onednn onnx threadripper

AMD Ryzen Threadripper 3990X 64-Core testing with a Gigabyte TRX40 AORUS PRO WIFI (F4p BIOS) and AMD Radeon RX 5700 8GB on Pop 21.10 via the Phoronix Test Suite.

Compare your own system(s) to this result file with the Phoronix Test Suite by running the command: phoronix-test-suite benchmark 2203314-PTS-ONEDNNON39
Jump To Table - Results

View

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

Limit displaying results to tests within:

HPC - High Performance Computing 2 Tests
Machine Learning 2 Tests

Statistics

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

Graph Settings

Force Line Graphs Where Applicable
Convert To Scalar Where Applicable
Prefer Vertical Bar Graphs

Multi-Way Comparison

Condense Multi-Option Tests Into Single Result Graphs

Table

Show Detailed System Result Table

Run Management

Highlight
Result
Hide
Result
Result
Identifier
Performance Per
Dollar
Date
Run
  Test
  Duration
A
March 31 2022
  2 Hours, 35 Minutes
B
March 31 2022
  2 Hours, 31 Minutes
C
March 31 2022
  1 Hour, 55 Minutes
D
March 31 2022
  2 Hours, 48 Minutes
Invert Hiding All Results Option
  2 Hours, 27 Minutes

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

A

Kernel Notes: Transparent Huge Pages: madvise
Compiler 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,brig,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-targets=nvptx-none=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/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 -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0x8301039
Python Notes: Python 3.9.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected

B

C

D

Processor: AMD Ryzen Threadripper 3990X 64-Core @ 2.90GHz (64 Cores / 128 Threads), Motherboard: Gigabyte TRX40 AORUS PRO WIFI (F4p BIOS), Chipset: AMD Starship/Matisse, Memory: 128GB, Disk: Samsung SSD 970 EVO Plus 500GB, Graphics: AMD Radeon RX 5700 8GB (1750/875MHz), Audio: AMD Navi 10 HDMI Audio, Monitor: DELL P2415Q, Network: Intel I211 + Intel Wi-Fi 6 AX200

OS: Pop 21.10, Kernel: 5.17.0-rc1-sched-core-phx (x86_64), Desktop: GNOME Shell 40.5, Display Server: X Server, OpenGL: 4.6 Mesa 21.2.2 (LLVM 12.0.1), Vulkan: 1.2.182, Compiler: GCC 11.2.0, File-System: ext4, Screen Resolution: 3840x2160

onednn onnx threadripperOpenBenchmarking.orgPhoronix Test SuiteAMD Ryzen Threadripper 3990X 64-Core @ 2.90GHz (64 Cores / 128 Threads)Gigabyte TRX40 AORUS PRO WIFI (F4p BIOS)AMD Starship/Matisse128GBSamsung SSD 970 EVO Plus 500GBAMD Radeon RX 5700 8GB (1750/875MHz)AMD Navi 10 HDMI AudioDELL P2415QIntel I211 + Intel Wi-Fi 6 AX200Pop 21.105.17.0-rc1-sched-core-phx (x86_64)GNOME Shell 40.5X Server4.6 Mesa 21.2.2 (LLVM 12.0.1)1.2.182GCC 11.2.0ext43840x2160ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerOpenGLVulkanCompilerFile-SystemScreen ResolutionOnednn Onnx Threadripper BenchmarksSystem Logs- Transparent Huge Pages: madvise- --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,brig,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-targets=nvptx-none=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/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 -v - Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0x8301039- Python 3.9.7- itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected

ABCDResult OverviewPhoronix Test Suite100%105%111%116%122%ONNX RuntimeONNX RuntimeoneDNNONNX RuntimeoneDNNoneDNNoneDNNoneDNNoneDNNoneDNNoneDNNoneDNNoneDNNoneDNNoneDNNONNX RuntimeONNX RuntimeONNX RuntimeONNX RuntimeoneDNNONNX RuntimeONNX RuntimeONNX RuntimeoneDNNoneDNNONNX RuntimeoneDNNoneDNNoneDNNONNX Runtimebertsquad-12 - CPU - Standardsuper-resolution-10 - CPU - StandardIP Shapes 3D - f32 - CPUGPT-2 - CPU - StandardIP Shapes 1D - u8s8f32 - CPUM.M.B.S.T - f32 - CPUD.B.s - u8s8f32 - CPUM.M.B.S.T - u8s8f32 - CPUIP Shapes 1D - f32 - CPUC.B.S.A - f32 - CPUR.N.N.I - f32 - CPUR.N.N.I - bf16bf16bf16 - CPUD.B.s - f32 - CPUR.N.N.T - bf16bf16bf16 - CPUIP Shapes 3D - u8s8f32 - CPUArcFace ResNet-100 - CPU - Standardfcn-resnet101-11 - CPU - Standardbertsquad-12 - CPU - Parallelfcn-resnet101-11 - CPU - ParallelR.N.N.T - f32 - CPUyolov4 - CPU - Standardsuper-resolution-10 - CPU - ParallelGPT-2 - CPU - ParallelR.N.N.I - u8s8f32 - CPUR.N.N.T - u8s8f32 - CPUArcFace ResNet-100 - CPU - ParallelD.B.s - u8s8f32 - CPUC.B.S.A - u8s8f32 - CPUD.B.s - f32 - CPUyolov4 - CPU - Parallel

onednn onnx threadripperonnx: bertsquad-12 - CPU - Standardonednn: IP Shapes 3D - f32 - CPUonnx: GPT-2 - CPU - Standardonednn: Convolution Batch Shapes Auto - f32 - CPUonednn: Recurrent Neural Network Inference - f32 - CPUonednn: Deconvolution Batch shapes_1d - f32 - CPUonednn: Recurrent Neural Network Training - bf16bf16bf16 - CPUonednn: IP Shapes 3D - u8s8f32 - CPUonnx: ArcFace ResNet-100 - CPU - Standardonnx: fcn-resnet101-11 - CPU - Standardonnx: bertsquad-12 - CPU - Parallelonnx: fcn-resnet101-11 - CPU - Parallelonednn: Recurrent Neural Network Training - f32 - CPUonnx: yolov4 - CPU - Standardonnx: super-resolution-10 - CPU - Parallelonnx: GPT-2 - CPU - Parallelonednn: Recurrent Neural Network Inference - u8s8f32 - CPUonednn: Recurrent Neural Network Training - u8s8f32 - CPUonnx: ArcFace ResNet-100 - CPU - Parallelonednn: Deconvolution Batch shapes_3d - u8s8f32 - CPUonednn: Convolution Batch Shapes Auto - u8s8f32 - CPUonednn: Deconvolution Batch shapes_3d - f32 - CPUonnx: yolov4 - CPU - Parallelonnx: super-resolution-10 - CPU - Standardonednn: Matrix Multiply Batch Shapes Transformer - u8s8f32 - CPUonednn: Recurrent Neural Network Inference - bf16bf16bf16 - CPUonednn: Matrix Multiply Batch Shapes Transformer - f32 - CPUonednn: Deconvolution Batch shapes_1d - u8s8f32 - CPUonednn: IP Shapes 1D - u8s8f32 - CPUonednn: IP Shapes 1D - f32 - CPUABCD5315.5438742190.9412661260.206.680055028.061.13774995153424824959.96293381534611236.754954.3410880.9791356.394302.10617361732311.60371208.5237.591651.526192.184332.009536476.2707247100.9100561251.246.821664997.821.117741010156425805003.99293378035121246.075034.8310720.9927136.433302.11025362640111.34491242.446.930131.498712.351611.964206466.2866348230.9041101221.126.850395011.491.119271017153432814964.59295378435291238.605003.3810790.9870206.446892.11212362756011.90351250.997.579411.560002.374031.993836426.4080644410.9284041211.446.906074884.901.10705991157421814882.28300373134951223.534950.9710790.9848196.443302.11146361737511.752151254.397.049811.455112.421761.91681OpenBenchmarking.org

