.55 MIN: 12.04 / MAX: 25.89 batches/sec Per Watt
OpenBenchmarking.org batches/sec Per Watt, More Is Better PyTorch 2.1 Device: CPU - Batch Size: 1 - Model: ResNet-50 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1816 0.3632 0.5448 0.7264 0.908 0.552 0.807 0.784 0.781 0.162 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.4 82.0 98.6 EPYC 8324PN - Zen 4C 12.9 65.6 79.0 EPYC 8324P - Zen 4C 12.8 67.3 81.3 EPYC 8324P - Zen 4C, 155W 13.3 67.8 81.9 EPYC 7601 - Zen 1 63.8 153.6 174.3 OpenBenchmarking.org Watts, Fewer Is Better PyTorch 2.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Op/s, More Is Better Speedb 2.7 Test: Read While Writing EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 3M 6M 9M 12M 15M SE +/- 117697.08, N = 15 SE +/- 50044.04, N = 3 SE +/- 52967.61, N = 15 SE +/- 7924.49, N = 3 SE +/- 81525.26, N = 3 12585315 6934544 7290610 7162203 5954251 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread Op/s Per Watt
OpenBenchmarking.org Op/s Per Watt, More Is Better Speedb 2.7 Test: Read While Writing EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 20K 40K 60K 80K 100K 108503.02 88296.45 64987.07 74216.65 35300.60 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.6 116.0 134.1 EPYC 8324PN - Zen 4C 12.8 78.5 91.1 EPYC 8324P - Zen 4C 11.8 112.2 124.5 EPYC 8324P - Zen 4C, 155W 13.6 96.5 108.8 EPYC 7601 - Zen 1 65.3 168.7 178.1 OpenBenchmarking.org Watts, Fewer Is Better Speedb 2.7 CPU Power Consumption Monitor 50 100 150 200 250
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: BERT-Large, NLP Question Answering - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 300 600 900 1200 1500 SE +/- 0.25, N = 3 SE +/- 0.30, N = 3 SE +/- 0.30, N = 3 SE +/- 0.27, N = 3 SE +/- 8.56, N = 3 776.54 722.10 611.15 647.89 1289.82 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 80.1 127.4 EPYC 8324PN - Zen 4C 12.8 60.3 86.0 EPYC 8324P - Zen 4C 12.1 71.4 119.7 EPYC 8324P - Zen 4C, 155W 13.2 67.0 108.3 EPYC 7601 - Zen 1 64.8 109.6 177.6 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better uvg266 0.4.1 Video Input: Bosphorus 4K - Video Preset: Super Fast EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 11 22 33 44 55 SE +/- 0.06, N = 4 SE +/- 0.03, N = 4 SE +/- 0.03, N = 4 SE +/- 0.08, N = 4 SE +/- 0.02, N = 3 49.22 36.75 42.94 40.50 23.58 Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better uvg266 0.4.1 Video Input: Bosphorus 4K - Video Preset: Super Fast EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1287 0.2574 0.3861 0.5148 0.6435 0.572 0.535 0.485 0.506 0.155 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.0 86.0 119.7 EPYC 8324PN - Zen 4C 12.5 68.7 89.9 EPYC 8324P - Zen 4C 12.9 88.5 116.2 EPYC 8324P - Zen 4C, 155W 13.6 80.0 102.0 EPYC 7601 - Zen 1 65.3 152.6 171.2 OpenBenchmarking.org Watts, Fewer Is Better uvg266 0.4.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 300 600 900 1200 1500 SE +/- 0.09, N = 3 SE +/- 0.79, N = 3 SE +/- 1.23, N = 3 SE +/- 0.47, N = 3 SE +/- 0.89, N = 3 982.33 894.20 752.51 801.19 1515.68 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.5 101.9 133.4 EPYC 8324PN - Zen 4C 12.5 69.7 89.0 EPYC 8324P - Zen 4C 11.8 88.8 121.7 EPYC 8324P - Zen 4C, 155W 13.2 80.1 102.2 EPYC 7601 - Zen 1 65.5 140.7 177.7 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: ResNet-50, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 20 40 60 80 100 SE +/- 0.0062, N = 3 SE +/- 0.0084, N = 3 SE +/- 0.0229, N = 3 SE +/- 0.0078, N = 3 SE +/- 0.1244, N = 3 9.2520 8.3745 7.2816 7.6064 104.4436 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.7 96.8 114.2 EPYC 8324PN - Zen 4C 12.4 66.5 76.5 EPYC 8324P - Zen 4C 12.9 93.1 108.1 EPYC 8324P - Zen 4C, 155W 13.2 77.0 95.3 EPYC 7601 - Zen 1 64.5 159.1 177.2 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
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.org FPS, More Is Better OpenVINO 2023.2.dev Model: Face Detection FP16-INT8 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 11 22 33 44 55 SE +/- 0.04, N = 3 SE +/- 0.01, N = 3 SE +/- 0.03, N = 3 SE +/- 0.04, N = 3 SE +/- 0.00, N = 3 49.43 25.41 31.25 28.58 3.81 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
OpenBenchmarking.org FPS, More Is Better OpenVINO 2023.2.dev Model: Weld Porosity Detection FP16-INT8 - Device: CPU EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 1000 2000 3000 4000 5000 SE +/- 1.98, N = 3 SE +/- 0.51, N = 3 SE +/- 0.48, N = 3 SE +/- 0.72, N = 3 SE +/- 0.03, N = 3 4878.00 2520.77 3102.13 2851.01 376.97 1. (CXX) g++ options: -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -pie
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 300 600 900 1200 1500 SE +/- 0.18, N = 3 SE +/- 0.38, N = 3 SE +/- 1.08, N = 3 SE +/- 1.07, N = 3 SE +/- 7.41, N = 3 982.61 892.10 751.62 803.82 1501.61 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.3 86.8 132.4 EPYC 8324PN - Zen 4C 12.5 69.4 84.7 EPYC 8324P - Zen 4C 12.1 89.9 119.5 EPYC 8324P - Zen 4C, 155W 13.1 80.9 105.1 EPYC 7601 - Zen 1 64.7 140.6 177.4 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
