Microsoft Azure EPYC 7003 HBv3 Benchmarks

Azure HBv3 vs. Azure HBv2 benchmarks.

HTML result view exported from: https://openbenchmarking.org/result/2104116-PTS-AZURE71063&grs.

Microsoft Azure EPYC 7003 HBv3 BenchmarksProcessorMotherboardMemoryDiskGraphicsOSKernelCompilerFile-SystemScreen ResolutionSystem LayerAzure HBv3Azure HBv22 x AMD EPYC 7V13 64-Core (120 Cores)Microsoft Virtual Machine (Hyper-V UEFI v4.1 BIOS)442GB2 x 960GB Microsoft NVMe Direct Disk + 32GB Virtual Disk + 515GB Virtual Diskhyperv_fbCentOS Linux 84.18.0-147.8.1.el8_1.x86_64 (x86_64)GCC 8.3.1 20190507nfs1152x864microsoft2 x AMD EPYC 7V12 64-Core (120 Cores)Microsoft Virtual Machine (Hyper-V UEFI v4.0 BIOS)450GB960GB Microsoft NVMe Direct Disk + 32GB Virtual Disk + 515GB Virtual DiskOpenBenchmarking.orgKernel Details- Transparent Huge Pages: alwaysCompiler Details- --build=x86_64-redhat-linux --disable-libmpx --disable-libunwind-exceptions --enable-__cxa_atexit --enable-bootstrap --enable-cet --enable-checking=release --enable-gnu-indirect-function --enable-gnu-unique-object --enable-initfini-array --enable-languages=c,c++,fortran,lto --enable-multilib --enable-offload-targets=nvptx-none --enable-plugin --enable-shared --enable-threads=posix --mandir=/usr/share/man --with-arch_32=x86-64 --with-gcc-major-version-only --with-isl --with-linker-hash-style=gnu --with-tune=generic --without-cuda-driver Processor Details- CPU Microcode: 0xffffffffPython Details- Python 3.6.8Security Details- Azure HBv3: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Vulnerable + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full generic retpoline STIBP: disabled RSB filling + tsx_async_abort: Not affected - Azure HBv2: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full generic retpoline STIBP: disabled RSB filling + tsx_async_abort: Not affected