ONNX Runtime

ONNX Runtime is developed by Microsoft and partners as a open-source, cross-platform, high performance machine learning inferencing and training accelerator. This test profile runs the ONNX Runtime with various models available from the ONNX Zoo. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: bertsquad-12 - Device: CPU - Executor: StandardADCB140280420560700SE +/- 3.69, N = 3SE +/- 1.32, N = 3SE +/- 2.02, N = 3SE +/- 0.67, N = 35316426466471. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: bertsquad-12 - Device: CPU - Executor: StandardADCB110220330440550Min: 525 / Avg: 530.5 / Max: 537.5Min: 639.5 / Avg: 642 / Max: 644Min: 642 / Avg: 645.5 / Max: 649Min: 646 / Avg: 646.67 / Max: 6481. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of Intel oneAPI. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 3D - Data Type: f32 - Engine: CPUDCBA246810SE +/- 0.07364, N = 3SE +/- 0.06711, N = 3SE +/- 0.00710, N = 3SE +/- 0.01655, N = 36.408066.286636.270725.54387MIN: 6.17MIN: 6.01MIN: 6.08MIN: 5.31. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 3D - Data Type: f32 - Engine: CPUDCBA3691215Min: 6.3 / Avg: 6.41 / Max: 6.55Min: 6.2 / Avg: 6.29 / Max: 6.42Min: 6.26 / Avg: 6.27 / Max: 6.28Min: 5.51 / Avg: 5.54 / Max: 5.561. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

ONNX Runtime

ONNX Runtime is developed by Microsoft and partners as a open-source, cross-platform, high performance machine learning inferencing and training accelerator. This test profile runs the ONNX Runtime with various models available from the ONNX Zoo. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: GPT-2 - Device: CPU - Executor: StandardADBC10002000300040005000SE +/- 60.06, N = 12SE +/- 44.12, N = 12SE +/- 30.47, N = 3SE +/- 17.68, N = 342194441471048231. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: GPT-2 - Device: CPU - Executor: StandardADBC8001600240032004000Min: 3868 / Avg: 4219.25 / Max: 4428.5Min: 4188.5 / Avg: 4440.5 / Max: 4623Min: 4656 / Avg: 4709.67 / Max: 4761.5Min: 4788 / Avg: 4822.83 / Max: 4845.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of Intel oneAPI. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPUADBC0.21180.42360.63540.84721.059SE +/- 0.010786, N = 3SE +/- 0.009748, N = 15SE +/- 0.010535, N = 15SE +/- 0.009118, N = 30.9412660.9284040.9100560.904110MIN: 0.86MIN: 0.83MIN: 0.8MIN: 0.841. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPUADBC246810Min: 0.92 / Avg: 0.94 / Max: 0.96Min: 0.88 / Avg: 0.93 / Max: 1.01Min: 0.83 / Avg: 0.91 / Max: 0.98Min: 0.89 / Avg: 0.9 / Max: 0.921. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPUABCD30060090012001500SE +/- 1.48, N = 3SE +/- 15.69, N = 3SE +/- 2.45, N = 3SE +/- 9.49, N = 31260.201251.241221.121211.44MIN: 1234.63MIN: 1199.03MIN: 1196.31MIN: 1174.611. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPUABCD2004006008001000Min: 1257.24 / Avg: 1260.2 / Max: 1261.8Min: 1220.24 / Avg: 1251.24 / Max: 1270.93Min: 1216.93 / Avg: 1221.12 / Max: 1225.43Min: 1195.5 / Avg: 1211.44 / Max: 1228.341. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_1d - Data Type: f32 - Engine: CPUDCBA246810SE +/- 0.03878, N = 3SE +/- 0.03051, N = 3SE +/- 0.03741, N = 3SE +/- 0.04611, N = 36.906076.850396.821666.68005MIN: 6.2MIN: 6.18MIN: 6.16MIN: 5.951. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_1d - Data Type: f32 - Engine: CPUDCBA3691215Min: 6.84 / Avg: 6.91 / Max: 6.98Min: 6.81 / Avg: 6.85 / Max: 6.91Min: 6.76 / Avg: 6.82 / Max: 6.89Min: 6.61 / Avg: 6.68 / Max: 6.771. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Training - Data Type: bf16bf16bf16 - Engine: CPUACBD11002200330044005500SE +/- 3.11, N = 3SE +/- 10.61, N = 3SE +/- 9.21, N = 3SE +/- 80.48, N = 135028.065011.494997.824884.90MIN: 4972.23MIN: 4942.23MIN: 4933.96MIN: 4023.921. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Training - Data Type: bf16bf16bf16 - Engine: CPUACBD9001800270036004500Min: 5022.06 / Avg: 5028.06 / Max: 5032.5Min: 4991.3 / Avg: 5011.49 / Max: 5027.22Min: 4979.77 / Avg: 4997.82 / Max: 5009.98Min: 4041.91 / Avg: 4884.9 / Max: 5051.311. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 3D - Data Type: u8s8f32 - Engine: CPUACBD0.2560.5120.7681.0241.28SE +/- 0.00225, N = 3SE +/- 0.00210, N = 3SE +/- 0.00878, N = 9SE +/- 0.01354, N = 31.137741.119271.117741.10705MIN: 1.04MIN: 1.04MIN: 1MIN: 1.031. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 3D - Data Type: u8s8f32 - Engine: CPUACBD246810Min: 1.14 / Avg: 1.14 / Max: 1.14Min: 1.12 / Avg: 1.12 / Max: 1.12Min: 1.05 / Avg: 1.12 / Max: 1.14Min: 1.09 / Avg: 1.11 / Max: 1.131. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

