phoronix-machine-learning.txt

AMD Ryzen Threadripper 7960X 24-Cores testing with a Gigabyte TRX50 AERO D (FA BIOS) and Sapphire AMD Radeon RX 7900 XTX 24GB on Ubuntu 24.04 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 2411137-NE-PHORONIXM28
Jump To Table - Results

Statistics

Remove Outliers Before Calculating Averages

Graph Settings

Prefer Vertical Bar Graphs

Multi-Way Comparison

Condense Multi-Option Tests Into Single Result Graphs

Table

Show Detailed System Result Table

Run Management

Result
Identifier
Performance Per
Dollar
Date
Run
  Test
  Duration
phoronix-ml.txt
November 10
  3 Days, 15 Hours, 7 Minutes
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):


phoronix-machine-learning.txtOpenBenchmarking.orgPhoronix Test SuiteAMD Ryzen Threadripper 7960X 24-Cores @ 7.79GHz (24 Cores / 48 Threads)Gigabyte TRX50 AERO D (FA BIOS)AMD Device 14a44 x 32GB DDR5-5200MT/s Micron MTC20F1045S1RC56BG11000GB GIGABYTE AG512K1TBSapphire AMD Radeon RX 7900 XTX 24GBAMD Device 14ccHP E273Aquantia AQC113C NBase-T/IEEE + Realtek RTL8125 2.5GbE + Qualcomm WCN785x Wi-Fi 7Ubuntu 24.046.8.0-48-generic (x86_64)GNOME Shell 46.0X Server + Wayland4.6 Mesa 24.2.0-devel (LLVM 18.1.7 DRM 3.58)OpenCL 2.1 AMD-APP (3625.0)GCC 13.2.0ext41920x1080ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerOpenGLOpenCLCompilerFile-SystemScreen ResolutionPhoronix-machine-learning.txt BenchmarksSystem Logs- Transparent Huge Pages: madvise- --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-backtrace --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-uJ7kn6/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-uJ7kn6/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v - Scaling Governor: amd-pstate-epp powersave (EPP: balance_performance) - CPU Microcode: 0xa108105- BAR1 / Visible vRAM Size: 24560 MB- Python 3.12.3- gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: Not affected + retbleed: Not affected + spec_rstack_overflow: 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; BHI: Not affected + srbds: Not affected + tsx_async_abort: Not affected