Microsoft Azure EPYC 7003 HBv3 Benchmarkssvt-vp9: Visual Quality Optimized - Bosphorus 1080ponednn: Deconvolution Batch shapes_3d - u8s8f32 - CPUmnn: inception-v3plaidml: No - Inference - VGG16 - CPUplaidml: No - Inference - VGG19 - CPUonednn: Recurrent Neural Network Inference - bf16bf16bf16 - CPUcompress-zstd: 19 - Decompression Speedcompress-zstd: 8, Long Mode - Decompression Speedcompress-zstd: 19, Long Mode - Decompression Speedbotan: AES-256botan: AES-256 - Decryptquantlib: compress-zstd: 8, Long Mode - Compression Speedrodinia: OpenMP HotSpot3Dsvt-av1: Enc Mode 0 - 1080ponednn: IP Shapes 1D - f32 - CPUfinancebench: Bonds OpenMPonednn: Deconvolution Batch shapes_1d - f32 - CPUfinancebench: Repo OpenMPsvt-hevc: 1 - Bosphorus 1080pbotan: Twofishbotan: Twofish - Decryptbotan: Blowfishbotan: Blowfish - Decryptsvt-av1: Enc Mode 8 - 1080pcompress-zstd: 8 - Compression Speedlulesh: botan: KASUMItnn: CPU - SqueezeNet v1.1hmmer: Pfam Database Searchbotan: KASUMI - Decryptonednn: Matrix Multiply Batch Shapes Transformer - u8s8f32 - CPUgmpbench: Total Timebotan: CAST-256botan: CAST-256 - Decryptrodinia: OpenMP Leukocytemafft: Multiple Sequence Alignment - LSU RNAincompact3d: X3D-benchmarking input.i3dnamd: ATPase Simulation - 327,506 Atomscompress-zstd: 19, Long Mode - Compression Speedbotan: ChaCha20Poly1305botan: ChaCha20Poly1305 - Decryptbuild-linux-kernel: Time To Compilegromacs: Water Benchmarkbuild-nodejs: Time To Compilebuild-llvm: Time To Compilenpb: LU.Chpcg: pennant: leblancbigrodinia: OpenMP LavaMDpennant: sedovbigonednn: Convolution Batch Shapes Auto - f32 - CPUonednn: Deconvolution Batch shapes_3d - f32 - CPUkripke: plaidml: No - Inference - ResNet 50 - CPUncnn: CPU - resnet50ncnn: CPU - alexnetncnn: CPU - vgg16ncnn: CPU - efficientnet-b0ncnn: CPU - shufflenet-v2ncnn: CPU-v2-v2 - mobilenet-v2ncnn: CPU - mobilenetmnn: mobilenet-v1-1.0mnn: MobileNetV2_224mnn: resnet-v2-50mnn: SqueezeNetV1.0tensorflow-lite: SqueezeNetonednn: Recurrent Neural Network Training - bf16bf16bf16 - CPUonednn: Matrix Multiply Batch Shapes Transformer - f32 - CPUonednn: Recurrent Neural Network Inference - u8s8f32 - CPUonednn: Recurrent Neural Network Training - u8s8f32 - CPUonednn: Recurrent Neural Network Inference - f32 - CPUonednn: Recurrent Neural Network Training - f32 - CPUonednn: Deconvolution Batch shapes_1d - u8s8f32 - CPUonednn: IP Shapes 3D - u8s8f32 - CPUonednn: IP Shapes 3D - f32 - CPUsvt-vp9: PSNR/SSIM Optimized - Bosphorus 1080psvt-vp9: VMAF Optimized - Bosphorus 1080psvt-hevc: 10 - Bosphorus 1080psvt-hevc: 7 - Bosphorus 1080psvt-av1: Enc Mode 4 - 1080pcompress-zstd: 19 - Compression Speedrodinia: OpenMP Streamclusterrodinia: OpenMP CFD Solvercloverleaf: Lagrangian-Eulerian Hydrodynamicsminife: SmallAzure HBv3Azure HBv2337.710.45568234.71038.3934.53540.1943717.54256.83727.55412.1315407.4772299.3771.982.4890.1610.881307104884.2942719.5044559196.54687545.39346.319345.776424.738425.15894.0283184.241636.60387.480272.614175.23984.0590.4452234893.4133.691133.66549.35314.326287.5995080.2756636.3667.765658.37042.0089.041111.327163.76056682.8239.06203.33720137.8035.8334190.5752381.58182823418656.2730.036.8947.6721.0413.7123.2427.873.9504.78229.0868.22466320.4843.1030.255501561.773962.336530.898875.1660.4066860.3838553.21204376.91357.06548.33378.0112.27578.27.3798.46016.6613785.3140.590.73213055.35625.5723.12796.8162631.03037.52674.23934.4013938.4441725.2588.0107.9760.1241.12484133719.32812512.105175154.25781336.12280.902283.257348.425349.14778.1042653.134803.53973.175323.615207.86771.1010.5247374155.5113.600113.72457.04016.064318.9050090.3005633.4615.795610.86144.9228.458118.969174.74353829.3437.26863.48650339.1986.0264490.5819811.59557438555505.6064.8010.80133.7832.4221.3937.2052.568.03012.52752.53314.13674885.41294.530.350347791.4601287.03778.3051314.001.0319420.4410129.68321163.51140.30379.91166.909.51269.512.91213.04223.7813165.09OpenBenchmarking.org

SVT-VP9

Tuning: Visual Quality Optimized - Input: Bosphorus 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-VP9 0.3Tuning: Visual Quality Optimized - Input: Bosphorus 1080pAzure HBv3Azure HBv270140210280350SE +/- 4.36, N = 3SE +/- 0.20, N = 3337.71140.591. (CC) gcc options: -O3 -fcommon -fPIE -fPIC -fvisibility=hidden -O2 -pie -rdynamic -lpthread -lrt -lm

oneDNN

Harness: Deconvolution Batch shapes_3d - Data Type: u8s8f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Deconvolution Batch shapes_3d - Data Type: u8s8f32 - Engine: CPUAzure HBv3Azure HBv20.16470.32940.49410.65880.8235SE +/- 0.002442, N = 3SE +/- 0.002363, N = 30.4556820.732130MIN: 0.38MIN: 0.651. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

Mobile Neural Network

Model: inception-v3

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 1.1.3Model: inception-v3Azure HBv3Azure HBv21224364860SE +/- 0.24, N = 15SE +/- 0.95, N = 1234.7155.36MIN: 31.09 / MAX: 427.77MIN: 47.56 / MAX: 509.741. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -O2 -rdynamic -pthread -ldl