ONNX Runtime

ONNX Runtime is developed by Microsoft and partners as a open-source, cross-platform, high performance machine learning inferencing and training accelerator. This test profile runs the ONNX Runtime with various models available from the ONNX Zoo. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: ArcFace ResNet-100 - Device: CPU - Executor: StandardDABC2004006008001000SE +/- 7.44, N = 3SE +/- 5.53, N = 3SE +/- 5.18, N = 3SE +/- 3.28, N = 3991995101010171. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: ArcFace ResNet-100 - Device: CPU - Executor: StandardDABC2004006008001000Min: 977.5 / Avg: 991.33 / Max: 1003Min: 984 / Avg: 995 / Max: 1001.5Min: 1003.5 / Avg: 1010.33 / Max: 1020.5Min: 1010.5 / Avg: 1016.83 / Max: 1021.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: fcn-resnet101-11 - Device: CPU - Executor: StandardACBD306090120150SE +/- 0.33, N = 3SE +/- 0.73, N = 3SE +/- 0.44, N = 3SE +/- 0.44, N = 31531531561571. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: fcn-resnet101-11 - Device: CPU - Executor: StandardACBD306090120150Min: 153 / Avg: 153.33 / Max: 154Min: 152 / Avg: 153.33 / Max: 154.5Min: 155 / Avg: 155.83 / Max: 156.5Min: 156 / Avg: 156.67 / Max: 157.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: bertsquad-12 - Device: CPU - Executor: ParallelDABC90180270360450SE +/- 0.93, N = 3SE +/- 2.93, N = 3SE +/- 1.44, N = 3SE +/- 1.80, N = 34214244254321. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: bertsquad-12 - Device: CPU - Executor: ParallelDABC80160240320400Min: 419.5 / Avg: 421.33 / Max: 422.5Min: 418.5 / Avg: 424 / Max: 428.5Min: 422.5 / Avg: 425 / Max: 427.5Min: 428 / Avg: 431.5 / Max: 4341. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: fcn-resnet101-11 - Device: CPU - Executor: ParallelBCDA20406080100SE +/- 0.17, N = 3SE +/- 0.17, N = 3SE +/- 0.17, N = 3SE +/- 0.17, N = 3808181821. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: fcn-resnet101-11 - Device: CPU - Executor: ParallelBCDA1632486480Min: 80 / Avg: 80.33 / Max: 80.5Min: 80.5 / Avg: 80.83 / Max: 81Min: 80.5 / Avg: 80.83 / Max: 81Min: 81.5 / Avg: 81.67 / Max: 821. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of Intel oneAPI. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPUBCAD11002200330044005500SE +/- 8.67, N = 3SE +/- 48.98, N = 3SE +/- 28.21, N = 3SE +/- 59.27, N = 155003.994964.594959.964882.28MIN: 4937.34MIN: 4820.84MIN: 4866.16MIN: 4219.231. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPUBCAD9001800270036004500Min: 4987.66 / Avg: 5003.99 / Max: 5017.18Min: 4867.03 / Avg: 4964.59 / Max: 5020.96Min: 4912.96 / Avg: 4959.96 / Max: 5010.48Min: 4237.72 / Avg: 4882.28 / Max: 5065.061. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