phoronix-machine-learning.txttensorflow: GPU - 512 - VGG-16tensorflow: GPU - 256 - VGG-16tensorflow: GPU - 512 - ResNet-50scikit-learn: Isotonic / Pathologicaltensorflow: GPU - 256 - ResNet-50tensorflow: GPU - 64 - VGG-16scikit-learn: Isotonic / Perturbed Logarithmtensorflow: GPU - 512 - GoogLeNetscikit-learn: Isotonic / Logistictensorflow: CPU - 512 - VGG-16tensorflow: CPU - 256 - ResNet-50tensorflow: GPU - 32 - VGG-16tensorflow: GPU - 512 - AlexNetlczero: BLASscikit-learn: SAGAtensorflow: GPU - 256 - GoogLeNettensorflow: CPU - 512 - ResNet-50tensorflow: CPU - 256 - VGG-16pytorch: CPU - 512 - Efficientnet_v2_ltensorflow: GPU - 64 - ResNet-50tensorflow: CPU - 512 - GoogLeNettensorflow: GPU - 16 - VGG-16scikit-learn: Sparse Rand Projections / 100 Iterationsscikit-learn: Hist Gradient Boosting Adultwhisper-cpp: ggml-medium.en - 2016 State of the Uniontensorflow: GPU - 256 - AlexNetscikit-learn: Plot Parallel Pairwisescikit-learn: Hist Gradient Boosting Higgs Bosonncnn: CPU - FastestDetncnn: CPU - vision_transformerncnn: CPU - regnety_400mncnn: CPU - squeezenet_ssdncnn: CPU - yolov4-tinyncnn: CPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3ncnn: CPU - resnet50ncnn: CPU - alexnetncnn: CPU - resnet18ncnn: CPU - vgg16ncnn: CPU - googlenetncnn: CPU - blazefacencnn: CPU - efficientnet-b0ncnn: CPU - mnasnetncnn: CPU - shufflenet-v2ncnn: CPU-v3-v3 - mobilenet-v3ncnn: CPU-v2-v2 - mobilenet-v2ncnn: CPU - mobilenetscikit-learn: Covertype Dataset Benchmarkscikit-learn: Lassotensorflow: GPU - 32 - ResNet-50scikit-learn: SGDOneClassSVMscikit-learn: TSNE MNIST Datasetopenvino: Noise Suppression Poconet-Like FP16 - CPUopenvino: Noise Suppression Poconet-Like FP16 - CPUopenvino: Person Detection FP16 - CPUopenvino: Person Detection FP16 - CPUopenvino: Person Detection FP32 - CPUopenvino: Person Detection FP32 - CPUtensorflow-lite: Inception V4tensorflow-lite: NASNet Mobiletensorflow-lite: SqueezeNetopenvino: Road Segmentation ADAS FP16 - CPUopenvino: Road Segmentation ADAS FP16 - CPUopenvino: Vehicle Detection FP16 - CPUopenvino: Vehicle Detection FP16 - CPUonnx: CaffeNet 12-int8 - CPU - Parallelonnx: CaffeNet 12-int8 - CPU - Parallelscikit-learn: Isolation Foresttensorflow: GPU - 64 - GoogLeNetonnx: fcn-resnet101-11 - CPU - Parallelonnx: fcn-resnet101-11 - CPU - Paralleltensorflow: CPU - 64 - VGG-16openvino: Machine Translation EN To DE FP16 - CPUopenvino: Machine Translation EN To DE FP16 - CPUscikit-learn: GLMscikit-learn: Hist Gradient Boostingwhisper-cpp: ggml-small.en - 2016 State of the Uniontensorflow: GPU - 16 - ResNet-50pytorch: CPU - 32 - Efficientnet_v2_lpytorch: CPU - 16 - Efficientnet_v2_lpytorch: CPU - 64 - Efficientnet_v2_lmnn: inception-v3mnn: mobilenet-v1-1.0mnn: MobileNetV2_224mnn: SqueezeNetV1.0mnn: resnet-v2-50mnn: squeezenetv1.1mnn: mobilenetV3mnn: nasnetpytorch: CPU - 256 - Efficientnet_v2_lscikit-learn: Plot Hierarchicalxnnpack: QS8MobileNetV2xnnpack: FP16MobileNetV3Smallxnnpack: FP16MobileNetV3Largexnnpack: FP16MobileNetV2xnnpack: FP16MobileNetV1xnnpack: FP32MobileNetV3Smallxnnpack: FP32MobileNetV3Largexnnpack: FP32MobileNetV2xnnpack: FP32MobileNetV1shoc: OpenCL - S3Dopencv: DNN - Deep Neural Networkscikit-learn: Hist Gradient Boosting Categorical Onlyscikit-learn: Plot Neighborstensorflow: GPU - 64 - AlexNetscikit-learn: Sparsifyscikit-learn: Plot Polynomial Kernel Approximationscikit-learn: Feature Expansionstensorflow: GPU - 32 - GoogLeNettensorflow: CPU - 256 - GoogLeNetscikit-learn: Plot Wardtensorflow: CPU - 32 - VGG-16scikit-learn: Sample Without Replacementpytorch: CPU - 64 - ResNet-152pytorch: CPU - 256 - ResNet-152pytorch: CPU - 512 - ResNet-152pytorch: CPU - 32 - ResNet-152pytorch: CPU - 16 - ResNet-152numpy: tensorflow: CPU - 64 - ResNet-50whisper-cpp: ggml-base.en - 2016 State of the Unionscikit-learn: Treetensorflow: CPU - 512 - AlexNetncnn: Vulkan GPU - FastestDetncnn: Vulkan GPU - vision_transformerncnn: Vulkan GPU - regnety_400mncnn: Vulkan GPU - squeezenet_ssdncnn: Vulkan GPU - yolov4-tinyncnn: Vulkan GPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3ncnn: Vulkan GPU - resnet50ncnn: Vulkan GPU - alexnetncnn: Vulkan GPU - resnet18ncnn: Vulkan GPU - vgg16ncnn: Vulkan GPU - googlenetncnn: Vulkan GPU - blazefacencnn: Vulkan GPU - efficientnet-b0ncnn: Vulkan GPU - mnasnetncnn: Vulkan GPU - shufflenet-v2ncnn: Vulkan GPU-v3-v3 - mobilenet-v3ncnn: Vulkan GPU-v2-v2 - mobilenet-v2ncnn: Vulkan GPU - mobilenetscikit-learn: Hist Gradient Boosting Threadingscikit-learn: SGD Regressionscikit-learn: Kernel PCA Solvers / Time vs. N Samplespytorch: CPU - 1 - Efficientnet_v2_lonnx: ZFNet-512 - CPU - Parallelonnx: ZFNet-512 - CPU - Parallelonnx: ZFNet-512 - CPU - Standardonnx: ZFNet-512 - CPU - Standardonnx: T5 Encoder - CPU - Standardonnx: T5 Encoder - CPU - Standardtensorflow: GPU - 32 - AlexNetonednn: Recurrent Neural Network Training - CPUonednn: IP Shapes 1D - CPUshoc: OpenCL - Max SP Flopsonednn: Recurrent Neural Network Inference - CPUscikit-learn: MNIST Datasettensorflow: GPU - 16 - GoogLeNettensorflow: CPU - 16 - VGG-16scikit-learn: Plot Incremental PCAscikit-learn: Text Vectorizersopenvino: Face Detection FP16 - CPUopenvino: Face Detection FP16 - CPUopenvino: Face Detection FP16-INT8 - CPUopenvino: Face Detection FP16-INT8 - CPUonnx: ResNet101_DUC_HDC-12 - CPU - Parallelonnx: ResNet101_DUC_HDC-12 - CPU - Parallelonnx: ResNet101_DUC_HDC-12 - CPU - Standardonnx: ResNet101_DUC_HDC-12 - CPU - Standardonnx: fcn-resnet101-11 - CPU - Standardonnx: fcn-resnet101-11 - CPU - Standardonnx: yolov4 - CPU - Parallelonnx: yolov4 - CPU - Parallelonnx: T5 Encoder - CPU - Parallelonnx: T5 Encoder - CPU - Parallelonnx: yolov4 - CPU - Standardonnx: yolov4 - CPU - Standardopenvino: Road Segmentation ADAS FP16-INT8 - CPUopenvino: Road Segmentation ADAS FP16-INT8 - CPUopenvino: Person Vehicle Bike Detection FP16 - CPUopenvino: Person Vehicle Bike Detection FP16 - CPUtensorflow-lite: Inception ResNet V2tensorflow-lite: Mobilenet Floattensorflow-lite: Mobilenet Quantopenvino: Person Re-Identification Retail FP16 - CPUopenvino: Person Re-Identification Retail FP16 - CPUopenvino: Face Detection Retail FP16-INT8 - CPUopenvino: Face Detection Retail FP16-INT8 - CPUopenvino: Handwritten English Recognition FP16-INT8 - CPUopenvino: Handwritten English Recognition FP16-INT8 - CPUopenvino: Age Gender Recognition Retail 0013 FP16-INT8 - CPUopenvino: Age Gender Recognition Retail 0013 FP16-INT8 - CPUopenvino: Vehicle Detection FP16-INT8 - CPUopenvino: Vehicle Detection FP16-INT8 - CPUopenvino: Handwritten English Recognition FP16 - CPUopenvino: Handwritten English Recognition FP16 - CPUopenvino: Age Gender Recognition Retail 0013 FP16 - CPUopenvino: Age Gender Recognition Retail 0013 FP16 - CPUopenvino: Weld Porosity Detection FP16 - CPUopenvino: Weld Porosity Detection FP16 - CPUopenvino: Weld Porosity Detection FP16-INT8 - CPUopenvino: Weld Porosity Detection FP16-INT8 - CPUopenvino: Face Detection Retail FP16 - CPUopenvino: Face Detection Retail FP16 - CPUonnx: CaffeNet 12-int8 - CPU - Standardonnx: CaffeNet 12-int8 - CPU - Standardonnx: ResNet50 v1-12-int8 - CPU - Parallelonnx: ResNet50 v1-12-int8 - CPU - Parallelonnx: ResNet50 v1-12-int8 - CPU - Standardonnx: ResNet50 v1-12-int8 - CPU - Standardonnx: super-resolution-10 - CPU - Parallelonnx: super-resolution-10 - CPU - Parallelonnx: super-resolution-10 - CPU - Standardonnx: super-resolution-10 - CPU - Standardtensorflow: CPU - 32 - ResNet-50scikit-learn: Plot OMP vs. LARStensorflow: GPU - 1 - VGG-16pytorch: CPU - 1 - ResNet-152tensorflow: GPU - 1 - AlexNettensorflow: CPU - 256 - AlexNetonednn: IP Shapes 3D - CPUpytorch: CPU - 16 - ResNet-50pytorch: CPU - 512 - ResNet-50pytorch: CPU - 256 - ResNet-50pytorch: CPU - 32 - ResNet-50pytorch: CPU - 64 - ResNet-50tensorflow: GPU - 16 - AlexNetscikit-learn: Kernel PCA Solvers / Time vs. N Componentsdeepspeech: CPUtensorflow: CPU - 64 - GoogLeNetscikit-learn: LocalOutlierFactortensorflow: CPU - 16 - ResNet-50onednn: Deconvolution Batch shapes_1d - CPUpytorch: CPU - 1 - ResNet-50tensorflow: GPU - 1 - ResNet-50rbenchmark: tensorflow: CPU - 32 - GoogLeNetscikit-learn: 20 Newsgroups / Logistic Regressiontensorflow: CPU - 64 - AlexNettensorflow: CPU - 1 - VGG-16tensorflow: CPU - 16 - GoogLeNettensorflow: CPU - 32 - AlexNettensorflow: CPU - 1 - ResNet-50onednn: Convolution Batch Shapes Auto - CPUrnnoise: 26 Minute Long Talking Sampletensorflow: CPU - 16 - AlexNettensorflow: GPU - 1 - GoogLeNetshoc: OpenCL - Texture Read Bandwidthtensorflow: CPU - 1 - AlexNettensorflow: CPU - 1 - GoogLeNetonednn: Deconvolution Batch shapes_3d - CPUshoc: OpenCL - Triadshoc: OpenCL - GEMM SGEMM_Nshoc: OpenCL - Bus Speed Downloadshoc: OpenCL - Bus Speed Readbackshoc: OpenCL - Reductionshoc: OpenCL - FFT SPshoc: OpenCL - MD5 Hashdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Asynchronous Multi-Streamphoronix-ml.txt2.552.559.343843.2589.362.551528.96628.731406.45230.4359.702.5449.28184669.11829.0358.5030.169.949.24185.252.51504.829153.338579.1707748.92167.99765.7369.8040.5918.5814.3424.1413.8513.335.568.0225.7116.423.118.286.118.156.456.3013.85320.394308.2259.14233.407247.55611.532052.0293.29129.45100.46119.4120372.633662.51836.2825.83465.8611.121077.874.92768203.059176.28728.53834.5721.2033929.0563.99187.88168.333166.776218.185238.949.859.909.9836.4523.7843.2686.42918.5344.3272.53615.29710.11141.463139814642128149511441503246518731233289.5433308030.188114.83947.85108.455104.700100.35327.91227.0642.10428.3290.64017.6417.9217.9917.7817.97715.5070.8992.7526146.970643.449.8241.0518.6316.0423.6413.7914.655.287.8625.1316.013.148.685.998.066.496.3113.7952.72964.36861.61114.1817.287557.86129.10376109.8635.81489172.03746.141261.401.1365793757.3736.40052.73626.9127.3431.20945.340607.4819.69320.4237.36770.9061.29718452.3142.21086242.9084.11676185.3605.396843.67307272.184103.5799.6544217.62679.666.181930.0833356.31381.252501.214.722523.593.586458.3821.581108.650.367537.775.512160.5723.121035.290.4348433.9212.251947.676.313742.772.764273.091.56253639.9059.18292108.8753.04551328.2938.08149123.73610.275497.317767.9441.4762.2023.1915.38627.501.3959145.5945.7546.0646.4246.6942.4331.03746.23475225.1621.61662.293.7756760.176.690.1252218.0710.450516.189.70198.11409.5618.382.363177.852288.7121.031003.32130.6660.921.8520613.81587615.3524.989326.252542.9449752.83746.5084OpenBenchmarking.org

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 512 - Model: VGG-16phoronix-ml.txt0.57381.14761.72142.29522.869SE +/- 0.00, N = 32.55

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 256 - Model: VGG-16phoronix-ml.txt0.57381.14761.72142.29522.869SE +/- 0.01, N = 32.55

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 512 - Model: ResNet-50phoronix-ml.txt3691215SE +/- 0.03, N = 39.34

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Isotonic / Pathologicalphoronix-ml.txt8001600240032004000SE +/- 9.06, N = 33843.261. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 256 - Model: ResNet-50phoronix-ml.txt3691215SE +/- 0.01, N = 39.36

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 64 - Model: VGG-16phoronix-ml.txt0.57381.14761.72142.29522.869SE +/- 0.00, N = 32.55

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Isotonic / Perturbed Logarithmphoronix-ml.txt30060090012001500SE +/- 2.36, N = 31528.971. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 512 - Model: GoogLeNetphoronix-ml.txt714212835SE +/- 0.10, N = 328.73

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Isotonic / Logisticphoronix-ml.txt30060090012001500SE +/- 0.82, N = 31406.451. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 512 - Model: VGG-16phoronix-ml.txt714212835SE +/- 0.01, N = 330.43

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: ResNet-50phoronix-ml.txt1326395265SE +/- 0.91, N = 959.70

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 32 - Model: VGG-16phoronix-ml.txt0.57151.1431.71452.2862.8575SE +/- 0.00, N = 32.54

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 512 - Model: AlexNetphoronix-ml.txt1122334455SE +/- 0.09, N = 349.28

LeelaChessZero

OpenBenchmarking.orgNodes Per Second, More Is BetterLeelaChessZero 0.31.1Backend: BLASphoronix-ml.txt4080120160200SE +/- 12.39, N = 91841. (CXX) g++ options: -flto -pthread