PlaidML

FP16: No - Mode: Inference - Network: VGG16 - Device: CPU

OpenBenchmarking.orgFPS, More Is BetterPlaidMLFP16: No - Mode: Inference - Network: VGG16 - Device: CPUAzure HBv3Azure HBv2918273645SE +/- 0.39, N = 3SE +/- 0.21, N = 338.3925.57

PlaidML

FP16: No - Mode: Inference - Network: VGG19 - Device: CPU

OpenBenchmarking.orgFPS, More Is BetterPlaidMLFP16: No - Mode: Inference - Network: VGG19 - Device: CPUAzure HBv3Azure HBv2816243240SE +/- 0.42, N = 4SE +/- 0.31, N = 1534.5323.12

oneDNN

Harness: Recurrent Neural Network Inference - Data Type: bf16bf16bf16 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Recurrent Neural Network Inference - Data Type: bf16bf16bf16 - Engine: CPUAzure HBv3Azure HBv22004006008001000SE +/- 8.06, N = 15SE +/- 6.35, N = 15540.19796.82MIN: 462.7MIN: 726.181. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

Zstd Compression

Compression Level: 19 - Decompression Speed

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.4.9Compression Level: 19 - Decompression SpeedAzure HBv3Azure HBv28001600240032004000SE +/- 6.14, N = 15SE +/- 1.17, N = 153717.52631.01. (CC) gcc options: -O3 -pthread -lz -llzma

Zstd Compression

Compression Level: 8, Long Mode - Decompression Speed

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.4.9Compression Level: 8, Long Mode - Decompression SpeedAzure HBv3Azure HBv29001800270036004500SE +/- 4.89, N = 7SE +/- 4.57, N = 34256.83037.51. (CC) gcc options: -O3 -pthread -lz -llzma

Zstd Compression

Compression Level: 19, Long Mode - Decompression Speed

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.4.9Compression Level: 19, Long Mode - Decompression SpeedAzure HBv3Azure HBv28001600240032004000SE +/- 12.70, N = 15SE +/- 2.70, N = 153727.52674.21. (CC) gcc options: -O3 -pthread -lz -llzma

Botan

Test: AES-256

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: AES-256Azure HBv3Azure HBv212002400360048006000SE +/- 3.73, N = 3SE +/- 2.68, N = 35412.133934.401. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Botan

Test: AES-256 - Decrypt

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: AES-256 - DecryptAzure HBv3Azure HBv212002400360048006000SE +/- 7.85, N = 3SE +/- 3.26, N = 35407.483938.441. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

QuantLib

OpenBenchmarking.orgMFLOPS, More Is BetterQuantLib 1.21Azure HBv3Azure HBv25001000150020002500SE +/- 4.80, N = 3SE +/- 0.77, N = 32299.31725.21. (CXX) g++ options: -O3 -march=native -O2 -rdynamic -lboost_timer -lboost_system -lboost_chrono

Zstd Compression

Compression Level: 8, Long Mode - Compression Speed

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.4.9Compression Level: 8, Long Mode - Compression SpeedAzure HBv3Azure HBv2170340510680850SE +/- 6.99, N = 7SE +/- 0.81, N = 3771.9588.01. (CC) gcc options: -O3 -pthread -lz -llzma

Rodinia

Test: OpenMP HotSpot3D

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 3.1Test: OpenMP HotSpot3DAzure HBv3Azure HBv220406080100SE +/- 1.18, N = 15SE +/- 0.89, N = 382.49107.981. (CXX) g++ options: -O2 -lOpenCL

SVT-AV1

Encoder Mode: Enc Mode 0 - Input: 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 0.8Encoder Mode: Enc Mode 0 - Input: 1080pAzure HBv3Azure HBv20.03620.07240.10860.14480.181SE +/- 0.000, N = 3SE +/- 0.001, N = 30.1610.1241. (CXX) g++ options: -O3 -fcommon -fPIE -fPIC -pie

oneDNN

Harness: IP Shapes 1D - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: IP Shapes 1D - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv20.25310.50620.75931.01241.2655SE +/- 0.012358, N = 3SE +/- 0.011740, N = 150.8813071.124840MIN: 0.76MIN: 1.011. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

FinanceBench

Benchmark: Bonds OpenMP

OpenBenchmarking.orgms, Fewer Is BetterFinanceBench 2016-07-25Benchmark: Bonds OpenMPAzure HBv3Azure HBv230K60K90K120K150KSE +/- 454.05, N = 3SE +/- 476.22, N = 3104884.29133719.331. (CXX) g++ options: -O3 -march=native -fopenmp

oneDNN

Harness: Deconvolution Batch shapes_1d - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Deconvolution Batch shapes_1d - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv23691215SE +/- 0.09389, N = 3SE +/- 0.11811, N = 69.5044512.10510MIN: 4.19MIN: 5.81. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