ONNX Runtime

ONNX Runtime is developed by Microsoft and partners as a open-source, cross-platform, high performance machine learning inferencing and training accelerator. This test profile runs the ONNX Runtime with various models available from the ONNX Zoo. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: yolov4 - Device: CPU - Executor: StandardABCD70140210280350SE +/- 3.18, N = 4SE +/- 1.42, N = 3SE +/- 1.64, N = 3SE +/- 1.04, N = 32932932953001. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: yolov4 - Device: CPU - Executor: StandardABCD50100150200250Min: 283.5 / Avg: 293 / Max: 297Min: 290 / Avg: 292.83 / Max: 294.5Min: 293 / Avg: 295.33 / Max: 298.5Min: 298 / Avg: 300 / Max: 301.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: super-resolution-10 - Device: CPU - Executor: ParallelDBCA8001600240032004000SE +/- 42.45, N = 4SE +/- 19.55, N = 3SE +/- 26.17, N = 3SE +/- 34.71, N = 337313780378438151. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: super-resolution-10 - Device: CPU - Executor: ParallelDBCA7001400210028003500Min: 3654 / Avg: 3731.25 / Max: 3848Min: 3753 / Avg: 3780 / Max: 3818Min: 3757.5 / Avg: 3783.67 / Max: 3836Min: 3771 / Avg: 3815 / Max: 3883.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: GPT-2 - Device: CPU - Executor: ParallelADBC8001600240032004000SE +/- 7.49, N = 3SE +/- 4.21, N = 3SE +/- 0.44, N = 3SE +/- 6.29, N = 334613495351235291. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: GPT-2 - Device: CPU - Executor: ParallelADBC6001200180024003000Min: 3452 / Avg: 3461.17 / Max: 3476Min: 3487 / Avg: 3495.33 / Max: 3500.5Min: 3511.5 / Avg: 3512.17 / Max: 3513Min: 3516.5 / Avg: 3528.67 / Max: 3537.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of Intel oneAPI. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Inference - Data Type: u8s8f32 - Engine: CPUBCAD30060090012001500SE +/- 10.61, N = 15SE +/- 2.90, N = 3SE +/- 9.32, N = 3SE +/- 5.02, N = 31246.071238.601236.751223.53MIN: 1122.63MIN: 1199.98MIN: 1201.87MIN: 1198.191. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Inference - Data Type: u8s8f32 - Engine: CPUBCAD2004006008001000Min: 1142.49 / Avg: 1246.07 / Max: 1298.02Min: 1234.9 / Avg: 1238.6 / Max: 1244.31Min: 1221.4 / Avg: 1236.75 / Max: 1253.59Min: 1218.21 / Avg: 1223.53 / Max: 1233.561. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Training - Data Type: u8s8f32 - Engine: CPUBCAD11002200330044005500SE +/- 24.43, N = 3SE +/- 9.55, N = 3SE +/- 40.36, N = 9SE +/- 19.43, N = 35034.835003.384954.344950.97MIN: 4959.6MIN: 4934.99MIN: 4613.84MIN: 4866.671. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Training - Data Type: u8s8f32 - Engine: CPUBCAD9001800270036004500Min: 5002.87 / Avg: 5034.83 / Max: 5082.81Min: 4984.65 / Avg: 5003.38 / Max: 5015.97Min: 4653.29 / Avg: 4954.34 / Max: 5040.04Min: 4913.29 / Avg: 4950.97 / Max: 4978.071. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