PyTorch Result
OpenBenchmarking.org batches/sec, More Is Better PyTorch 2.1 Device: CPU - Batch Size: 256 - Model: ResNet-50 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 9 18 27 36 45 SE +/- 0.14, N = 3 SE +/- 0.10, N = 3 SE +/- 0.06, N = 3 SE +/- 0.09, N = 3 SE +/- 0.12, N = 3 36.23 40.36 40.59 40.54 20.48 MIN: 14.59 / MAX: 36.98 MIN: 18.81 / MAX: 40.93 MIN: 14.97 / MAX: 41.27 MIN: 14.7 / MAX: 41.23 MIN: 11.38 / MAX: 21.17 batches/sec Per Watt
OpenBenchmarking.org batches/sec Per Watt, More Is Better PyTorch 2.1 Device: CPU - Batch Size: 256 - Model: ResNet-50 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1211 0.2422 0.3633 0.4844 0.6055 0.389 0.538 0.527 0.519 0.127 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.8 93.2 106.3 EPYC 8324PN - Zen 4C 12.6 75.0 85.7 EPYC 8324P - Zen 4C 12.8 77.1 88.7 EPYC 8324P - Zen 4C, 155W 13.1 78.0 89.1 EPYC 7601 - Zen 1 64.7 161.6 173.2 OpenBenchmarking.org Watts, Fewer Is Better PyTorch 2.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org FPS, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Live EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 30 60 90 120 150 SE +/- 0.20, N = 3 SE +/- 0.17, N = 3 SE +/- 0.30, N = 3 SE +/- 0.31, N = 3 SE +/- 0.11, N = 3 115.11 109.77 109.84 109.71 58.09 1. (CXX) g++ options: -O3 -rdynamic -lpthread -lrt -ldl -lnuma FPS Per Watt
OpenBenchmarking.org FPS Per Watt, More Is Better FFmpeg 6.1 Encoder: libx265 - Scenario: Live EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.459 0.918 1.377 1.836 2.295 2.005 2.040 1.988 1.977 0.704 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.2 57.4 78.9 EPYC 8324PN - Zen 4C 12.3 53.8 71.4 EPYC 8324P - Zen 4C 12.0 55.2 74.9 EPYC 8324P - Zen 4C, 155W 13.5 55.5 74.6 EPYC 7601 - Zen 1 64.0 82.5 112.1 OpenBenchmarking.org Watts, Fewer Is Better FFmpeg 6.1 CPU Power Consumption Monitor 40 80 120 160 200
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.
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 30 60 90 120 150 SE +/- 0.02, N = 3 SE +/- 0.01, N = 3 SE +/- 0.02, N = 3 SE +/- 0.01, N = 3 SE +/- 0.35, N = 3 74.43 69.60 59.87 62.82 115.78 CPU Power Consumption
Min Avg Max EPYC 7601 - Zen 1 66.0 156.5 177.4 EPYC 8324P - Zen 4C 13.0 96.8 111.3 EPYC 8324P - Zen 4C, 155W 13.3 80.5 100.1 EPYC 8324PN - Zen 4C 11.6 67.8 79.3 EPYC 8534PN - Zen 4C 13.1 101.0 122.6 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Op/s, More Is Better RocksDB 8.0 Test: Read While Writing EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 1.4M 2.8M 4.2M 5.6M 7M SE +/- 64443.01, N = 15 SE +/- 3126.89, N = 3 SE +/- 38567.21, N = 15 SE +/- 47925.84, N = 5 SE +/- 36400.82, N = 3 6688089 4107313 4394310 4301343 3463695 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread Op/s Per Watt
OpenBenchmarking.org Op/s Per Watt, More Is Better RocksDB 8.0 Test: Read While Writing EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 12K 24K 36K 48K 60K 57031.64 51316.29 38762.72 43556.26 20446.35 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 117.3 138.2 EPYC 8324PN - Zen 4C 12.1 80.0 93.0 EPYC 8324P - Zen 4C 13.2 113.4 129.0 EPYC 8324P - Zen 4C, 155W 13.5 98.8 110.3 EPYC 7601 - Zen 1 65.8 169.4 177.9 OpenBenchmarking.org Watts, Fewer Is Better RocksDB 8.0 CPU Power Consumption Monitor 50 100 150 200 250
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: ResNet-50, Baseline - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 30 60 90 120 150 SE +/- 0.05, N = 3 SE +/- 0.05, N = 3 SE +/- 0.07, N = 3 SE +/- 0.02, N = 3 SE +/- 0.43, N = 3 74.41 69.51 59.87 62.88 115.35
x265 This is a simple test of the x265 encoder run on the CPU with 1080p and 4K options for H.265 video encode performance with x265. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Frames Per Second, More Is Better x265 3.4 Video Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 6 12 18 24 30 SE +/- 0.17, N = 3 SE +/- 0.05, N = 3 SE +/- 0.04, N = 3 SE +/- 0.04, N = 3 SE +/- 0.10, N = 3 27.51 24.03 25.55 25.45 14.29 1. (CXX) g++ options: -O3 -rdynamic -lpthread -lrt -ldl -lnuma Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better x265 3.4 Video Input: Bosphorus 4K EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0878 0.1756 0.2634 0.3512 0.439 0.362 0.390 0.343 0.342 0.112 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.6 76.1 96.9 EPYC 8324PN - Zen 4C 12.1 61.6 76.0 EPYC 8324P - Zen 4C 10.4 74.6 95.5 EPYC 8324P - Zen 4C, 155W 13.2 74.4 91.2 EPYC 7601 - Zen 1 64.3 128.0 150.6 OpenBenchmarking.org Watts, Fewer Is Better x265 3.4 CPU Power Consumption Monitor 40 80 120 160 200
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.