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: SAGAphoronix-ml.txt140280420560700SE +/- 3.66, N = 3669.121. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 256 - Model: GoogLeNetphoronix-ml.txt714212835SE +/- 0.01, N = 329.03

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 512 - Model: ResNet-50phoronix-ml.txt1326395265SE +/- 0.33, N = 358.50

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: VGG-16phoronix-ml.txt714212835SE +/- 0.08, N = 330.16

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 512 - Model: Efficientnet_v2_lphoronix-ml.txt3691215SE +/- 0.09, N = 129.94MIN: 7.81 / MAX: 10.45

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 64 - Model: ResNet-50phoronix-ml.txt3691215SE +/- 0.00, N = 39.24

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 512 - Model: GoogLeNetphoronix-ml.txt4080120160200SE +/- 1.69, N = 7185.25

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 16 - Model: VGG-16phoronix-ml.txt0.56481.12961.69442.25922.824SE +/- 0.00, N = 32.51

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Sparse Random Projections / 100 Iterationsphoronix-ml.txt110220330440550SE +/- 2.62, N = 3504.831. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Hist Gradient Boosting Adultphoronix-ml.txt306090120150SE +/- 1.23, N = 12153.341. (F9X) gfortran options: -O0

Whisper.cpp

OpenBenchmarking.orgSeconds, Fewer Is BetterWhisper.cpp 1.6.2Model: ggml-medium.en - Input: 2016 State of the Unionphoronix-ml.txt130260390520650SE +/- 1.41, N = 3579.171. (CXX) g++ options: -O3 -std=c++11 -fPIC -pthread -msse3 -mssse3 -mavx -mf16c -mfma -mavx2 -mavx512f -mavx512cd -mavx512vl -mavx512dq -mavx512bw -mavx512vbmi -mavx512vnni

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 256 - Model: AlexNetphoronix-ml.txt1122334455SE +/- 0.03, N = 348.92

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Plot Parallel Pairwisephoronix-ml.txt4080120160200SE +/- 4.47, N = 9168.001. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Hist Gradient Boosting Higgs Bosonphoronix-ml.txt1530456075SE +/- 0.83, N = 365.741. (F9X) gfortran options: -O0

NCNN

NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: FastestDetphoronix-ml.txt3691215SE +/- 0.30, N = 159.80MIN: 6.73 / MAX: 273.491. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: vision_transformerphoronix-ml.txt918273645SE +/- 0.18, N = 1540.59MIN: 37.83 / MAX: 299.431. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: regnety_400mphoronix-ml.txt510152025SE +/- 0.12, N = 1518.58MIN: 17.32 / MAX: 295.211. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: squeezenet_ssdphoronix-ml.txt48121620SE +/- 0.12, N = 1514.34MIN: 13.13 / MAX: 263.271. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: yolov4-tinyphoronix-ml.txt612182430SE +/- 0.11, N = 1524.14MIN: 21.58 / MAX: 105.121. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPUv2-yolov3v2-yolov3 - Model: mobilenetv2-yolov3phoronix-ml.txt48121620SE +/- 0.12, N = 1513.85MIN: 12.83 / MAX: 247.131. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: resnet50phoronix-ml.txt3691215SE +/- 0.13, N = 1513.33MIN: 11.94 / MAX: 281.31. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: alexnetphoronix-ml.txt1.2512.5023.7535.0046.255SE +/- 0.07, N = 155.56MIN: 5.07 / MAX: 35.351. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: resnet18phoronix-ml.txt246810SE +/- 0.07, N = 158.02MIN: 7.54 / MAX: 17.961. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: vgg16phoronix-ml.txt612182430SE +/- 0.34, N = 1525.71MIN: 22.56 / MAX: 344.21. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: googlenetphoronix-ml.txt48121620SE +/- 0.15, N = 1516.42MIN: 15.38 / MAX: 271.341. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: blazefacephoronix-ml.txt0.69981.39962.09942.79923.499SE +/- 0.02, N = 153.11MIN: 2.85 / MAX: 11.531. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: efficientnet-b0phoronix-ml.txt246810SE +/- 0.09, N = 158.28MIN: 7.57 / MAX: 296.151. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: mnasnetphoronix-ml.txt246810SE +/- 0.11, N = 156.11MIN: 5.26 / MAX: 321.291. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: shufflenet-v2phoronix-ml.txt246810SE +/- 0.09, N = 158.15MIN: 7.52 / MAX: 291.181. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3 - Model: mobilenet-v3phoronix-ml.txt246810SE +/- 0.04, N = 156.45MIN: 5.96 / MAX: 63.771. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v2-v2 - Model: mobilenet-v2phoronix-ml.txt246810SE +/- 0.05, N = 156.30MIN: 5.63 / MAX: 33.151. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: mobilenetphoronix-ml.txt48121620SE +/- 0.12, N = 1513.85MIN: 12.83 / MAX: 247.131. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Covertype Dataset Benchmarkphoronix-ml.txt70140210280350SE +/- 0.61, N = 3320.391. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Lassophoronix-ml.txt70140210280350SE +/- 0.07, N = 3308.231. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 32 - Model: ResNet-50phoronix-ml.txt3691215SE +/- 0.01, N = 39.14

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: SGDOneClassSVMphoronix-ml.txt50100150200250SE +/- 0.33, N = 3233.411. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: TSNE MNIST Datasetphoronix-ml.txt50100150200250SE +/- 0.66, N = 3247.561. (F9X) gfortran options: -O0

OpenVINO

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

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Noise Suppression Poconet-Like FP16 - Device: CPUphoronix-ml.txt3691215SE +/- 0.18, N = 1511.53MIN: 5.76 / MAX: 42.761. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Noise Suppression Poconet-Like FP16 - Device: CPUphoronix-ml.txt400800120016002000SE +/- 35.59, N = 152052.021. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Person Detection FP16 - Device: CPUphoronix-ml.txt20406080100SE +/- 2.03, N = 1593.29MIN: 31.12 / MAX: 185.661. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Person Detection FP16 - Device: CPUphoronix-ml.txt306090120150SE +/- 3.20, N = 15129.451. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Person Detection FP32 - Device: CPUphoronix-ml.txt20406080100SE +/- 0.82, N = 15100.46MIN: 32.5 / MAX: 161.811. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Person Detection FP32 - Device: CPUphoronix-ml.txt306090120150SE +/- 1.07, N = 15119.411. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

TensorFlow Lite

This is a benchmark of the TensorFlow Lite implementation focused on TensorFlow machine learning for mobile, IoT, edge, and other cases. The current Linux support is limited to running on CPUs. This test profile is measuring the average inference time. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2022-05-18Model: Inception V4phoronix-ml.txt4K8K12K16K20KSE +/- 520.83, N = 1520372.6

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2022-05-18Model: NASNet Mobilephoronix-ml.txt7K14K21K28K35KSE +/- 419.25, N = 1533662.5

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2022-05-18Model: SqueezeNetphoronix-ml.txt400800120016002000SE +/- 17.45, N = 151836.28

OpenVINO

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

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Road Segmentation ADAS FP16 - Device: CPUphoronix-ml.txt612182430SE +/- 0.43, N = 1525.83MIN: 10.2 / MAX: 57.071. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Road Segmentation ADAS FP16 - Device: CPUphoronix-ml.txt100200300400500SE +/- 9.05, N = 15465.861. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Vehicle Detection FP16 - Device: CPUphoronix-ml.txt3691215SE +/- 0.14, N = 1511.12MIN: 4.52 / MAX: 34.071. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Vehicle Detection FP16 - Device: CPUphoronix-ml.txt2004006008001000SE +/- 15.55, N = 151077.871. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

ONNX Runtime

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: CaffeNet 12-int8 - Device: CPU - Executor: Parallelphoronix-ml.txt1.10872.21743.32614.43485.5435SE +/- 0.04054, N = 154.927681. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: CaffeNet 12-int8 - Device: CPU - Executor: Parallelphoronix-ml.txt4080120160200SE +/- 1.63, N = 15203.061. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Isolation Forestphoronix-ml.txt4080120160200SE +/- 0.54, N = 3176.291. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 64 - Model: GoogLeNetphoronix-ml.txt714212835SE +/- 0.02, N = 328.53

ONNX Runtime

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: fcn-resnet101-11 - Device: CPU - Executor: Parallelphoronix-ml.txt2004006008001000SE +/- 16.79, N = 12834.571. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: fcn-resnet101-11 - Device: CPU - Executor: Parallelphoronix-ml.txt0.27080.54160.81241.08321.354SE +/- 0.02346, N = 121.203391. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 64 - Model: VGG-16phoronix-ml.txt714212835SE +/- 0.02, N = 329.05

OpenVINO

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

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Machine Translation EN To DE FP16 - Device: CPUphoronix-ml.txt1428425670SE +/- 1.08, N = 1263.99MIN: 29.61 / MAX: 110.091. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Machine Translation EN To DE FP16 - Device: CPUphoronix-ml.txt4080120160200SE +/- 3.37, N = 12187.881. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: GLMphoronix-ml.txt4080120160200SE +/- 0.81, N = 3168.331. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Hist Gradient Boostingphoronix-ml.txt4080120160200SE +/- 0.90, N = 3166.781. (F9X) gfortran options: -O0