ONNX Runtime

ONNX Runtime is developed by Microsoft and partners as a open-source, cross-platform, high performance machine learning inferencing and training accelerator. This test profile runs the ONNX Runtime with various models available from the ONNX Zoo. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: ArcFace ResNet-100 - Device: CPU - Executor: ParallelBCDA2004006008001000SE +/- 4.36, N = 3SE +/- 7.42, N = 3SE +/- 2.33, N = 3SE +/- 5.11, N = 310721079107910881. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: ArcFace ResNet-100 - Device: CPU - Executor: ParallelBCDA2004006008001000Min: 1064.5 / Avg: 1071.5 / Max: 1079.5Min: 1071 / Avg: 1078.67 / Max: 1093.5Min: 1075 / Avg: 1078.67 / Max: 1083Min: 1078 / Avg: 1087.5 / Max: 1095.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of Intel oneAPI. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_3d - Data Type: u8s8f32 - Engine: CPUBCDA0.22340.44680.67020.89361.117SE +/- 0.001082, N = 3SE +/- 0.004648, N = 3SE +/- 0.000768, N = 3SE +/- 0.001712, N = 30.9927130.9870200.9848190.979135MIN: 0.93MIN: 0.91MIN: 0.92MIN: 0.931. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_3d - Data Type: u8s8f32 - Engine: CPUBCDA246810Min: 0.99 / Avg: 0.99 / Max: 0.99Min: 0.98 / Avg: 0.99 / Max: 0.99Min: 0.98 / Avg: 0.98 / Max: 0.99Min: 0.98 / Avg: 0.98 / Max: 0.981. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Convolution Batch Shapes Auto - Data Type: u8s8f32 - Engine: CPUCDBA246810SE +/- 0.00816, N = 3SE +/- 0.01318, N = 3SE +/- 0.02160, N = 3SE +/- 0.01609, N = 36.446896.443306.433306.39430MIN: 6.34MIN: 6.36MIN: 6.33MIN: 6.311. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Convolution Batch Shapes Auto - Data Type: u8s8f32 - Engine: CPUCDBA3691215Min: 6.44 / Avg: 6.45 / Max: 6.46Min: 6.43 / Avg: 6.44 / Max: 6.47Min: 6.39 / Avg: 6.43 / Max: 6.47Min: 6.37 / Avg: 6.39 / Max: 6.421. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_3d - Data Type: f32 - Engine: CPUCDBA0.47520.95041.42561.90082.376SE +/- 0.00512, N = 3SE +/- 0.00569, N = 3SE +/- 0.00652, N = 3SE +/- 0.00321, N = 32.112122.111462.110252.10617MIN: 2.06MIN: 2.06MIN: 2.05MIN: 2.051. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_3d - Data Type: f32 - Engine: CPUCDBA246810Min: 2.11 / Avg: 2.11 / Max: 2.12Min: 2.1 / Avg: 2.11 / Max: 2.12Min: 2.1 / Avg: 2.11 / Max: 2.12Min: 2.1 / Avg: 2.11 / Max: 2.111. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