Result
OpenBenchmarking.org Frames Per Second, More Is Better uvg266 0.4.1 Video Input: Bosphorus 4K - Video Preset: Ultra Fast EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 12 24 36 48 60 SE +/- 0.14, N = 4 SE +/- 0.04, N = 4 SE +/- 0.05, N = 4 SE +/- 0.06, N = 4 SE +/- 0.03, N = 3 51.33 41.02 46.54 44.81 27.29 Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better uvg266 0.4.1 Video Input: Bosphorus 4K - Video Preset: Ultra Fast EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.1427 0.2854 0.4281 0.5708 0.7135 0.634 0.623 0.546 0.581 0.184 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.4 81.0 114.6 EPYC 8324PN - Zen 4C 12.6 65.8 87.1 EPYC 8324P - Zen 4C 12.8 85.3 111.9 EPYC 8324P - Zen 4C, 155W 13.5 77.1 102.8 EPYC 7601 - Zen 1 64.4 148.0 169.3 OpenBenchmarking.org Watts, Fewer Is Better uvg266 0.4.1 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org ns/day, More Is Better LAMMPS Molecular Dynamics Simulator 23Jun2022 Model: 20k Atoms EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 6 12 18 24 30 SE +/- 0.28, N = 3 SE +/- 0.02, N = 3 SE +/- 0.05, N = 3 SE +/- 0.03, N = 3 SE +/- 0.05, N = 3 25.72 15.69 18.41 17.42 13.78 1. (CXX) g++ options: -O3 -lm -ldl ns/day Per Watt
OpenBenchmarking.org ns/day Per Watt, More Is Better LAMMPS Molecular Dynamics Simulator 23Jun2022 Model: 20k Atoms EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0452 0.0904 0.1356 0.1808 0.226 0.201 0.192 0.167 0.181 0.080 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.9 128.0 132.4 EPYC 8324PN - Zen 4C 13.0 81.7 84.7 EPYC 8324P - Zen 4C 13.5 110.0 122.8 EPYC 8324P - Zen 4C, 155W 13.9 96.5 105.3 EPYC 7601 - Zen 1 65.6 172.4 175.7 OpenBenchmarking.org Watts, Fewer Is Better LAMMPS Molecular Dynamics Simulator 23Jun2022 CPU Power Consumption Monitor 50 100 150 200 250
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.
Neural Magic DeepSparse This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: CV Detection, YOLOv5s COCO - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50 100 150 200 250 SE +/- 0.14, N = 3 SE +/- 0.07, N = 3 SE +/- 0.08, N = 3 SE +/- 0.07, N = 3 SE +/- 0.88, N = 3 166.95 154.28 130.42 139.18 234.02 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.7 106.4 129.0 EPYC 8324PN - Zen 4C 12.3 70.8 81.0 EPYC 8324P - Zen 4C 13.0 91.9 117.7 EPYC 8324P - Zen 4C, 155W 13.0 82.6 98.1 EPYC 7601 - Zen 1 64.8 157.0 176.8 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
Result
OpenBenchmarking.org ms/batch, Fewer Is Better Neural Magic DeepSparse 1.6 Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Asynchronous Multi-Stream EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50 100 150 200 250 SE +/- 0.07, N = 3 SE +/- 0.09, N = 3 SE +/- 0.14, N = 3 SE +/- 0.04, N = 3 SE +/- 1.50, N = 3 166.03 153.40 129.84 138.23 231.40 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.8 105.5 121.8 EPYC 8324PN - Zen 4C 12.6 69.8 83.0 EPYC 8324P - Zen 4C 11.4 91.0 116.6 EPYC 8324P - Zen 4C, 155W 13.4 82.5 101.3 EPYC 7601 - Zen 1 64.4 156.5 176.5 OpenBenchmarking.org Watts, Fewer Is Better Neural Magic DeepSparse 1.6 CPU Power Consumption Monitor 50 100 150 200 250