FinanceBench

Benchmark: Repo OpenMP

OpenBenchmarking.orgms, Fewer Is BetterFinanceBench 2016-07-25Benchmark: Repo OpenMPAzure HBv3Azure HBv216K32K48K64K80KSE +/- 228.72, N = 3SE +/- 166.77, N = 359196.5575154.261. (CXX) g++ options: -O3 -march=native -fopenmp

SVT-HEVC

Tuning: 1 - Input: Bosphorus 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-HEVC 1.5.0Tuning: 1 - Input: Bosphorus 1080pAzure HBv3Azure HBv21020304050SE +/- 0.60, N = 3SE +/- 0.13, N = 345.3936.121. (CC) gcc options: -fPIE -fPIC -O2 -O3 -pie -rdynamic -lpthread -lrt

Botan

Test: Twofish

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: TwofishAzure HBv3Azure HBv280160240320400SE +/- 1.10, N = 3SE +/- 0.15, N = 3346.32280.901. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Botan

Test: Twofish - Decrypt

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: Twofish - DecryptAzure HBv3Azure HBv280160240320400SE +/- 0.77, N = 3SE +/- 0.22, N = 3345.78283.261. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Botan

Test: Blowfish

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: BlowfishAzure HBv3Azure HBv290180270360450SE +/- 0.47, N = 3SE +/- 0.42, N = 3424.74348.431. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Botan

Test: Blowfish - Decrypt

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: Blowfish - DecryptAzure HBv3Azure HBv290180270360450SE +/- 0.42, N = 3SE +/- 0.56, N = 3425.16349.151. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

SVT-AV1

Encoder Mode: Enc Mode 8 - Input: 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 0.8Encoder Mode: Enc Mode 8 - Input: 1080pAzure HBv3Azure HBv220406080100SE +/- 0.34, N = 3SE +/- 0.65, N = 394.0378.101. (CXX) g++ options: -O3 -fcommon -fPIE -fPIC -pie

Zstd Compression

Compression Level: 8 - Compression Speed

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.4.9Compression Level: 8 - Compression SpeedAzure HBv3Azure HBv27001400210028003500SE +/- 45.10, N = 3SE +/- 33.03, N = 153184.22653.11. (CC) gcc options: -O3 -pthread -lz -llzma

LULESH

OpenBenchmarking.orgz/s, More Is BetterLULESH 2.0.3Azure HBv3Azure HBv29K18K27K36K45KSE +/- 476.10, N = 3SE +/- 44.77, N = 341636.6034803.541. (CXX) g++ options: -O3 -fopenmp -lm -fexceptions -pthread -lmpi_cxx -lmpi

Botan

Test: KASUMI

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: KASUMIAzure HBv3Azure HBv220406080100SE +/- 0.04, N = 3SE +/- 0.07, N = 387.4873.181. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

TNN

Target: CPU - Model: SqueezeNet v1.1

OpenBenchmarking.orgms, Fewer Is BetterTNN 0.2.3Target: CPU - Model: SqueezeNet v1.1Azure HBv3Azure HBv270140210280350SE +/- 0.15, N = 3SE +/- 0.19, N = 3272.61323.62MIN: 272.1 / MAX: 273.5MIN: 322.04 / MAX: 326.051. (CXX) g++ options: -fopenmp -pthread -fvisibility=hidden -O2 -rdynamic -ldl

Timed HMMer Search

Pfam Database Search

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed HMMer Search 3.3.1Pfam Database SearchAzure HBv3Azure HBv250100150200250SE +/- 1.00, N = 3SE +/- 1.08, N = 3175.24207.871. (CC) gcc options: -O3 -pthread -lhmmer -leasel -lm

Botan

Test: KASUMI - Decrypt

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: KASUMI - DecryptAzure HBv3Azure HBv220406080100SE +/- 0.02, N = 3SE +/- 0.04, N = 384.0671.101. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

oneDNN

Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPUAzure HBv3Azure HBv20.11810.23620.35430.47240.5905SE +/- 0.004912, N = 4SE +/- 0.004967, N = 30.4452230.524737MIN: 0.39MIN: 0.481. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

GNU GMP GMPbench

Total Time

OpenBenchmarking.orgGMPbench Score, More Is BetterGNU GMP GMPbench 6.2.1Total TimeAzure HBv3Azure HBv2100020003000400050004893.44155.51. (CC) gcc options: -O3 -fomit-frame-pointer -lm