Whisper.cpp

OpenBenchmarking.orgSeconds, Fewer Is BetterWhisper.cpp 1.6.2Model: ggml-small.en - Input: 2016 State of the Unionphoronix-ml.txt50100150200250SE +/- 0.41, N = 3218.191. (CXX) g++ options: -O3 -std=c++11 -fPIC -pthread -msse3 -mssse3 -mavx -mf16c -mfma -mavx2 -mavx512f -mavx512cd -mavx512vl -mavx512dq -mavx512bw -mavx512vbmi -mavx512vnni

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 16 - Model: ResNet-50phoronix-ml.txt246810SE +/- 0.01, N = 38.94

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 32 - Model: Efficientnet_v2_lphoronix-ml.txt3691215SE +/- 0.07, N = 39.85MIN: 8.27 / MAX: 10.3

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 16 - Model: Efficientnet_v2_lphoronix-ml.txt3691215SE +/- 0.09, N = 39.90MIN: 8.07 / MAX: 10.35

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 64 - Model: Efficientnet_v2_lphoronix-ml.txt3691215SE +/- 0.05, N = 39.98MIN: 7.98 / MAX: 10.29

Mobile Neural Network

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: inception-v3phoronix-ml.txt816243240SE +/- 0.04, N = 336.45MIN: 36.16 / MAX: 50.971. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: mobilenet-v1-1.0phoronix-ml.txt0.85141.70282.55423.40564.257SE +/- 0.007, N = 33.784MIN: 3.71 / MAX: 6.581. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: MobileNetV2_224phoronix-ml.txt0.73531.47062.20592.94123.6765SE +/- 0.042, N = 33.268MIN: 3.15 / MAX: 5.051. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: SqueezeNetV1.0phoronix-ml.txt246810SE +/- 0.200, N = 36.429MIN: 5.97 / MAX: 7.141. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: resnet-v2-50phoronix-ml.txt510152025SE +/- 0.11, N = 318.53MIN: 18.26 / MAX: 28.91. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: squeezenetv1.1phoronix-ml.txt0.97361.94722.92083.89444.868SE +/- 0.117, N = 34.327MIN: 3.96 / MAX: 6.651. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: mobilenetV3phoronix-ml.txt0.57061.14121.71182.28242.853SE +/- 0.008, N = 32.536MIN: 2.4 / MAX: 3.471. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

OpenBenchmarking.orgms, Fewer Is BetterMobile Neural Network 2.9.b11b7037dModel: nasnetphoronix-ml.txt48121620SE +/- 0.02, N = 315.30MIN: 14.65 / MAX: 21.591. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -pthread -ldl

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 256 - Model: Efficientnet_v2_lphoronix-ml.txt3691215SE +/- 0.05, N = 310.11MIN: 8.31 / MAX: 10.38

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Plot Hierarchicalphoronix-ml.txt306090120150SE +/- 0.41, N = 3141.461. (F9X) gfortran options: -O0

XNNPACK

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: QS8MobileNetV2phoronix-ml.txt30060090012001500SE +/- 7.84, N = 313981. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP16MobileNetV3Smallphoronix-ml.txt30060090012001500SE +/- 5.55, N = 314641. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP16MobileNetV3Largephoronix-ml.txt5001000150020002500SE +/- 6.66, N = 321281. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP16MobileNetV2phoronix-ml.txt30060090012001500SE +/- 15.14, N = 314951. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP16MobileNetV1phoronix-ml.txt2004006008001000SE +/- 6.56, N = 311441. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP32MobileNetV3Smallphoronix-ml.txt30060090012001500SE +/- 3.61, N = 315031. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP32MobileNetV3Largephoronix-ml.txt5001000150020002500SE +/- 12.67, N = 324651. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP32MobileNetV2phoronix-ml.txt400800120016002000SE +/- 14.40, N = 318731. (CXX) g++ options: -O3 -lrt -lm

OpenBenchmarking.orgus, Fewer Is BetterXNNPACK b7b048Model: FP32MobileNetV1phoronix-ml.txt30060090012001500SE +/- 2.52, N = 312331. (CXX) g++ options: -O3 -lrt -lm

SHOC Scalable HeterOgeneous Computing

The CUDA and OpenCL version of Vetter's Scalable HeterOgeneous Computing benchmark suite. SHOC provides a number of different benchmark programs for evaluating the performance and stability of compute devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: S3Dphoronix-ml.txt70140210280350SE +/- 3.81, N = 15298.491. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenCV

This is a benchmark of the OpenCV (Computer Vision) library's built-in performance tests. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenCV 4.7Test: DNN - Deep Neural Networkphoronix-ml.txt7K14K21K28K35KSE +/- 1066.17, N = 15330801. (CXX) g++ options: -fsigned-char -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -msse -msse2 -msse3 -fvisibility=hidden -O3 -ldl -lm -lpthread -lrt

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Hist Gradient Boosting Categorical Onlyphoronix-ml.txt714212835SE +/- 0.30, N = 1530.191. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Plot Neighborsphoronix-ml.txt306090120150SE +/- 0.47, N = 3114.841. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 64 - Model: AlexNetphoronix-ml.txt1122334455SE +/- 0.04, N = 347.85

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Sparsifyphoronix-ml.txt20406080100SE +/- 0.30, N = 3108.461. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Plot Polynomial Kernel Approximationphoronix-ml.txt20406080100SE +/- 0.04, N = 3104.701. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Feature Expansionsphoronix-ml.txt20406080100SE +/- 0.56, N = 3100.351. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 32 - Model: GoogLeNetphoronix-ml.txt714212835SE +/- 0.04, N = 327.91

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: GoogLeNetphoronix-ml.txt50100150200250SE +/- 0.25, N = 3227.06

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Plot Wardphoronix-ml.txt1020304050SE +/- 0.35, N = 842.101. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 32 - Model: VGG-16phoronix-ml.txt714212835SE +/- 0.05, N = 328.32

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Sample Without Replacementphoronix-ml.txt20406080100SE +/- 0.76, N = 390.641. (F9X) gfortran options: -O0

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 64 - Model: ResNet-152phoronix-ml.txt48121620SE +/- 0.08, N = 317.64MIN: 14.41 / MAX: 18.1

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 256 - Model: ResNet-152phoronix-ml.txt48121620SE +/- 0.06, N = 317.92MIN: 14.68 / MAX: 18.36

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 512 - Model: ResNet-152phoronix-ml.txt48121620SE +/- 0.07, N = 317.99MIN: 14.67 / MAX: 18.38

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 32 - Model: ResNet-152phoronix-ml.txt48121620SE +/- 0.03, N = 317.78MIN: 15.02 / MAX: 18.13

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 16 - Model: ResNet-152phoronix-ml.txt48121620SE +/- 0.07, N = 317.97MIN: 14.56 / MAX: 18.28

Numpy Benchmark

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

OpenBenchmarking.orgScore, More Is BetterNumpy Benchmarkphoronix-ml.txt150300450600750SE +/- 5.45, N = 3715.50

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 64 - Model: ResNet-50phoronix-ml.txt1632486480SE +/- 0.33, N = 370.89

Whisper.cpp

OpenBenchmarking.orgSeconds, Fewer Is BetterWhisper.cpp 1.6.2Model: ggml-base.en - Input: 2016 State of the Unionphoronix-ml.txt20406080100SE +/- 0.44, N = 392.751. (CXX) g++ options: -O3 -std=c++11 -fPIC -pthread -msse3 -mssse3 -mavx -mf16c -mfma -mavx2 -mavx512f -mavx512cd -mavx512vl -mavx512dq -mavx512bw -mavx512vbmi -mavx512vnni

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Treephoronix-ml.txt1122334455SE +/- 0.48, N = 546.971. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 512 - Model: AlexNetphoronix-ml.txt140280420560700SE +/- 1.19, N = 3643.44

NCNN

NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: FastestDetphoronix-ml.txt3691215SE +/- 0.39, N = 39.82MIN: 8.74 / MAX: 19.331. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: vision_transformerphoronix-ml.txt918273645SE +/- 0.27, N = 341.05MIN: 38.99 / MAX: 101.631. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: regnety_400mphoronix-ml.txt510152025SE +/- 0.09, N = 318.63MIN: 18.07 / MAX: 108.41. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: squeezenet_ssdphoronix-ml.txt48121620SE +/- 1.72, N = 316.04MIN: 13.26 / MAX: 580.791. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: yolov4-tinyphoronix-ml.txt612182430SE +/- 0.41, N = 323.64MIN: 22.17 / MAX: 39.941. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPUv2-yolov3v2-yolov3 - Model: mobilenetv2-yolov3phoronix-ml.txt48121620SE +/- 0.10, N = 313.79MIN: 13.15 / MAX: 23.261. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: resnet50phoronix-ml.txt48121620SE +/- 1.16, N = 314.65MIN: 12.41 / MAX: 377.471. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: alexnetphoronix-ml.txt1.1882.3763.5644.7525.94SE +/- 0.02, N = 35.28MIN: 5.1 / MAX: 15.431. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: resnet18phoronix-ml.txt246810SE +/- 0.08, N = 37.86MIN: 7.54 / MAX: 15.181. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: vgg16phoronix-ml.txt612182430SE +/- 0.31, N = 325.13MIN: 23.1 / MAX: 139.051. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: googlenetphoronix-ml.txt48121620SE +/- 0.04, N = 316.01MIN: 15.51 / MAX: 26.581. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: blazefacephoronix-ml.txt0.70651.4132.11952.8263.5325SE +/- 0.00, N = 33.14MIN: 3.01 / MAX: 8.51. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: efficientnet-b0phoronix-ml.txt246810SE +/- 0.37, N = 38.68MIN: 7.76 / MAX: 202.361. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: mnasnetphoronix-ml.txt1.34782.69564.04345.39126.739SE +/- 0.01, N = 35.99MIN: 5.68 / MAX: 14.251. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: shufflenet-v2phoronix-ml.txt246810SE +/- 0.05, N = 38.06MIN: 7.83 / MAX: 15.081. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3phoronix-ml.txt246810SE +/- 0.03, N = 36.49MIN: 6.21 / MAX: 15.751. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2phoronix-ml.txt246810SE +/- 0.02, N = 36.31MIN: 5.98 / MAX: 14.561. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: mobilenetphoronix-ml.txt48121620SE +/- 0.10, N = 313.79MIN: 13.15 / MAX: 23.261. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Hist Gradient Boosting Threadingphoronix-ml.txt1224364860SE +/- 0.61, N = 452.731. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: SGD Regressionphoronix-ml.txt1428425670SE +/- 0.08, N = 364.371. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Kernel PCA Solvers / Time vs. N Samplesphoronix-ml.txt1428425670SE +/- 0.24, N = 361.611. (F9X) gfortran options: -O0