VVenC VVenC is the Fraunhofer Versatile Video Encoder as a fast/efficient H.266/VVC encoder. The vvenc encoder makes use of SIMD Everywhere (SIMDe). The vvenc software is published under the Clear BSD License. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Frames Per Second, More Is Better VVenC 1.9 Video Input: Bosphorus 4K - Video Preset: Faster EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 3 6 9 12 15 SE +/- 0.012, N = 3 SE +/- 0.030, N = 3 SE +/- 0.010, N = 3 SE +/- 0.030, N = 3 SE +/- 0.044, N = 3 11.272 10.291 10.741 10.635 6.358 1. (CXX) g++ options: -O3 -flto=auto -fno-fat-lto-objects Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better VVenC 1.9 Video Input: Bosphorus 4K - Video Preset: Faster EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0356 0.0712 0.1068 0.1424 0.178 0.158 0.147 0.134 0.135 0.049 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.0 71.1 115.0 EPYC 8324PN - Zen 4C 12.9 69.8 89.3 EPYC 8324P - Zen 4C 11.8 80.4 107.5 EPYC 8324P - Zen 4C, 155W 13.1 78.7 103.8 EPYC 7601 - Zen 1 64.8 129.1 162.9 OpenBenchmarking.org Watts, Fewer Is Better VVenC 1.9 CPU Power Consumption Monitor 50 100 150 200 250
rav1e Xiph rav1e is a Rust-written AV1 video encoder that claims to be the fastest and safest AV1 encoder. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Frames Per Second, More Is Better rav1e 0.7 Speed: 10 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 3 6 9 12 15 SE +/- 0.125, N = 5 SE +/- 0.146, N = 4 SE +/- 0.141, N = 4 SE +/- 0.138, N = 4 SE +/- 0.075, N = 4 12.186 11.824 11.790 11.818 6.995 Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better rav1e 0.7 Speed: 10 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0509 0.1018 0.1527 0.2036 0.2545 0.223 0.226 0.222 0.221 0.079 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.9 54.6 59.0 EPYC 8324PN - Zen 4C 12.2 52.3 56.9 EPYC 8324P - Zen 4C 12.6 53.1 58.3 EPYC 8324P - Zen 4C, 155W 13.3 53.6 58.5 EPYC 7601 - Zen 1 45.5 88.5 96.7 OpenBenchmarking.org Watts, Fewer Is Better rav1e 0.7 CPU Power Consumption Monitor 20 40 60 80 100
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.
Result
OpenBenchmarking.org Op/s, More Is Better Speedb 2.7 Test: Update Random EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 70K 140K 210K 280K 350K SE +/- 2986.90, N = 15 SE +/- 863.96, N = 3 SE +/- 579.72, N = 3 SE +/- 167.58, N = 3 SE +/- 190.36, N = 3 349745 305729 317250 317634 203823 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread Op/s Per Watt
OpenBenchmarking.org Op/s Per Watt, More Is Better Speedb 2.7 Test: Update Random EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 900 1800 2700 3600 4500 3874.59 4423.42 4050.08 4047.08 1640.50 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.1 90.3 116.2 EPYC 8324PN - Zen 4C 12.9 69.1 79.1 EPYC 8324P - Zen 4C 12.2 78.3 88.2 EPYC 8324P - Zen 4C, 155W 13.6 78.5 88.0 EPYC 7601 - Zen 1 65.1 124.2 142.7 OpenBenchmarking.org Watts, Fewer Is Better Speedb 2.7 CPU Power Consumption Monitor 40 80 120 160 200
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.
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.
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.
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.
Result
OpenBenchmarking.org Items Per Second, More Is Better OSPRay 2.12 Benchmark: particle_volume/pathtracer/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 SE +/- 0.09, N = 3 SE +/- 0.02, N = 3 SE +/- 0.07, N = 3 SE +/- 0.09, N = 3 SE +/- 0.06, N = 3 164.21 147.22 147.25 147.28 98.90 Items Per Second Per Watt
OpenBenchmarking.org Items Per Second Per Watt, More Is Better OSPRay 2.12 Benchmark: particle_volume/pathtracer/real_time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.441 0.882 1.323 1.764 2.205 1.460 1.960 1.501 1.721 0.651 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.8 112.4 131.6 EPYC 8324PN - Zen 4C 12.8 75.1 88.1 EPYC 8324P - Zen 4C 12.6 98.1 119.5 EPYC 8324P - Zen 4C, 155W 13.6 85.6 106.9 EPYC 7601 - Zen 1 65.4 151.9 175.1 OpenBenchmarking.org Watts, Fewer Is Better OSPRay 2.12 CPU Power Consumption Monitor 50 100 150 200 250
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.