Botan

Test: CAST-256

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: CAST-256Azure HBv3Azure HBv2306090120150SE +/- 0.03, N = 3SE +/- 0.07, N = 3133.69113.601. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Botan

Test: CAST-256 - Decrypt

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: CAST-256 - DecryptAzure HBv3Azure HBv2306090120150SE +/- 0.02, N = 3SE +/- 0.01, N = 3133.67113.721. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Rodinia

Test: OpenMP Leukocyte

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 3.1Test: OpenMP LeukocyteAzure HBv3Azure HBv21326395265SE +/- 0.43, N = 8SE +/- 0.57, N = 349.3557.041. (CXX) g++ options: -O2 -lOpenCL

Timed MAFFT Alignment

Multiple Sequence Alignment - LSU RNA

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed MAFFT Alignment 7.471Multiple Sequence Alignment - LSU RNAAzure HBv3Azure HBv248121620SE +/- 0.10, N = 15SE +/- 0.06, N = 314.3316.061. (CC) gcc options: -std=c99 -O3 -lm -lpthread

Xcompact3d Incompact3d

Input: X3D-benchmarking input.i3d

OpenBenchmarking.orgSeconds, Fewer Is BetterXcompact3d Incompact3d 2021-03-11Input: X3D-benchmarking input.i3dAzure HBv3Azure HBv270140210280350SE +/- 0.17, N = 3SE +/- 0.07, N = 3287.60318.911. (F9X) gfortran options: -cpp -O2 -funroll-loops -floop-optimize -fcray-pointer -fbacktrace -fexceptions -pthread -lmpi_usempif08 -lmpi_mpifh -lmpi

NAMD

ATPase Simulation - 327,506 Atoms

OpenBenchmarking.orgdays/ns, Fewer Is BetterNAMD 2.14ATPase Simulation - 327,506 AtomsAzure HBv3Azure HBv20.06760.13520.20280.27040.338SE +/- 0.00027, N = 3SE +/- 0.00059, N = 30.275660.30056

Zstd Compression

Compression Level: 19, Long Mode - Compression Speed

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.4.9Compression Level: 19, Long Mode - Compression SpeedAzure HBv3Azure HBv2816243240SE +/- 0.34, N = 15SE +/- 0.38, N = 1536.333.41. (CC) gcc options: -O3 -pthread -lz -llzma

Botan

Test: ChaCha20Poly1305

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: ChaCha20Poly1305Azure HBv3Azure HBv2140280420560700SE +/- 0.27, N = 3SE +/- 0.90, N = 3667.77615.801. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Botan

Test: ChaCha20Poly1305 - Decrypt

OpenBenchmarking.orgMiB/s, More Is BetterBotan 2.17.3Test: ChaCha20Poly1305 - DecryptAzure HBv3Azure HBv2140280420560700SE +/- 0.06, N = 3SE +/- 0.78, N = 3658.37610.861. (CXX) g++ options: -fstack-protector -m64 -pthread -lbotan-2 -ldl -lrt

Timed Linux Kernel Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Linux Kernel Compilation 5.10.20Time To CompileAzure HBv3Azure HBv21020304050SE +/- 0.58, N = 15SE +/- 0.58, N = 1342.0144.92

GROMACS

Water Benchmark

OpenBenchmarking.orgNs Per Day, More Is BetterGROMACS 2020.3Water BenchmarkAzure HBv3Azure HBv23691215SE +/- 0.009, N = 3SE +/- 0.009, N = 39.0418.4581. (CXX) g++ options: -O2 -pthread -lrt -lpthread -lm

Timed Node.js Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed Node.js Compilation 15.11Time To CompileAzure HBv3Azure HBv2306090120150SE +/- 1.09, N = 3SE +/- 0.83, N = 3111.33118.97

Timed LLVM Compilation

Time To Compile

OpenBenchmarking.orgSeconds, Fewer Is BetterTimed LLVM Compilation 10.0Time To CompileAzure HBv3Azure HBv24080120160200SE +/- 1.89, N = 3SE +/- 1.99, N = 3163.76174.74

NAS Parallel Benchmarks

Test / Class: LU.C

OpenBenchmarking.orgTotal Mop/s, More Is BetterNAS Parallel Benchmarks 3.4Test / Class: LU.CAzure HBv3Azure HBv212K24K36K48K60KSE +/- 428.57, N = 14SE +/- 21.24, N = 356682.8253829.341. (F9X) gfortran options: -O3 -march=native -fexceptions -pthread -lmpi_usempif08 -lmpi_mpifh -lmpi