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 1 - Model: Efficientnet_v2_lphoronix-ml.txt48121620SE +/- 0.19, N = 314.18MIN: 12.11 / MAX: 14.81

ONNX Runtime

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: ZFNet-512 - Device: CPU - Executor: Parallelphoronix-ml.txt48121620SE +/- 0.20, N = 417.291. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: ZFNet-512 - Device: CPU - Executor: Parallelphoronix-ml.txt1326395265SE +/- 0.67, N = 457.861. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: ZFNet-512 - Device: CPU - Executor: Standardphoronix-ml.txt3691215SE +/- 0.10702, N = 49.103761. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: ZFNet-512 - Device: CPU - Executor: Standardphoronix-ml.txt20406080100SE +/- 1.32, N = 4109.861. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: T5 Encoder - Device: CPU - Executor: Standardphoronix-ml.txt1.30842.61683.92525.23366.542SE +/- 0.07434, N = 45.814891. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: T5 Encoder - Device: CPU - Executor: Standardphoronix-ml.txt4080120160200SE +/- 2.15, N = 4172.041. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 32 - Model: AlexNetphoronix-ml.txt1020304050SE +/- 0.04, N = 346.14

oneDNN

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.6Harness: Recurrent Neural Network Training - Engine: CPUphoronix-ml.txt30060090012001500SE +/- 9.63, N = 31261.40MIN: 1196.781. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -fcf-protection=full -pie -ldl

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.6Harness: IP Shapes 1D - Engine: CPUphoronix-ml.txt0.25570.51140.76711.02281.2785SE +/- 0.00979, N = 151.13657MIN: 1.011. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -fcf-protection=full -pie -ldl

SHOC Scalable HeterOgeneous Computing

The CUDA and OpenCL version of Vetter's Scalable HeterOgeneous Computing benchmark suite. SHOC provides a number of different benchmark programs for evaluating the performance and stability of compute devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Max SP Flopsphoronix-ml.txt20K40K60K80K100KSE +/- 230.38, N = 393757.31. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

oneDNN

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.6Harness: Recurrent Neural Network Inference - Engine: CPUphoronix-ml.txt160320480640800SE +/- 8.66, N = 3736.40MIN: 639.281. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -fcf-protection=full -pie -ldl

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: MNIST Datasetphoronix-ml.txt1224364860SE +/- 0.42, N = 352.741. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 16 - Model: GoogLeNetphoronix-ml.txt612182430SE +/- 0.01, N = 326.91

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 16 - Model: VGG-16phoronix-ml.txt612182430SE +/- 0.05, N = 327.34

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Plot Incremental PCAphoronix-ml.txt714212835SE +/- 0.08, N = 331.211. (F9X) gfortran options: -O0

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Text Vectorizersphoronix-ml.txt1020304050SE +/- 0.18, N = 345.341. (F9X) gfortran options: -O0

OpenVINO

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

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Face Detection FP16 - Device: CPUphoronix-ml.txt130260390520650SE +/- 2.89, N = 3607.48MIN: 575.29 / MAX: 657.061. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Face Detection FP16 - Device: CPUphoronix-ml.txt510152025SE +/- 0.09, N = 319.691. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Face Detection FP16-INT8 - Device: CPUphoronix-ml.txt70140210280350SE +/- 0.15, N = 3320.42MIN: 299.06 / MAX: 380.181. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Face Detection FP16-INT8 - Device: CPUphoronix-ml.txt918273645SE +/- 0.02, N = 337.361. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

ONNX Runtime

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: ResNet101_DUC_HDC-12 - Device: CPU - Executor: Parallelphoronix-ml.txt170340510680850SE +/- 2.03, N = 3770.911. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: ResNet101_DUC_HDC-12 - Device: CPU - Executor: Parallelphoronix-ml.txt0.29190.58380.87571.16761.4595SE +/- 0.00341, N = 31.297181. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: ResNet101_DUC_HDC-12 - Device: CPU - Executor: Standardphoronix-ml.txt100200300400500SE +/- 1.06, N = 3452.311. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: ResNet101_DUC_HDC-12 - Device: CPU - Executor: Standardphoronix-ml.txt0.49740.99481.49221.98962.487SE +/- 0.00518, N = 32.210861. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: fcn-resnet101-11 - Device: CPU - Executor: Standardphoronix-ml.txt50100150200250SE +/- 0.28, N = 3242.911. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: fcn-resnet101-11 - Device: CPU - Executor: Standardphoronix-ml.txt0.92631.85262.77893.70524.6315SE +/- 0.00482, N = 34.116761. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: yolov4 - Device: CPU - Executor: Parallelphoronix-ml.txt4080120160200SE +/- 2.57, N = 3185.361. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: yolov4 - Device: CPU - Executor: Parallelphoronix-ml.txt1.21432.42863.64294.85726.0715SE +/- 0.07557, N = 35.396841. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: T5 Encoder - Device: CPU - Executor: Parallelphoronix-ml.txt0.82641.65282.47923.30564.132SE +/- 0.01909, N = 33.673071. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: T5 Encoder - Device: CPU - Executor: Parallelphoronix-ml.txt60120180240300SE +/- 1.41, N = 3272.181. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: yolov4 - Device: CPU - Executor: Standardphoronix-ml.txt20406080100SE +/- 0.38, N = 3103.581. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: yolov4 - Device: CPU - Executor: Standardphoronix-ml.txt3691215SE +/- 0.03544, N = 39.654421. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenVINO

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

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Road Segmentation ADAS FP16-INT8 - Device: CPUphoronix-ml.txt48121620SE +/- 0.06, N = 317.62MIN: 9.01 / MAX: 33.691. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Road Segmentation ADAS FP16-INT8 - Device: CPUphoronix-ml.txt150300450600750SE +/- 2.47, N = 3679.661. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Person Vehicle Bike Detection FP16 - Device: CPUphoronix-ml.txt246810SE +/- 0.01, N = 36.18MIN: 3.57 / MAX: 20.461. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Person Vehicle Bike Detection FP16 - Device: CPUphoronix-ml.txt400800120016002000SE +/- 3.67, N = 31930.081. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

TensorFlow Lite

This is a benchmark of the TensorFlow Lite implementation focused on TensorFlow machine learning for mobile, IoT, edge, and other cases. The current Linux support is limited to running on CPUs. This test profile is measuring the average inference time. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2022-05-18Model: Inception ResNet V2phoronix-ml.txt7K14K21K28K35KSE +/- 434.39, N = 333356.3

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2022-05-18Model: Mobilenet Floatphoronix-ml.txt30060090012001500SE +/- 4.96, N = 31381.25

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2022-05-18Model: Mobilenet Quantphoronix-ml.txt5001000150020002500SE +/- 12.96, N = 32501.21