rav1e Xiph rav1e is a Rust-written AV1 video encoder that claims to be the fastest and safest AV1 encoder. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Frames Per Second, More Is Better rav1e 0.7 Speed: 6 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 1.0953 2.1906 3.2859 4.3812 5.4765 SE +/- 0.006, N = 3 SE +/- 0.017, N = 3 SE +/- 0.003, N = 3 SE +/- 0.014, N = 3 SE +/- 0.020, N = 3 4.868 4.727 4.706 4.731 2.963 Frames Per Second Per Watt
OpenBenchmarking.org Frames Per Second Per Watt, More Is Better rav1e 0.7 Speed: 6 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0203 0.0406 0.0609 0.0812 0.1015 0.090 0.090 0.089 0.088 0.032 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.2 54.2 61.9 EPYC 8324PN - Zen 4C 12.7 52.4 59.7 EPYC 8324P - Zen 4C 12.5 53.1 60.2 EPYC 8324P - Zen 4C, 155W 13.3 54.0 61.3 EPYC 7601 - Zen 1 65.7 93.2 104.7 OpenBenchmarking.org Watts, Fewer Is Better rav1e 0.7 CPU Power Consumption Monitor 20 40 60 80 100
Quicksilver Quicksilver is a proxy application that represents some elements of the Mercury workload by solving a simplified dynamic Monte Carlo particle transport problem. Quicksilver is developed by Lawrence Livermore National Laboratory (LLNL) and this test profile currently makes use of the OpenMP CPU threaded code path. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Figure Of Merit, More Is Better Quicksilver 20230818 Input: CORAL2 P1 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 5M 10M 15M 20M 25M SE +/- 18559.21, N = 3 SE +/- 24037.01, N = 3 SE +/- 66583.28, N = 3 SE +/- 106926.77, N = 3 SE +/- 105830.05, N = 3 21233333 18273333 18880000 18950000 13030000 1. (CXX) g++ options: -fopenmp -O3 -march=native Figure Of Merit Per Watt
OpenBenchmarking.org Figure Of Merit Per Watt, More Is Better Quicksilver 20230818 Input: CORAL2 P1 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50K 100K 150K 200K 250K 195063.32 243826.72 208439.28 208244.59 87231.32 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.3 108.9 118.9 EPYC 8324PN - Zen 4C 12.5 74.9 88.5 EPYC 8324P - Zen 4C 13.2 90.6 97.7 EPYC 8324P - Zen 4C, 155W 14.0 91.0 98.6 EPYC 7601 - Zen 1 65.4 149.4 162.5 OpenBenchmarking.org Watts, Fewer Is Better Quicksilver 20230818 CPU Power Consumption Monitor 50 100 150 200 250
Apache Cassandra This is a benchmark of the Apache Cassandra NoSQL database management system making use of cassandra-stress. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Op/s, More Is Better Apache Cassandra 4.1.3 Test: Writes EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50K 100K 150K 200K 250K SE +/- 1997.81, N = 3 SE +/- 423.47, N = 3 SE +/- 371.97, N = 3 SE +/- 522.48, N = 3 SE +/- 1316.75, N = 8 244177 212207 223293 224034 150277 Op/s Per Watt
OpenBenchmarking.org Op/s Per Watt, More Is Better Apache Cassandra 4.1.3 Test: Writes EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 600 1200 1800 2400 3000 2486.66 2829.25 2451.59 2447.90 977.63 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 12.5 98.2 123.2 EPYC 8324PN - Zen 4C 11.8 75.0 87.7 EPYC 8324P - Zen 4C 12.7 91.1 108.3 EPYC 8324P - Zen 4C, 155W 13.4 91.5 104.6 EPYC 7601 - Zen 1 63.3 153.7 181.9 OpenBenchmarking.org Watts, Fewer Is Better Apache Cassandra 4.1.3 CPU Power Consumption Monitor 50 100 150 200 250
Apache IoTDB Apache IotDB is a time series database and this benchmark is facilitated using the IoT Benchmaark [https://github.com/thulab/iot-benchmark/]. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org point/sec, More Is Better Apache IoTDB 1.2 Device Count: 500 - Batch Size Per Write: 100 - Sensor Count: 500 - Client Number: 100 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 20M 40M 60M 80M 100M SE +/- 820005.27, N = 3 SE +/- 90002.32, N = 3 SE +/- 458351.52, N = 3 SE +/- 1107129.95, N = 12 SE +/- 395656.07, N = 3 82731495 77285152 80321359 70294901 51950735
Kripke Kripke is a simple, scalable, 3D Sn deterministic particle transport code. Its primary purpose is to research how data layout, programming paradigms and architectures effect the implementation and performance of Sn transport. Kripke is developed by LLNL. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Throughput FoM, More Is Better Kripke 1.2.6 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 60M 120M 180M 240M 300M SE +/- 1948629.96, N = 3 SE +/- 363531.75, N = 3 SE +/- 634968.78, N = 3 SE +/- 2369703.74, N = 3 SE +/- 1494376.42, N = 3 288394300 247500133 268773267 258716167 181795967 1. (CXX) g++ options: -O3 -fopenmp -ldl Throughput FoM Per Watt
OpenBenchmarking.org Throughput FoM Per Watt, More Is Better Kripke 1.2.6 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 600K 1200K 1800K 2400K 3000K 2490692.28 2943707.26 2566916.47 2674408.73 1133698.87 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.7 115.8 137.0 EPYC 8324PN - Zen 4C 13.0 84.1 103.7 EPYC 8324P - Zen 4C 12.6 104.7 123.6 EPYC 8324P - Zen 4C, 155W 14.2 96.7 118.5 EPYC 7601 - Zen 1 66.0 160.4 176.7 OpenBenchmarking.org Watts, Fewer Is Better Kripke 1.2.6 CPU Power Consumption Monitor 50 100 150 200 250
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.
Result
OpenBenchmarking.org Op/s, More Is Better Speedb 2.7 Test: Read Random Write Random EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 500K 1000K 1500K 2000K 2500K SE +/- 2240.30, N = 3 SE +/- 1833.01, N = 3 SE +/- 3849.48, N = 3 SE +/- 1300.80, N = 3 SE +/- 4528.27, N = 3 2272026 1982257 2233744 2133311 1447819 1. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread Op/s Per Watt
OpenBenchmarking.org Op/s Per Watt, More Is Better Speedb 2.7 Test: Read Random Write Random EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 6K 12K 18K 24K 30K 20292.36 25808.68 20899.63 22326.89 9121.91 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.1 112.0 127.0 EPYC 8324PN - Zen 4C 12.2 76.8 90.2 EPYC 8324P - Zen 4C 13.1 106.9 119.8 EPYC 8324P - Zen 4C, 155W 13.7 95.5 110.7 EPYC 7601 - Zen 1 65.3 158.7 177.5 OpenBenchmarking.org Watts, Fewer Is Better Speedb 2.7 CPU Power Consumption Monitor 50 100 150 200 250
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.org Seconds, Fewer Is Better OpenFOAM 10 Input: drivaerFastback, Medium Mesh Size - Mesh Time EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 50 100 150 200 250 184.16 172.69 153.15 163.04 229.49 1. (CXX) g++ options: -std=c++14 -m64 -O3 -ftemplate-depth-100 -fPIC -fuse-ld=bfd -Xlinker --add-needed --no-as-needed -lfoamToVTK -ldynamicMesh -llagrangian -lgenericPatchFields -lfileFormats -lOpenFOAM -ldl -lm
Quicksilver Quicksilver is a proxy application that represents some elements of the Mercury workload by solving a simplified dynamic Monte Carlo particle transport problem. Quicksilver is developed by Lawrence Livermore National Laboratory (LLNL) and this test profile currently makes use of the OpenMP CPU threaded code path. Learn more via the OpenBenchmarking.org test page.