High Performance Conjugate Gradient

OpenBenchmarking.orgGFLOP/s, More Is BetterHigh Performance Conjugate Gradient 3.1Azure HBv3Azure HBv2918273645SE +/- 0.06, N = 3SE +/- 0.06, N = 339.0637.271. (CXX) g++ options: -O3 -ffast-math -ftree-vectorize -fexceptions -pthread -lmpi_cxx -lmpi

Pennant

Test: leblancbig

OpenBenchmarking.orgHydro Cycle Time - Seconds, Fewer Is BetterPennant 1.0.1Test: leblancbigAzure HBv3Azure HBv20.78451.5692.35353.1383.9225SE +/- 0.017738, N = 3SE +/- 0.009939, N = 33.3372013.4865031. (CXX) g++ options: -fopenmp -fexceptions -pthread -lmpi_cxx -lmpi

Rodinia

Test: OpenMP LavaMD

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 3.1Test: OpenMP LavaMDAzure HBv3Azure HBv2918273645SE +/- 0.23, N = 3SE +/- 0.20, N = 337.8039.201. (CXX) g++ options: -O2 -lOpenCL

Pennant

Test: sedovbig

OpenBenchmarking.orgHydro Cycle Time - Seconds, Fewer Is BetterPennant 1.0.1Test: sedovbigAzure HBv3Azure HBv2246810SE +/- 0.003905, N = 3SE +/- 0.012548, N = 35.8334196.0264491. (CXX) g++ options: -fopenmp -fexceptions -pthread -lmpi_cxx -lmpi

oneDNN

Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv20.13090.26180.39270.52360.6545SE +/- 0.005232, N = 3SE +/- 0.000573, N = 30.5752380.581981MIN: 0.5MIN: 0.531. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: Deconvolution Batch shapes_3d - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Deconvolution Batch shapes_3d - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv20.3590.7181.0771.4361.795SE +/- 0.00509, N = 3SE +/- 0.01470, N = 31.581821.59557MIN: 1.49MIN: 1.51. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

Kripke

OpenBenchmarking.orgThroughput FoM, More Is BetterKripke 1.2.4Azure HBv3Azure HBv220M40M60M80M100MSE +/- 1839362.01, N = 15SE +/- 812916.11, N = 1582341865438555501. (CXX) g++ options: -O2 -fopenmp

PlaidML

FP16: No - Mode: Inference - Network: ResNet 50 - Device: CPU

OpenBenchmarking.orgFPS, More Is BetterPlaidMLFP16: No - Mode: Inference - Network: ResNet 50 - Device: CPUAzure HBv3Azure HBv2246810SE +/- 0.16, N = 9SE +/- 0.03, N = 36.275.60

NCNN

Target: CPU - Model: resnet50

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20201218Target: CPU - Model: resnet50Azure HBv3Azure HBv21428425670SE +/- 0.90, N = 9SE +/- 2.16, N = 1230.0364.80MIN: 25.77 / MAX: 962.03MIN: 44.36 / MAX: 2196.771. (CXX) g++ options: -O2 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: alexnet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20201218Target: CPU - Model: alexnetAzure HBv3Azure HBv23691215SE +/- 0.57, N = 9SE +/- 0.21, N = 126.8910.80MIN: 5.85 / MAX: 759.55MIN: 9.2 / MAX: 35.441. (CXX) g++ options: -O2 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: vgg16

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20201218Target: CPU - Model: vgg16Azure HBv3Azure HBv2306090120150SE +/- 2.09, N = 9SE +/- 6.58, N = 1247.67133.78MIN: 40.11 / MAX: 1444.13MIN: 82.15 / MAX: 2364.481. (CXX) g++ options: -O2 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: efficientnet-b0

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20201218Target: CPU - Model: efficientnet-b0Azure HBv3Azure HBv2816243240SE +/- 3.62, N = 9SE +/- 1.94, N = 1221.0432.42MIN: 12.75 / MAX: 4928.42MIN: 20.32 / MAX: 786.111. (CXX) g++ options: -O2 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: shufflenet-v2

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20201218Target: CPU - Model: shufflenet-v2Azure HBv3Azure HBv2510152025SE +/- 0.83, N = 9SE +/- 1.32, N = 1213.7121.39MIN: 9.5 / MAX: 286.31MIN: 14.34 / MAX: 1188.391. (CXX) g++ options: -O2 -rdynamic -lgomp -lpthread