OpenVINO

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

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Person Re-Identification Retail FP16 - Device: CPUphoronix-ml.txt1.0622.1243.1864.2485.31SE +/- 0.01, N = 34.72MIN: 2.72 / MAX: 17.211. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Person Re-Identification Retail FP16 - Device: CPUphoronix-ml.txt5001000150020002500SE +/- 3.45, N = 32523.591. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Face Detection Retail FP16-INT8 - Device: CPUphoronix-ml.txt0.80551.6112.41653.2224.0275SE +/- 0.00, N = 33.58MIN: 1.95 / MAX: 16.911. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Face Detection Retail FP16-INT8 - Device: CPUphoronix-ml.txt14002800420056007000SE +/- 8.36, N = 36458.381. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Handwritten English Recognition FP16-INT8 - Device: CPUphoronix-ml.txt510152025SE +/- 0.08, N = 321.58MIN: 16.4 / MAX: 43.471. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Handwritten English Recognition FP16-INT8 - Device: CPUphoronix-ml.txt2004006008001000SE +/- 4.28, N = 31108.651. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPUphoronix-ml.txt0.06750.1350.20250.270.3375SE +/- 0.00, N = 30.3MIN: 0.17 / MAX: 9.441. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPUphoronix-ml.txt14K28K42K56K70KSE +/- 38.55, N = 367537.771. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Vehicle Detection FP16-INT8 - Device: CPUphoronix-ml.txt1.23982.47963.71944.95926.199SE +/- 0.01, N = 35.51MIN: 2.97 / MAX: 19.151. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Vehicle Detection FP16-INT8 - Device: CPUphoronix-ml.txt5001000150020002500SE +/- 3.34, N = 32160.571. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Handwritten English Recognition FP16 - Device: CPUphoronix-ml.txt612182430SE +/- 0.09, N = 323.12MIN: 14.9 / MAX: 38.921. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Handwritten English Recognition FP16 - Device: CPUphoronix-ml.txt2004006008001000SE +/- 4.12, N = 31035.291. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Age Gender Recognition Retail 0013 FP16 - Device: CPUphoronix-ml.txt0.09680.19360.29040.38720.484SE +/- 0.00, N = 30.43MIN: 0.23 / MAX: 11.961. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Age Gender Recognition Retail 0013 FP16 - Device: CPUphoronix-ml.txt10K20K30K40K50KSE +/- 17.34, N = 348433.921. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Weld Porosity Detection FP16 - Device: CPUphoronix-ml.txt3691215SE +/- 0.01, N = 312.25MIN: 6.32 / MAX: 26.61. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Weld Porosity Detection FP16 - Device: CPUphoronix-ml.txt400800120016002000SE +/- 1.95, N = 31947.671. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Weld Porosity Detection FP16-INT8 - Device: CPUphoronix-ml.txt246810SE +/- 0.01, N = 36.31MIN: 3.31 / MAX: 21.371. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Weld Porosity Detection FP16-INT8 - Device: CPUphoronix-ml.txt8001600240032004000SE +/- 3.08, N = 33742.771. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2024.0Model: Face Detection Retail FP16 - Device: CPUphoronix-ml.txt0.6211.2421.8632.4843.105SE +/- 0.01, N = 32.76MIN: 1.41 / MAX: 15.611. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2024.0Model: Face Detection Retail FP16 - Device: CPUphoronix-ml.txt9001800270036004500SE +/- 13.60, N = 34273.091. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared -ldl

ONNX Runtime

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: CaffeNet 12-int8 - Device: CPU - Executor: Standardphoronix-ml.txt0.35160.70321.05481.40641.758SE +/- 0.01747, N = 31.562531. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: CaffeNet 12-int8 - Device: CPU - Executor: Standardphoronix-ml.txt140280420560700SE +/- 7.24, N = 3639.911. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: ResNet50 v1-12-int8 - Device: CPU - Executor: Parallelphoronix-ml.txt3691215SE +/- 0.03096, N = 39.182921. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: ResNet50 v1-12-int8 - Device: CPU - Executor: Parallelphoronix-ml.txt20406080100SE +/- 0.36, N = 3108.881. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: ResNet50 v1-12-int8 - Device: CPU - Executor: Standardphoronix-ml.txt0.68521.37042.05562.74083.426SE +/- 0.00627, N = 33.045511. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: ResNet50 v1-12-int8 - Device: CPU - Executor: Standardphoronix-ml.txt70140210280350SE +/- 0.68, N = 3328.291. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: super-resolution-10 - Device: CPU - Executor: Parallelphoronix-ml.txt246810SE +/- 0.06021, N = 38.081491. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: super-resolution-10 - Device: CPU - Executor: Parallelphoronix-ml.txt306090120150SE +/- 0.93, N = 3123.741. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInference Time Cost (ms), Fewer Is BetterONNX Runtime 1.19Model: super-resolution-10 - Device: CPU - Executor: Standardphoronix-ml.txt3691215SE +/- 0.02, N = 310.281. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

OpenBenchmarking.orgInferences Per Second, More Is BetterONNX Runtime 1.19Model: super-resolution-10 - Device: CPU - Executor: Standardphoronix-ml.txt20406080100SE +/- 0.21, N = 397.321. (CXX) g++ options: -O3 -march=native -ffunction-sections -fdata-sections -mtune=native -flto=auto -fno-fat-lto-objects -ldl -lrt

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

Benchmark: Plot Non-Negative Matrix Factorization

phoronix-ml.txt: The test quit with a non-zero exit status. E: KeyError:

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 32 - Model: ResNet-50phoronix-ml.txt1530456075SE +/- 0.07, N = 367.94

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Plot OMP vs. LARSphoronix-ml.txt918273645SE +/- 0.09, N = 341.481. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 1 - Model: VGG-16phoronix-ml.txt0.4950.991.4851.982.475SE +/- 0.00, N = 32.20

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 1 - Model: ResNet-152phoronix-ml.txt612182430SE +/- 0.14, N = 323.19MIN: 19.02 / MAX: 24.35

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 1 - Model: AlexNetphoronix-ml.txt48121620SE +/- 0.14, N = 1515.38

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 256 - Model: AlexNetphoronix-ml.txt140280420560700SE +/- 0.20, N = 3627.50

oneDNN

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.6Harness: IP Shapes 3D - Engine: CPUphoronix-ml.txt0.31410.62820.94231.25641.5705SE +/- 0.01618, N = 151.39591MIN: 1.161. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -fcf-protection=full -pie -ldl

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 16 - Model: ResNet-50phoronix-ml.txt1020304050SE +/- 0.20, N = 345.59MIN: 38.82 / MAX: 46.87

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 512 - Model: ResNet-50phoronix-ml.txt1020304050SE +/- 0.29, N = 345.75MIN: 41.27 / MAX: 46.82

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 256 - Model: ResNet-50phoronix-ml.txt1020304050SE +/- 0.23, N = 346.06MIN: 38.74 / MAX: 46.92

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 32 - Model: ResNet-50phoronix-ml.txt1122334455SE +/- 0.47, N = 346.42MIN: 41.66 / MAX: 47.49

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 64 - Model: ResNet-50phoronix-ml.txt1122334455SE +/- 0.10, N = 346.69MIN: 42.52 / MAX: 47.36

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 16 - Model: AlexNetphoronix-ml.txt1020304050SE +/- 0.02, N = 342.43

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: Kernel PCA Solvers / Time vs. N Componentsphoronix-ml.txt714212835SE +/- 0.33, N = 331.041. (F9X) gfortran options: -O0

DeepSpeech

Mozilla DeepSpeech is a speech-to-text engine powered by TensorFlow for machine learning and derived from Baidu's Deep Speech research paper. This test profile times the speech-to-text process for a roughly three minute audio recording. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterDeepSpeech 0.6Acceleration: CPUphoronix-ml.txt1020304050SE +/- 0.15, N = 346.23

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 64 - Model: GoogLeNetphoronix-ml.txt50100150200250SE +/- 0.29, N = 3225.16

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: LocalOutlierFactorphoronix-ml.txt510152025SE +/- 0.13, N = 321.621. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 16 - Model: ResNet-50phoronix-ml.txt1428425670SE +/- 0.06, N = 362.29

oneDNN

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.6Harness: Deconvolution Batch shapes_1d - Engine: CPUphoronix-ml.txt0.84951.6992.54853.3984.2475SE +/- 0.00801, N = 33.77567MIN: 2.811. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -fcf-protection=full -pie -ldl

PyTorch

This is a benchmark of PyTorch making use of pytorch-benchmark [https://github.com/LukasHedegaard/pytorch-benchmark]. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgbatches/sec, More Is BetterPyTorch 2.2.1Device: CPU - Batch Size: 1 - Model: ResNet-50phoronix-ml.txt1326395265SE +/- 0.03, N = 360.17MIN: 49.64 / MAX: 62.8

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 1 - Model: ResNet-50phoronix-ml.txt246810SE +/- 0.03, N = 36.69

R Benchmark

This test is a quick-running survey of general R performance Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterR Benchmarkphoronix-ml.txt0.02820.05640.08460.11280.141SE +/- 0.0007, N = 30.1252

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 32 - Model: GoogLeNetphoronix-ml.txt50100150200250SE +/- 0.23, N = 3218.07

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterScikit-Learn 1.2.2Benchmark: 20 Newsgroups / Logistic Regressionphoronix-ml.txt3691215SE +/- 0.06, N = 310.451. (F9X) gfortran options: -O0

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 64 - Model: AlexNetphoronix-ml.txt110220330440550SE +/- 0.16, N = 3516.18

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 1 - Model: VGG-16phoronix-ml.txt3691215SE +/- 0.00, N = 39.70

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 16 - Model: GoogLeNetphoronix-ml.txt4080120160200SE +/- 0.22, N = 3198.11

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 32 - Model: AlexNetphoronix-ml.txt90180270360450SE +/- 0.28, N = 3409.56

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

Benchmark: RCV1 Logreg Convergencet

phoronix-ml.txt: The test quit with a non-zero exit status. E: IndexError: list index out of range

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 1 - Model: ResNet-50phoronix-ml.txt510152025SE +/- 0.11, N = 318.38

oneDNN

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.6Harness: Convolution Batch Shapes Auto - Engine: CPUphoronix-ml.txt0.53171.06341.59512.12682.6585SE +/- 0.02737, N = 42.36317MIN: 1.971. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -fcf-protection=full -pie -ldl