Result
OpenBenchmarking.org Figure Of Merit, More Is Better Quicksilver 20230818 Input: CTS2 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 3M 6M 9M 12M 15M SE +/- 23094.01, N = 3 SE +/- 25166.11, N = 3 SE +/- 29627.31, N = 3 SE +/- 43333.33, N = 3 SE +/- 54569.02, N = 3 16300000 14030000 14313333 14376667 11443333 1. (CXX) g++ options: -fopenmp -O3 -march=native Figure Of Merit Per Watt
OpenBenchmarking.org Figure Of Merit Per Watt, More Is Better Quicksilver 20230818 Input: CTS2 EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40K 80K 120K 160K 200K 144074.42 171302.45 156624.76 156169.03 72917.48 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.7 113.1 116.6 EPYC 8324PN - Zen 4C 12.8 81.9 87.7 EPYC 8324P - Zen 4C 11.8 91.4 93.8 EPYC 8324P - Zen 4C, 155W 13.7 92.1 94.7 EPYC 7601 - Zen 1 66.8 156.9 165.2 OpenBenchmarking.org Watts, Fewer Is Better Quicksilver 20230818 CPU Power Consumption Monitor 50 100 150 200 250
CPU Power Consumption Monitor OpenBenchmarking.org Watts CPU Power Consumption Monitor Phoronix Test Suite System Monitoring EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 Min: 6.99 / Avg: 97.94 / Max: 154.25 Min: 6.71 / Avg: 71.8 / Max: 108.62 Min: 6.76 / Avg: 88.61 / Max: 141.34 Min: 11.47 / Avg: 82.06 / Max: 125.14 Min: 32.69 / Avg: 143.24 / Max: 192.83
Xmrig Min Avg Max EPYC 8534PN - Zen 4C 14.1 131.4 143.8 EPYC 8324PN - Zen 4C 13.1 89.2 95.6 EPYC 8324P - Zen 4C 13.6 126.9 138.0 EPYC 8324P - Zen 4C, 155W 13.0 105.5 115.0 EPYC 7601 - Zen 1 65.7 150.1 166.7 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: KawPow - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 155.95 189.15 148.73 171.50 31.23
Result
OpenBenchmarking.org H/s, More Is Better Xmrig 6.21 Variant: KawPow - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4K 8K 12K 16K 20K SE +/- 5.33, N = 3 SE +/- 33.52, N = 3 SE +/- 43.40, N = 3 SE +/- 61.47, N = 3 SE +/- 218.18, N = 11 20487.2 16877.9 18875.2 18088.7 4687.9 1. (CXX) g++ options: -fexceptions -fno-rtti -maes -O3 -Ofast -static-libgcc -static-libstdc++ -rdynamic -lssl -lcrypto -luv -lpthread -lrt -ldl -lhwloc H/s Per Watt
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: KawPow - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 155.95 189.15 148.73 171.50 31.23 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.1 131.4 143.8 EPYC 8324PN - Zen 4C 13.1 89.2 95.6 EPYC 8324P - Zen 4C 13.6 126.9 138.0 EPYC 8324P - Zen 4C, 155W 13.0 105.5 115.0 EPYC 7601 - Zen 1 65.7 150.1 166.7 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
Min Avg Max EPYC 8534PN - Zen 4C 14.9 130.3 143.1 EPYC 8324PN - Zen 4C 13.1 89.0 95.9 EPYC 8324P - Zen 4C 13.5 127.5 138.3 EPYC 8324P - Zen 4C, 155W 13.6 104.6 114.3 EPYC 7601 - Zen 1 65.9 152.8 165.7 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: CryptoNight-Heavy - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 157.28 190.30 148.12 173.49 33.30
Result
OpenBenchmarking.org H/s, More Is Better Xmrig 6.21 Variant: CryptoNight-Heavy - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4K 8K 12K 16K 20K SE +/- 6.48, N = 3 SE +/- 35.06, N = 3 SE +/- 41.98, N = 3 SE +/- 44.77, N = 3 SE +/- 166.19, N = 12 20489.8 16932.8 18883.2 18155.0 5087.4 1. (CXX) g++ options: -fexceptions -fno-rtti -maes -O3 -Ofast -static-libgcc -static-libstdc++ -rdynamic -lssl -lcrypto -luv -lpthread -lrt -ldl -lhwloc H/s Per Watt
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: CryptoNight-Heavy - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 157.28 190.30 148.12 173.49 33.30 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.9 130.3 143.1 EPYC 8324PN - Zen 4C 13.1 89.0 95.9 EPYC 8324P - Zen 4C 13.5 127.5 138.3 EPYC 8324P - Zen 4C, 155W 13.6 104.6 114.3 EPYC 7601 - Zen 1 65.9 152.8 165.7 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
Min Avg Max EPYC 8534PN - Zen 4C 14.6 129.8 142.6 EPYC 8324PN - Zen 4C 13.1 89.0 95.5 EPYC 8324P - Zen 4C 11.2 126.8 137.8 EPYC 8324P - Zen 4C, 155W 13.4 104.8 115.5 EPYC 7601 - Zen 1 65.4 152.1 165.1 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: CryptoNight-Femto UPX2 - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 157.77 188.75 149.20 172.66 33.27
Result
OpenBenchmarking.org H/s, More Is Better Xmrig 6.21 Variant: CryptoNight-Femto UPX2 - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4K 8K 12K 16K 20K SE +/- 12.46, N = 3 SE +/- 103.09, N = 3 SE +/- 50.80, N = 3 SE +/- 49.53, N = 3 SE +/- 202.60, N = 12 20477.1 16799.2 18922.7 18090.9 5061.3 1. (CXX) g++ options: -fexceptions -fno-rtti -maes -O3 -Ofast -static-libgcc -static-libstdc++ -rdynamic -lssl -lcrypto -luv -lpthread -lrt -ldl -lhwloc H/s Per Watt