ONNX Runtime

ONNX Runtime is developed by Microsoft and partners as a open-source, cross-platform, high performance machine learning inferencing and training accelerator. This test profile runs the ONNX Runtime with various models available from the ONNX Zoo. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: yolov4 - Device: CPU - Executor: ParallelADBC80160240320400SE +/- 0.50, N = 3SE +/- 0.17, N = 3SE +/- 0.29, N = 3SE +/- 0.33, N = 33613613623621. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: yolov4 - Device: CPU - Executor: ParallelADBC60120180240300Min: 360 / Avg: 360.5 / Max: 361.5Min: 361 / Avg: 361.33 / Max: 361.5Min: 361 / Avg: 361.5 / Max: 362Min: 362 / Avg: 362.33 / Max: 3631. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: super-resolution-10 - Device: CPU - Executor: StandardBADC16003200480064008000SE +/- 409.51, N = 12SE +/- 58.03, N = 3SE +/- 46.97, N = 3SE +/- 47.29, N = 364017323737575601. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt
OpenBenchmarking.orgInferences Per Minute, More Is BetterONNX Runtime 1.11Model: super-resolution-10 - Device: CPU - Executor: StandardBADC13002600390052006500Min: 4695.5 / Avg: 6401.25 / Max: 7610Min: 7214 / Avg: 7323 / Max: 7412Min: 7291 / Avg: 7374.67 / Max: 7453.5Min: 7475 / Avg: 7559.67 / Max: 7638.51. (CXX) g++ options: -ffunction-sections -fdata-sections -march=native -mtune=native -O3 -flto -fno-fat-lto-objects -ldl -lrt

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of Intel oneAPI. Learn more via the OpenBenchmarking.org test page.

Harness: Matrix Multiply Batch Shapes Transformer - Data Type: bf16bf16bf16 - Engine: CPU

A: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

B: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

C: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

D: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPUCDAB3691215SE +/- 0.32, N = 15SE +/- 0.39, N = 12SE +/- 0.23, N = 15SE +/- 0.07, N = 311.9011.7511.6011.34MIN: 8.22MIN: 8.18MIN: 9.98MIN: 10.781. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPUCDAB3691215Min: 9.3 / Avg: 11.9 / Max: 13.59Min: 8.54 / Avg: 11.75 / Max: 12.91Min: 10.29 / Avg: 11.6 / Max: 13.75Min: 11.2 / Avg: 11.34 / Max: 11.421. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Inference - Data Type: bf16bf16bf16 - Engine: CPUDCBA30060090012001500SE +/- 10.63, N = 3SE +/- 14.38, N = 3SE +/- 8.78, N = 3SE +/- 37.60, N = 121254.391250.991242.441208.52MIN: 1209.18MIN: 1202.91MIN: 1204.89MIN: 796.321. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Recurrent Neural Network Inference - Data Type: bf16bf16bf16 - Engine: CPUDCBA2004006008001000Min: 1233.91 / Avg: 1254.39 / Max: 1269.58Min: 1224.32 / Avg: 1250.99 / Max: 1273.65Min: 1226.55 / Avg: 1242.44 / Max: 1256.87Min: 806.68 / Avg: 1208.52 / Max: 1299.111. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPUACDB246810SE +/- 0.31854, N = 15SE +/- 0.41090, N = 15SE +/- 0.36772, N = 12SE +/- 0.38999, N = 157.591657.579417.049816.93013MIN: 5.06MIN: 4.43MIN: 4.64MIN: 4.811. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPUACDB3691215Min: 5.29 / Avg: 7.59 / Max: 9.2Min: 4.6 / Avg: 7.58 / Max: 9.39Min: 4.85 / Avg: 7.05 / Max: 8.72Min: 4.99 / Avg: 6.93 / Max: 9.41. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

Harness: Deconvolution Batch shapes_3d - Data Type: bf16bf16bf16 - Engine: CPU

A: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

B: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

C: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

D: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

Harness: Deconvolution Batch shapes_1d - Data Type: bf16bf16bf16 - Engine: CPU

A: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

B: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

C: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

D: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

Harness: Convolution Batch Shapes Auto - Data Type: bf16bf16bf16 - Engine: CPU

A: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

B: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

C: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

D: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_1d - Data Type: u8s8f32 - Engine: CPUCABD0.3510.7021.0531.4041.755SE +/- 0.00131, N = 3SE +/- 0.00789, N = 3SE +/- 0.04182, N = 12SE +/- 0.02072, N = 31.560001.526191.498711.45511MIN: 1.41MIN: 1.38MIN: 0.93MIN: 1.31. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: Deconvolution Batch shapes_1d - Data Type: u8s8f32 - Engine: CPUCABD246810Min: 1.56 / Avg: 1.56 / Max: 1.56Min: 1.51 / Avg: 1.53 / Max: 1.54Min: 1.04 / Avg: 1.5 / Max: 1.55Min: 1.42 / Avg: 1.46 / Max: 1.491. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