RNNoise

RNNoise is a recurrent neural network for audio noise reduction developed by Mozilla and Xiph.Org. This test profile is a single-threaded test measuring the time to denoise a sample 26 minute long 16-bit RAW audio file using this recurrent neural network noise suppression library. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterRNNoise 0.2Input: 26 Minute Long Talking Samplephoronix-ml.txt246810SE +/- 0.019, N = 37.8521. (CC) gcc options: -O2 -pedantic -fvisibility=hidden

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 16 - Model: AlexNetphoronix-ml.txt60120180240300SE +/- 0.39, N = 3288.71

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: GPU - Batch Size: 1 - Model: GoogLeNetphoronix-ml.txt510152025SE +/- 0.15, N = 321.03

SHOC Scalable HeterOgeneous Computing

The CUDA and OpenCL version of Vetter's Scalable HeterOgeneous Computing benchmark suite. SHOC provides a number of different benchmark programs for evaluating the performance and stability of compute devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Texture Read Bandwidthphoronix-ml.txt2004006008001000SE +/- 5.65, N = 31003.321. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

TensorFlow

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

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 1 - Model: AlexNetphoronix-ml.txt714212835SE +/- 0.00, N = 330.66

OpenBenchmarking.orgimages/sec, More Is BetterTensorFlow 2.16.1Device: CPU - Batch Size: 1 - Model: GoogLeNetphoronix-ml.txt1428425670SE +/- 0.30, N = 360.92

oneDNN

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.6Harness: Deconvolution Batch shapes_3d - Engine: CPUphoronix-ml.txt0.41670.83341.25011.66682.0835SE +/- 0.01905, N = 41.85206MIN: 1.731. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -fcf-protection=full -pie -ldl

SHOC Scalable HeterOgeneous Computing

The CUDA and OpenCL version of Vetter's Scalable HeterOgeneous Computing benchmark suite. SHOC provides a number of different benchmark programs for evaluating the performance and stability of compute devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Triadphoronix-ml.txt612182430SE +/- 0.23, N = 623.051. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: GEMM SGEMM_Nphoronix-ml.txt2K4K6K8K10KSE +/- 23.01, N = 38470.021. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Bus Speed Downloadphoronix-ml.txt612182430SE +/- 0.00, N = 324.991. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Bus Speed Readbackphoronix-ml.txt612182430SE +/- 0.00, N = 326.251. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Reductionphoronix-ml.txt130260390520650SE +/- 0.41, N = 3595.051. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: FFT SPphoronix-ml.txt6001200180024003000SE +/- 2.81, N = 32703.371. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGHash/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: MD5 Hashphoronix-ml.txt1122334455SE +/- 0.68, N = 349.641. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

Scikit-Learn

Scikit-learn is a Python module for machine learning built on NumPy, SciPy, and is BSD-licensed. Learn more via the OpenBenchmarking.org test page.

Benchmark: Plot Fast KMeans

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'matplotlib.tri.triangulation'

Benchmark: Plot Lasso Path

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'matplotlib.tri.triangulation'

Benchmark: Plot Singular Value Decomposition

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'matplotlib.tri.triangulation'

Benchmark: Glmnet

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'glmnet'

PlaidML

This test profile uses PlaidML deep learning framework developed by Intel for offering up various benchmarks. Learn more via the OpenBenchmarking.org test page.

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

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'tensorflow'

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

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'tensorflow'

Mlpack Benchmark

Mlpack benchmark scripts for machine learning libraries Learn more via the OpenBenchmarking.org test page.

Benchmark: scikit_svm

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'imp'

Benchmark: scikit_linearridgeregression

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'imp'

Benchmark: scikit_qda

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'imp'

Benchmark: scikit_ica

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'imp'

Numenta Anomaly Benchmark

Numenta Anomaly Benchmark (NAB) is a benchmark for evaluating algorithms for anomaly detection in streaming, real-time applications. It is comprised of over 50 labeled real-world and artificial time-series data files plus a novel scoring mechanism designed for real-time applications. This test profile currently measures the time to run various detectors. Learn more via the OpenBenchmarking.org test page.

Detector: Contextual Anomaly Detector OSE

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'pandas'

Detector: Bayesian Changepoint

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'pandas'

Detector: Earthgecko Skyline

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'pandas'

Detector: Windowed Gaussian

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'pandas'

Detector: Relative Entropy

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'pandas'

Detector: KNN CAD

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'pandas'

spaCy

The spaCy library is an open-source solution for advanced neural language processing (NLP). The spaCy library leverages Python and is a leading neural language processing solution. This test profile times the spaCy CPU performance with various models. Learn more via the OpenBenchmarking.org test page.

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'tqdm'

AI Benchmark Alpha

AI Benchmark Alpha is a Python library for evaluating artificial intelligence (AI) performance on diverse hardware platforms and relies upon the TensorFlow machine learning library. Learn more via the OpenBenchmarking.org test page.

phoronix-ml.txt: The test quit with a non-zero exit status. E: ModuleNotFoundError: No module named 'tensorflow'

ONNX Runtime

Model: Faster R-CNN R-50-FPN-int8 - Device: CPU - Executor: Standard

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "FasterRCNN-12-int8/FasterRCNN-12-int8.onnx" failed: No such file or directory

Model: GPT-2 - Device: CPU - Executor: Standard

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "GPT2/model.onnx" failed: No such file or directory

Model: bertsquad-12 - Device: CPU - Executor: Parallel

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "bertsquad-12/bertsquad-12.onnx" failed: No such file or directory

Model: Faster R-CNN R-50-FPN-int8 - Device: CPU - Executor: Parallel

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "FasterRCNN-12-int8/FasterRCNN-12-int8.onnx" failed: No such file or directory

Model: ArcFace ResNet-100 - Device: CPU - Executor: Parallel

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "resnet100/resnet100.onnx" failed: No such file or directory

Model: GPT-2 - Device: CPU - Executor: Parallel

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "GPT2/model.onnx" failed: No such file or directory

Llama.cpp

Model: llama-2-70b-chat.Q5_0.gguf

phoronix-ml.txt: The test quit with a non-zero exit status. E: main: error: unable to load model

Model: llama-2-13b.Q4_0.gguf

phoronix-ml.txt: The test quit with a non-zero exit status. E: main: error: unable to load model

Model: llama-2-7b.Q4_0.gguf

phoronix-ml.txt: The test quit with a non-zero exit status. E: main: error: unable to load model

ONNX Runtime

Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "resnet100/resnet100.onnx" failed: No such file or directory

Model: bertsquad-12 - Device: CPU - Executor: Standard

phoronix-ml.txt: The test quit with a non-zero exit status. E: onnxruntime/onnxruntime/test/onnx/onnx_model_info.cc:45 void OnnxModelInfo::InitOnnxModelInfo(const std::filesystem::__cxx11::path&) open file "bertsquad-12/bertsquad-12.onnx" failed: No such file or directory

Llamafile

Test: wizardcoder-python-34b-v1.0.Q6_K - Acceleration: CPU

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./run-wizardcoder: line 2: ./wizardcoder-python-34b-v1.0.Q6_K.llamafile.86: No such file or directory

Test: mistral-7b-instruct-v0.2.Q8_0 - Acceleration: CPU

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./run-mistral: line 2: ./mistral-7b-instruct-v0.2.Q5_K_M.llamafile.86: No such file or directory

Test: llava-v1.5-7b-q4 - Acceleration: CPU

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./run-llava: line 2: ./llava-v1.6-mistral-7b.Q8_0.llamafile.86: No such file or directory

TNN

TNN is an open-source deep learning reasoning framework developed by Tencent. Learn more via the OpenBenchmarking.org test page.

Target: CPU - Model: SqueezeNet v1.1

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./tnn: 3: ./test/TNNTest: not found

Target: CPU - Model: SqueezeNet v2

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./tnn: 3: ./test/TNNTest: not found

Target: CPU - Model: MobileNet v2

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./tnn: 3: ./test/TNNTest: not found

Target: CPU - Model: DenseNet

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./tnn: 3: ./test/TNNTest: not found

Caffe

This is a benchmark of the Caffe deep learning framework and currently supports the AlexNet and Googlenet model and execution on both CPUs and NVIDIA GPUs. Learn more via the OpenBenchmarking.org test page.

Model: GoogleNet - Acceleration: CPU - Iterations: 1000

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./caffe: 3: ./tools/caffe: not found

Model: GoogleNet - Acceleration: CPU - Iterations: 200

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./caffe: 3: ./tools/caffe: not found

Model: GoogleNet - Acceleration: CPU - Iterations: 100

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./caffe: 3: ./tools/caffe: not found

Model: AlexNet - Acceleration: CPU - Iterations: 1000

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./caffe: 3: ./tools/caffe: not found

Model: AlexNet - Acceleration: CPU - Iterations: 200

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./caffe: 3: ./tools/caffe: not found

Model: AlexNet - Acceleration: CPU - Iterations: 100

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./caffe: 3: ./tools/caffe: not found

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.

Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: NLP Text Classification, DistilBERT mnli - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: CV Classification, ResNet-50 ImageNet - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: Llama2 Chat 7b Quantized - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: Llama2 Chat 7b Quantized - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: ResNet-50, Sparse INT8 - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: ResNet-50, Sparse INT8 - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: ResNet-50, Baseline - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: ResNet-50, Baseline - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Synchronous Single-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Stream

phoronix-ml.txt: The test quit with a non-zero exit status. E: ./deepsparse: 2: /.local/bin/deepsparse.benchmark: not found

257 Results Shown

TensorFlow:
  GPU - 512 - VGG-16
  GPU - 256 - VGG-16
  GPU - 512 - ResNet-50
Scikit-Learn
TensorFlow:
  GPU - 256 - ResNet-50
  GPU - 64 - VGG-16
Scikit-Learn
TensorFlow
Scikit-Learn
TensorFlow:
  CPU - 512 - VGG-16
  CPU - 256 - ResNet-50
  GPU - 32 - VGG-16
  GPU - 512 - AlexNet
LeelaChessZero
Scikit-Learn
TensorFlow:
  GPU - 256 - GoogLeNet
  CPU - 512 - ResNet-50
  CPU - 256 - VGG-16
PyTorch
TensorFlow:
  GPU - 64 - ResNet-50
  CPU - 512 - GoogLeNet
  GPU - 16 - VGG-16
Scikit-Learn:
  Sparse Rand Projections / 100 Iterations
  Hist Gradient Boosting Adult
Whisper.cpp
TensorFlow
Scikit-Learn:
  Plot Parallel Pairwise
  Hist Gradient Boosting Higgs Boson
NCNN:
  CPU - FastestDet
  CPU - vision_transformer
  CPU - regnety_400m
  CPU - squeezenet_ssd
  CPU - yolov4-tiny
  CPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3
  CPU - resnet50
  CPU - alexnet
  CPU - resnet18
  CPU - vgg16
  CPU - googlenet
  CPU - blazeface
  CPU - efficientnet-b0
  CPU - mnasnet
  CPU - shufflenet-v2
  CPU-v3-v3 - mobilenet-v3
  CPU-v2-v2 - mobilenet-v2
  CPU - mobilenet
Scikit-Learn:
  Covertype Dataset Benchmark
  Lasso
TensorFlow
Scikit-Learn:
  SGDOneClassSVM
  TSNE MNIST Dataset
OpenVINO:
  Noise Suppression Poconet-Like FP16 - CPU:
    ms
    FPS
  Person Detection FP16 - CPU:
    ms
    FPS
  Person Detection FP32 - CPU:
    ms
    FPS
TensorFlow Lite:
  Inception V4
  NASNet Mobile
  SqueezeNet
OpenVINO:
  Road Segmentation ADAS FP16 - CPU:
    ms
    FPS
  Vehicle Detection FP16 - CPU:
    ms
    FPS
ONNX Runtime:
  CaffeNet 12-int8 - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
Scikit-Learn
TensorFlow
ONNX Runtime:
  fcn-resnet101-11 - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
TensorFlow
OpenVINO:
  Machine Translation EN To DE FP16 - CPU:
    ms
    FPS
Scikit-Learn:
  GLM
  Hist Gradient Boosting
Whisper.cpp
TensorFlow
PyTorch:
  CPU - 32 - Efficientnet_v2_l
  CPU - 16 - Efficientnet_v2_l
  CPU - 64 - Efficientnet_v2_l
Mobile Neural Network:
  inception-v3
  mobilenet-v1-1.0
  MobileNetV2_224
  SqueezeNetV1.0
  resnet-v2-50
  squeezenetv1.1
  mobilenetV3
  nasnet
PyTorch
Scikit-Learn
XNNPACK:
  QS8MobileNetV2
  FP16MobileNetV3Small
  FP16MobileNetV3Large
  FP16MobileNetV2
  FP16MobileNetV1
  FP32MobileNetV3Small
  FP32MobileNetV3Large
  FP32MobileNetV2
  FP32MobileNetV1
SHOC Scalable HeterOgeneous Computing
OpenCV
Scikit-Learn:
  Hist Gradient Boosting Categorical Only
  Plot Neighbors
TensorFlow
Scikit-Learn:
  Sparsify
  Plot Polynomial Kernel Approximation
  Feature Expansions
TensorFlow:
  GPU - 32 - GoogLeNet
  CPU - 256 - GoogLeNet
Scikit-Learn
TensorFlow
Scikit-Learn
PyTorch:
  CPU - 64 - ResNet-152
  CPU - 256 - ResNet-152
  CPU - 512 - ResNet-152
  CPU - 32 - ResNet-152
  CPU - 16 - ResNet-152
Numpy Benchmark
TensorFlow
Whisper.cpp
Scikit-Learn
TensorFlow
NCNN:
  Vulkan GPU - FastestDet
  Vulkan GPU - vision_transformer
  Vulkan GPU - regnety_400m
  Vulkan GPU - squeezenet_ssd
  Vulkan GPU - yolov4-tiny
  Vulkan GPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3
  Vulkan GPU - resnet50
  Vulkan GPU - alexnet
  Vulkan GPU - resnet18
  Vulkan GPU - vgg16
  Vulkan GPU - googlenet
  Vulkan GPU - blazeface
  Vulkan GPU - efficientnet-b0
  Vulkan GPU - mnasnet
  Vulkan GPU - shufflenet-v2
  Vulkan GPU-v3-v3 - mobilenet-v3
  Vulkan GPU-v2-v2 - mobilenet-v2
  Vulkan GPU - mobilenet
Scikit-Learn:
  Hist Gradient Boosting Threading
  SGD Regression
  Kernel PCA Solvers / Time vs. N Samples
PyTorch
ONNX Runtime:
  ZFNet-512 - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
  ZFNet-512 - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
  T5 Encoder - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
TensorFlow
oneDNN:
  Recurrent Neural Network Training - CPU
  IP Shapes 1D - CPU
SHOC Scalable HeterOgeneous Computing
oneDNN
Scikit-Learn
TensorFlow:
  GPU - 16 - GoogLeNet
  CPU - 16 - VGG-16
Scikit-Learn:
  Plot Incremental PCA
  Text Vectorizers
OpenVINO:
  Face Detection FP16 - CPU:
    ms
    FPS
  Face Detection FP16-INT8 - CPU:
    ms
    FPS
ONNX Runtime:
  ResNet101_DUC_HDC-12 - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
  ResNet101_DUC_HDC-12 - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
  fcn-resnet101-11 - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
  yolov4 - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
  T5 Encoder - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
  yolov4 - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
OpenVINO:
  Road Segmentation ADAS FP16-INT8 - CPU:
    ms
    FPS
  Person Vehicle Bike Detection FP16 - CPU:
    ms
    FPS
TensorFlow Lite:
  Inception ResNet V2
  Mobilenet Float
  Mobilenet Quant
OpenVINO:
  Person Re-Identification Retail FP16 - CPU:
    ms
    FPS
  Face Detection Retail FP16-INT8 - CPU:
    ms
    FPS
  Handwritten English Recognition FP16-INT8 - CPU:
    ms
    FPS
  Age Gender Recognition Retail 0013 FP16-INT8 - CPU:
    ms
    FPS
  Vehicle Detection FP16-INT8 - CPU:
    ms
    FPS
  Handwritten English Recognition FP16 - CPU:
    ms
    FPS
  Age Gender Recognition Retail 0013 FP16 - CPU:
    ms
    FPS
  Weld Porosity Detection FP16 - CPU:
    ms
    FPS
  Weld Porosity Detection FP16-INT8 - CPU:
    ms
    FPS
  Face Detection Retail FP16 - CPU:
    ms
    FPS
ONNX Runtime:
  CaffeNet 12-int8 - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
  ResNet50 v1-12-int8 - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
  ResNet50 v1-12-int8 - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
  super-resolution-10 - CPU - Parallel:
    Inference Time Cost (ms)
    Inferences Per Second
  super-resolution-10 - CPU - Standard:
    Inference Time Cost (ms)
    Inferences Per Second
TensorFlow
Scikit-Learn
TensorFlow
PyTorch
TensorFlow:
  GPU - 1 - AlexNet
  CPU - 256 - AlexNet
oneDNN
PyTorch:
  CPU - 16 - ResNet-50
  CPU - 512 - ResNet-50
  CPU - 256 - ResNet-50
  CPU - 32 - ResNet-50
  CPU - 64 - ResNet-50
TensorFlow
Scikit-Learn
DeepSpeech
TensorFlow
Scikit-Learn
TensorFlow
oneDNN
PyTorch
TensorFlow
R Benchmark
TensorFlow
Scikit-Learn
TensorFlow:
  CPU - 64 - AlexNet
  CPU - 1 - VGG-16
  CPU - 16 - GoogLeNet
  CPU - 32 - AlexNet
  CPU - 1 - ResNet-50
oneDNN
RNNoise
TensorFlow:
  CPU - 16 - AlexNet
  GPU - 1 - GoogLeNet
SHOC Scalable HeterOgeneous Computing
TensorFlow:
  CPU - 1 - AlexNet
  CPU - 1 - GoogLeNet
oneDNN
SHOC Scalable HeterOgeneous Computing:
  OpenCL - Triad
  OpenCL - GEMM SGEMM_N
  OpenCL - Bus Speed Download
  OpenCL - Bus Speed Readback
  OpenCL - Reduction
  OpenCL - FFT SP
  OpenCL - MD5 Hash