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: CryptoNight-Femto UPX2 - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 157.77 188.75 149.20 172.66 33.27 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 14.6 129.8 142.6 EPYC 8324PN - Zen 4C 13.1 89.0 95.5 EPYC 8324P - Zen 4C 11.2 126.8 137.8 EPYC 8324P - Zen 4C, 155W 13.4 104.8 115.5 EPYC 7601 - Zen 1 65.4 152.1 165.1 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
Min Avg Max EPYC 8534PN - Zen 4C 13.6 127.4 140.9 EPYC 8324PN - Zen 4C 12.8 89.3 95.4 EPYC 8324P - Zen 4C 12.8 125.6 136.7 EPYC 8324P - Zen 4C, 155W 12.8 104.1 113.6 EPYC 7601 - Zen 1 66.9 152.3 164.3 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: Monero - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 160.74 189.12 150.56 174.20 32.66
Result
OpenBenchmarking.org H/s, More Is Better Xmrig 6.21 Variant: Monero - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 4K 8K 12K 16K 20K SE +/- 5.73, N = 3 SE +/- 34.05, N = 3 SE +/- 56.24, N = 3 SE +/- 43.26, N = 3 SE +/- 155.10, N = 12 20484.3 16892.3 18908.9 18129.0 4973.3 1. (CXX) g++ options: -fexceptions -fno-rtti -maes -O3 -Ofast -static-libgcc -static-libstdc++ -rdynamic -lssl -lcrypto -luv -lpthread -lrt -ldl -lhwloc H/s Per Watt
OpenBenchmarking.org H/s Per Watt, More Is Better Xmrig 6.21 Variant: Monero - Hash Count: 1M EPYC 8534PN - Zen 4C EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 40 80 120 160 200 160.74 189.12 150.56 174.20 32.66 CPU Power Consumption
Min Avg Max EPYC 8534PN - Zen 4C 13.6 127.4 140.9 EPYC 8324PN - Zen 4C 12.8 89.3 95.4 EPYC 8324P - Zen 4C 12.8 125.6 136.7 EPYC 8324P - Zen 4C, 155W 12.8 104.1 113.6 EPYC 7601 - Zen 1 66.9 152.3 164.3 OpenBenchmarking.org Watts, Fewer Is Better Xmrig 6.21 CPU Power Consumption Monitor 50 100 150 200 250
Llama.cpp Min Avg Max EPYC 8324PN - Zen 4C 12.6 82.4 100.5 EPYC 8324P - Zen 4C 11.8 92.9 119.1 EPYC 8324P - Zen 4C, 155W 12.5 89.1 113.3 EPYC 7601 - Zen 1 64.9 132.4 154.0 OpenBenchmarking.org Watts, Fewer Is Better Llama.cpp b1808 CPU Power Consumption Monitor 40 80 120 160 200
OpenBenchmarking.org Tokens Per Second Per Watt, More Is Better Llama.cpp b1808 Model: llama-2-7b.Q4_0.gguf EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0779 0.1558 0.2337 0.3116 0.3895 0.346 0.320 0.330 0.032
Result
OpenBenchmarking.org Tokens Per Second, More Is Better Llama.cpp b1808 Model: llama-2-7b.Q4_0.gguf EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 7 14 21 28 35 SE +/- 0.05, N = 3 SE +/- 0.06, N = 3 SE +/- 0.04, N = 3 SE +/- 0.12, N = 15 28.56 29.72 29.38 4.20 1. (CXX) g++ options: -std=c++11 -fPIC -O3 -pthread -march=native -mtune=native -lopenblas Tokens Per Second Per Watt
OpenBenchmarking.org Tokens Per Second Per Watt, More Is Better Llama.cpp b1808 Model: llama-2-7b.Q4_0.gguf EPYC 8324PN - Zen 4C EPYC 8324P - Zen 4C EPYC 8324P - Zen 4C, 155W EPYC 7601 - Zen 1 0.0779 0.1558 0.2337 0.3116 0.3895 0.346 0.320 0.330 0.032 CPU Power Consumption
Min Avg Max EPYC 8324PN - Zen 4C 12.6 82.4 100.5 EPYC 8324P - Zen 4C 11.8 92.9 119.1 EPYC 8324P - Zen 4C, 155W 12.5 89.1 113.3 EPYC 7601 - Zen 1 64.9 132.4 154.0 OpenBenchmarking.org Watts, Fewer Is Better Llama.cpp b1808 CPU Power Consumption Monitor 40 80 120 160 200
EPYC 8534PN - Zen 4C Processor: AMD EPYC 8534PN 64-Core @ 2.00GHz (64 Cores / 128 Threads), Motherboard: AMD Cinnabar (RCB1009C BIOS), Chipset: AMD Device 14a4, Memory: 6 x 32 GB DRAM-4800MT/s Samsung M321R4GA0BB0-CQKMG, Disk: 1000GB INTEL SSDPE2KX010T8, Graphics: llvmpipe, Network: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, OpenGL: 4.5 Mesa 23.2.1-1ubuntu3.1 (LLVM 15.0.7 256 bits), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
Kernel Notes: Transparent Huge Pages: madviseCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/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-build-config=bootstrap-lto-lean --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 -vProcessor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xaa00212Java Notes: OpenJDK Runtime Environment (build 11.0.21+9-post-Ubuntu-0ubuntu123.10)Python Notes: Python 3.11.6Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of Safe RET + 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 + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 9 January 2024 15:21 by user phoronix.