NCNN

Target: CPU-v2-v2 - Model: mobilenet-v2

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20201218Target: CPU-v2-v2 - Model: mobilenet-v2Azure HBv3Azure HBv2918273645SE +/- 3.10, N = 9SE +/- 2.17, N = 1223.2437.20MIN: 10.66 / MAX: 3825.91MIN: 13.35 / MAX: 4343.961. (CXX) g++ options: -O2 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: mobilenet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20201218Target: CPU - Model: mobilenetAzure HBv3Azure HBv21224364860SE +/- 1.74, N = 9SE +/- 2.25, N = 1227.8752.56MIN: 21.41 / MAX: 63.52MIN: 40.97 / MAX: 511.31. (CXX) g++ options: -O2 -rdynamic -lgomp -lpthread

Mobile Neural Network

Model: mobilenet-v1-1.0

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 1.1.3Model: mobilenet-v1-1.0Azure HBv3Azure HBv2246810SE +/- 0.151, N = 15SE +/- 0.500, N = 123.9508.030MIN: 2.55 / MAX: 5.41MIN: 4.5 / MAX: 15.231. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -O2 -rdynamic -pthread -ldl

Mobile Neural Network

Model: MobileNetV2_224

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 1.1.3Model: MobileNetV2_224Azure HBv3Azure HBv23691215SE +/- 0.085, N = 15SE +/- 0.587, N = 124.78212.527MIN: 3.71 / MAX: 61.38MIN: 7.24 / MAX: 67.341. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -O2 -rdynamic -pthread -ldl

Mobile Neural Network

Model: resnet-v2-50

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 1.1.3Model: resnet-v2-50Azure HBv3Azure HBv21224364860SE +/- 0.25, N = 15SE +/- 1.18, N = 1229.0952.53MIN: 25.79 / MAX: 252.29MIN: 41.27 / MAX: 326.881. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -O2 -rdynamic -pthread -ldl

Mobile Neural Network

Model: SqueezeNetV1.0

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 1.1.3Model: SqueezeNetV1.0Azure HBv3Azure HBv248121620SE +/- 0.113, N = 15SE +/- 0.657, N = 128.22414.136MIN: 5.66 / MAX: 54.17MIN: 10.98 / MAX: 78.481. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -O2 -rdynamic -pthread -ldl

TensorFlow Lite

Model: SqueezeNet

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2020-08-23Model: SqueezeNetAzure HBv3Azure HBv216K32K48K64K80KSE +/- 2854.14, N = 15SE +/- 1151.94, N = 1566320.474885.4

oneDNN

Harness: Recurrent Neural Network Training - Data Type: bf16bf16bf16 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Recurrent Neural Network Training - Data Type: bf16bf16bf16 - Engine: CPUAzure HBv3Azure HBv230060090012001500SE +/- 14.29, N = 15SE +/- 11.10, N = 15843.101294.53MIN: 711.66MIN: 1179.191. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv20.07880.15760.23640.31520.394SE +/- 0.000278, N = 3SE +/- 0.006544, N = 120.2555010.350347MIN: 0.22MIN: 0.31. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: Recurrent Neural Network Inference - Data Type: u8s8f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Recurrent Neural Network Inference - Data Type: u8s8f32 - Engine: CPUAzure HBv3Azure HBv22004006008001000SE +/- 8.80, N = 15SE +/- 5.73, N = 15561.77791.46MIN: 471.83MIN: 718.741. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: Recurrent Neural Network Training - Data Type: u8s8f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Recurrent Neural Network Training - Data Type: u8s8f32 - Engine: CPUAzure HBv3Azure HBv230060090012001500SE +/- 102.39, N = 14SE +/- 13.52, N = 15962.341287.03MIN: 722.13MIN: 1149.251. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv22004006008001000SE +/- 8.62, N = 15SE +/- 8.13, N = 3530.90778.31MIN: 458.96MIN: 738.781. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv230060090012001500SE +/- 25.25, N = 12SE +/- 16.54, N = 15875.171314.00MIN: 727.38MIN: 1140.591. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: Deconvolution Batch shapes_1d - Data Type: u8s8f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: Deconvolution Batch shapes_1d - Data Type: u8s8f32 - Engine: CPUAzure HBv3Azure HBv20.23220.46440.69660.92881.161SE +/- 0.004961, N = 3SE +/- 0.030825, N = 150.4066861.031942MIN: 0.35MIN: 0.831. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: IP Shapes 3D - Data Type: u8s8f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: IP Shapes 3D - Data Type: u8s8f32 - Engine: CPUAzure HBv3Azure HBv20.09920.19840.29760.39680.496SE +/- 0.009558, N = 12SE +/- 0.003995, N = 70.3838550.441012MIN: 0.32MIN: 0.41. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