Harness: IP Shapes 3D - Data Type: bf16bf16bf16 - Engine: CPU

A: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

B: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

C: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

D: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

Harness: IP Shapes 1D - Data Type: bf16bf16bf16 - Engine: CPU

A: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

B: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

C: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

D: The test run did not produce a result. The test run did not produce a result. The test run did not produce a result.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 1D - Data Type: u8s8f32 - Engine: CPUDCBA0.54491.08981.63472.17962.7245SE +/- 0.08567, N = 15SE +/- 0.10516, N = 12SE +/- 0.08755, N = 15SE +/- 0.09135, N = 122.421762.374032.351612.18433MIN: 1.53MIN: 1.6MIN: 1.4MIN: 1.281. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 1D - Data Type: u8s8f32 - Engine: CPUDCBA246810Min: 1.7 / Avg: 2.42 / Max: 2.94Min: 1.77 / Avg: 2.37 / Max: 2.95Min: 2 / Avg: 2.35 / Max: 2.93Min: 1.46 / Avg: 2.18 / Max: 2.711. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 1D - Data Type: f32 - Engine: CPUACBD0.45210.90421.35631.80842.2605SE +/- 0.02176, N = 15SE +/- 0.05487, N = 15SE +/- 0.06425, N = 12SE +/- 0.06721, N = 152.009531.993831.964201.91681MIN: 1.59MIN: 1.39MIN: 1.46MIN: 1.251. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 2.6Harness: IP Shapes 1D - Data Type: f32 - Engine: CPUACBD246810Min: 1.86 / Avg: 2.01 / Max: 2.15Min: 1.57 / Avg: 1.99 / Max: 2.28Min: 1.66 / Avg: 1.96 / Max: 2.24Min: 1.34 / Avg: 1.92 / Max: 2.251. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -std=c++11 -pie -ldl -lpthread

A

Kernel Notes: Transparent Huge Pages: madvise
Compiler 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,brig,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-targets=nvptx-none=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/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 -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0x8301039
Python Notes: Python 3.9.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 31 March 2022 05:03 by user pts.

B

Kernel Notes: Transparent Huge Pages: madvise
Compiler 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,brig,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-targets=nvptx-none=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/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 -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0x8301039
Python Notes: Python 3.9.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 31 March 2022 08:24 by user pts.

C

Kernel Notes: Transparent Huge Pages: madvise
Compiler 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,brig,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-targets=nvptx-none=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/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 -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0x8301039
Python Notes: Python 3.9.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 31 March 2022 11:13 by user pts.

D

Processor: AMD Ryzen Threadripper 3990X 64-Core @ 2.90GHz (64 Cores / 128 Threads), Motherboard: Gigabyte TRX40 AORUS PRO WIFI (F4p BIOS), Chipset: AMD Starship/Matisse, Memory: 128GB, Disk: Samsung SSD 970 EVO Plus 500GB, Graphics: AMD Radeon RX 5700 8GB (1750/875MHz), Audio: AMD Navi 10 HDMI Audio, Monitor: DELL P2415Q, Network: Intel I211 + Intel Wi-Fi 6 AX200

OS: Pop 21.10, Kernel: 5.17.0-rc1-sched-core-phx (x86_64), Desktop: GNOME Shell 40.5, Display Server: X Server, OpenGL: 4.6 Mesa 21.2.2 (LLVM 12.0.1), Vulkan: 1.2.182, Compiler: GCC 11.2.0, File-System: ext4, Screen Resolution: 3840x2160

Kernel Notes: Transparent Huge Pages: madvise
Compiler 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,brig,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-targets=nvptx-none=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-ZPT0kp/gcc-11-11.2.0/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 -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0x8301039
Python Notes: Python 3.9.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 31 March 2022 13:27 by user pts.