EPYC 8324PN - Zen 4C Processor: AMD EPYC 8534PN 32-Core @ 2.05GHz (32 Cores / 64 Threads), Motherboard: AMD Cinnabar (RCB1009C BIOS), Chipset: AMD Device 14a4, Memory: 6 x 32 GB DRAM-4800MT/s Samsung M321R4GA0BB0-CQKMG, Disk: 1000GB INTEL SSDPE2KX010T8, Graphics: llvmpipe, Network: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, OpenGL: 4.5 Mesa 23.2.1-1ubuntu3.1 (LLVM 15.0.7 256 bits), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
Kernel Notes: Transparent Huge Pages: madviseCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/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-build-config=bootstrap-lto-lean --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 -vProcessor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xaa00212Java Notes: OpenJDK Runtime Environment (build 11.0.21+9-post-Ubuntu-0ubuntu123.10)Python Notes: Python 3.11.6Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of Safe RET + 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 + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 10 January 2024 14:58 by user phoronix.
EPYC 8324P - Zen 4C Processor: AMD EPYC 8324P 32-Core @ 2.65GHz (32 Cores / 64 Threads), Motherboard: AMD Cinnabar (RCB1009C BIOS), Chipset: AMD Device 14a4, Memory: 6 x 32 GB DRAM-4800MT/s Samsung M321R4GA0BB0-CQKMG, Disk: 1000GB INTEL SSDPE2KX010T8, Graphics: llvmpipe, Network: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, OpenGL: 4.5 Mesa 23.2.1-1ubuntu3.1 (LLVM 15.0.7 256 bits), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
Kernel Notes: Transparent Huge Pages: madviseCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/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-build-config=bootstrap-lto-lean --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 -vProcessor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xaa00212Java Notes: OpenJDK Runtime Environment (build 11.0.21+9-post-Ubuntu-0ubuntu123.10)Python Notes: Python 3.11.6Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of Safe RET + 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 + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 12 January 2024 00:50 by user phoronix.
EPYC 8324P - Zen 4C, 155W Processor: AMD EPYC 8324P 32-Core @ 2.65GHz (32 Cores / 64 Threads), Motherboard: AMD Cinnabar (RCB1009C BIOS), Chipset: AMD Device 14a4, Memory: 6 x 32 GB DRAM-4800MT/s Samsung M321R4GA0BB0-CQKMG, Disk: 1000GB INTEL SSDPE2KX010T8, Graphics: ASPEED, Network: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1200
Kernel Notes: Transparent Huge Pages: madviseCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/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-build-config=bootstrap-lto-lean --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 -vProcessor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xaa00212Java Notes: OpenJDK Runtime Environment (build 11.0.21+9-post-Ubuntu-0ubuntu123.10)Python Notes: Python 3.11.6Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of Safe RET + 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 + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 12 January 2024 19:57 by user phoronix.
EPYC 7601 - Zen 1 Processor: AMD EPYC 7601 32-Core @ 2.20GHz (32 Cores / 64 Threads), Motherboard: TYAN B8026T70AE24HR (V1.02.B10 BIOS), Chipset: AMD 17h, Memory: 8 x 16 GB 2667MT/s Samsung M393A2K40BB2-CTD, Disk: 1000GB INTEL SSDPE2KX010T8 + 280GB INTEL SSDPE21D280GA, Graphics: ASPEED, Monitor: VE228, Network: 2 x Broadcom NetXtreme BCM5720 PCIe
OS: Ubuntu 23.10, Kernel: 6.6.9-060609-generic (x86_64), Desktop: GNOME Shell 45.0, Display Server: X Server 1.21.1.7, OpenGL: 4.5 Mesa 23.2.1-1ubuntu3.1 (LLVM 15.0.7 256 bits), Compiler: GCC 13.2.0, File-System: ext4, Screen Resolution: 1920x1080
Kernel Notes: Transparent Huge Pages: madviseCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --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-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/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-build-config=bootstrap-lto-lean --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 -vProcessor Notes: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0x800126eJava Notes: OpenJDK Runtime Environment (build 11.0.21+9-post-Ubuntu-0ubuntu123.10)Python Notes: Python 3.11.6Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Mitigation of untrained return thunk; SMT vulnerable + spec_rstack_overflow: Mitigation of Safe RET + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional STIBP: disabled RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 13 January 2024 15:35 by user phoronix.