oneDNN

Harness: IP Shapes 3D - Data Type: f32 - Engine: CPU

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.1.2Harness: IP Shapes 3D - Data Type: f32 - Engine: CPUAzure HBv3Azure HBv23691215SE +/- 0.02162, N = 3SE +/- 0.28898, N = 153.212049.68321MIN: 2.58MIN: 5.561. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -O2 -pie -lpthread -ldl

SVT-VP9

Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-VP9 0.3Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080pAzure HBv3Azure HBv280160240320400SE +/- 5.29, N = 15SE +/- 9.44, N = 12376.91163.511. (CC) gcc options: -O3 -fcommon -fPIE -fPIC -fvisibility=hidden -O2 -pie -rdynamic -lpthread -lrt -lm

SVT-VP9

Tuning: VMAF Optimized - Input: Bosphorus 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-VP9 0.3Tuning: VMAF Optimized - Input: Bosphorus 1080pAzure HBv3Azure HBv280160240320400SE +/- 23.14, N = 12SE +/- 5.31, N = 15357.06140.301. (CC) gcc options: -O3 -fcommon -fPIE -fPIC -fvisibility=hidden -O2 -pie -rdynamic -lpthread -lrt -lm

SVT-HEVC

Tuning: 10 - Input: Bosphorus 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-HEVC 1.5.0Tuning: 10 - Input: Bosphorus 1080pAzure HBv3Azure HBv2120240360480600SE +/- 7.76, N = 3SE +/- 21.86, N = 15548.33379.911. (CC) gcc options: -fPIE -fPIC -O2 -O3 -pie -rdynamic -lpthread -lrt

SVT-HEVC

Tuning: 7 - Input: Bosphorus 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-HEVC 1.5.0Tuning: 7 - Input: Bosphorus 1080pAzure HBv3Azure HBv280160240320400SE +/- 4.22, N = 3SE +/- 6.53, N = 12378.01166.901. (CC) gcc options: -fPIE -fPIC -O2 -O3 -pie -rdynamic -lpthread -lrt

SVT-AV1

Encoder Mode: Enc Mode 4 - Input: 1080p

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 0.8Encoder Mode: Enc Mode 4 - Input: 1080pAzure HBv3Azure HBv23691215SE +/- 0.031, N = 3SE +/- 0.218, N = 1512.2759.5121. (CXX) g++ options: -O3 -fcommon -fPIE -fPIC -pie

Zstd Compression

Compression Level: 19 - Compression Speed

OpenBenchmarking.orgMB/s, More Is BetterZstd Compression 1.4.9Compression Level: 19 - Compression SpeedAzure HBv3Azure HBv220406080100SE +/- 0.71, N = 15SE +/- 1.18, N = 1578.269.51. (CC) gcc options: -O3 -pthread -lz -llzma

Rodinia

Test: OpenMP Streamcluster

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 3.1Test: OpenMP StreamclusterAzure HBv3Azure HBv23691215SE +/- 0.196, N = 15SE +/- 0.453, N = 157.37912.9121. (CXX) g++ options: -O2 -lOpenCL

Rodinia

Test: OpenMP CFD Solver

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 3.1Test: OpenMP CFD SolverAzure HBv3Azure HBv23691215SE +/- 0.801, N = 12SE +/- 0.748, N = 128.46013.0421. (CXX) g++ options: -O2 -lOpenCL

CloverLeaf

Lagrangian-Eulerian Hydrodynamics

OpenBenchmarking.orgSeconds, Fewer Is BetterCloverLeafLagrangian-Eulerian HydrodynamicsAzure HBv3Azure HBv2612182430SE +/- 0.83, N = 15SE +/- 0.45, N = 1216.6623.781. (F9X) gfortran options: -O3 -march=native -funroll-loops -fopenmp

miniFE

Problem Size: Small

OpenBenchmarking.orgCG Mflops, More Is BetterminiFE 2.2Problem Size: SmallAzure HBv3Azure HBv23K6K9K12K15KSE +/- 387.39, N = 15SE +/- 608.74, N = 1213785.3013165.091. (CXX) g++ options: -O3 -fopenmp -fexceptions -pthread -lmpi_cxx -lmpi

Phoronix Test Suite v10.8.4