AMD Ryzen 7 3800XT 8-Core testing with a MSI X370 XPOWER GAMING TITANIUM (MS-7A31) v1.0 (1.MS BIOS) and Sapphire AMD Radeon HD 4650 on Debian 10 via the Phoronix Test Suite.
1 Kernel Notes: Transparent Huge Pages: alwaysCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none --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 --with-tune=generic --without-cuda-driver -vProcessor Notes: Scaling Governor: acpi-cpufreq ondemand (Boost: Enabled) - CPU Microcode: 0x8701021Java Notes: OpenJDK Runtime Environment (build 11.0.11+9-post-Debian-1deb10u1)Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected
2 3 Processor: AMD Ryzen 7 3800XT 8-Core @ 3.90GHz (8 Cores / 16 Threads), Motherboard: MSI X370 XPOWER GAMING TITANIUM (MS-7A31) v1.0 (1.MS BIOS), Chipset: AMD Starship/Matisse, Memory: 16GB, Disk: 128GB INTEL SSDPEKKW128G7, Graphics: Sapphire AMD Radeon HD 4650, Audio: AMD RV710/730, Network: Intel I211
OS: Debian 10, Kernel: 4.19.0-17-amd64 (x86_64), Display Server: X Server 1.20.4, Compiler: GCC 8.3.0, File-System: ext4, Screen Resolution: 1024x768
3800xt nn Processor Motherboard Chipset Memory Disk Graphics Audio Network OS Kernel Display Server Compiler File-System Screen Resolution 1 2 3 AMD Ryzen 7 3800XT 8-Core @ 3.90GHz (8 Cores / 16 Threads) MSI X370 XPOWER GAMING TITANIUM (MS-7A31) v1.0 (1.MS BIOS) AMD Starship/Matisse 16GB 128GB INTEL SSDPEKKW128G7 Sapphire AMD Radeon HD 4650 AMD RV710/730 Intel I211 Debian 10 4.19.0-17-amd64 (x86_64) X Server 1.20.4 GCC 8.3.0 ext4 1024x768 OpenBenchmarking.org Kernel Details - Transparent Huge Pages: always Compiler Details - --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none --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 --with-tune=generic --without-cuda-driver -v Processor Details - Scaling Governor: acpi-cpufreq ondemand (Boost: Enabled) - CPU Microcode: 0x8701021 Java Details - OpenJDK Runtime Environment (build 11.0.11+9-post-Debian-1deb10u1) Security Details - itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected
1 2 3 Result Overview Phoronix Test Suite 100% 100% 101% 101% 101% Mobile Neural Network TNN NCNN Renaissance
3800xt nn renaissance: Scala Dotty renaissance: Apache Spark PageRank renaissance: In-Memory Database Shootout tnn: CPU - SqueezeNet v1.1 ncnn: CPU - alexnet renaissance: Apache Spark ALS ncnn: CPU - vgg16 renaissance: Genetic Algorithm Using Jenetics + Futures ncnn: CPU - resnet50 renaissance: Rand Forest ncnn: CPU - blazeface ncnn: CPU - mobilenet mnn: resnet-v2-50 tnn: CPU - SqueezeNet v2 mnn: squeezenetv1.1 mnn: SqueezeNetV1.0 renaissance: Savina Reactors.IO ncnn: CPU - yolov4-tiny renaissance: Apache Spark Bayes renaissance: Finagle HTTP Requests ncnn: CPU - mnasnet tnn: CPU - MobileNet v2 renaissance: Akka Unbalanced Cobwebbed Tree ncnn: CPU - squeezenet_ssd renaissance: ALS Movie Lens ncnn: CPU - shufflenet-v2 mnn: inception-v3 tnn: CPU - DenseNet ncnn: CPU-v2-v2 - mobilenet-v2 ncnn: CPU - googlenet ncnn: CPU - efficientnet-b0 mnn: mobilenetV3 ncnn: CPU - resnet18 ncnn: CPU-v3-v3 - mobilenet-v3 ncnn: CPU - regnety_400m mnn: mobilenet-v1-1.0 mnn: MobileNetV2_224 1 2 3 694.4 3623.6 3582.9 219.772 13.56 2052.5 65.35 1670.4 28.11 680.2 1.86 17.20 24.409 61.046 4.308 6.721 8181.2 26.96 1925.6 1729.8 4.29 247.988 10511.5 22.32 7176.6 4.31 32.807 2877.780 5.41 15.48 7.24 2.861 17.10 4.59 10.54 3.197 3.822 766.7 3492.3 3640.5 217.468 13.49 2027.5 66.67 1692.6 27.64 683.9 1.83 17.13 24.414 61.596 4.343 6.791 8109.3 27.18 1941.5 1742.2 4.26 248.411 10514.2 22.19 7141.8 4.32 32.924 2881.720 5.40 15.43 7.22 2.861 17.08 4.59 10.19 3.226 3.898 750.9 3421.7 3472.0 225.877 14.01 1977.4 65.82 1659.3 27.57 672.2 1.83 17.39 24.688 60.903 4.298 6.725 8110.2 26.95 1937.6 1736.4 4.27 249.550 10574.6 22.29 7153.2 4.30 32.795 2888.970 5.42 15.45 7.23 2.856 17.10 4.59 10.18 3.196 4.065 OpenBenchmarking.org
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Scala Dotty 1 2 3 170 340 510 680 850 SE +/- 4.26, N = 3 SE +/- 5.65, N = 3 SE +/- 10.32, N = 15 694.4 766.7 750.9 MIN: 568.04 / MAX: 1528.4 MIN: 563.93 / MAX: 1477.46 MIN: 559.99 / MAX: 2958.83
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Apache Spark PageRank 1 2 3 800 1600 2400 3200 4000 SE +/- 42.32, N = 3 SE +/- 50.56, N = 3 SE +/- 43.92, N = 15 3623.6 3492.3 3421.7 MIN: 3204.77 / MAX: 3716.12 MIN: 3104.55 / MAX: 3638.26 MIN: 2764.62 / MAX: 3771.68
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: In-Memory Database Shootout 1 2 3 800 1600 2400 3200 4000 SE +/- 28.02, N = 13 SE +/- 45.53, N = 5 SE +/- 43.56, N = 4 3582.9 3640.5 3472.0 MIN: 3224.53 / MAX: 3897.38 MIN: 3267.72 / MAX: 4627.58 MIN: 3137.95 / MAX: 3858.73
TNN TNN is an open-source deep learning reasoning framework developed by Tencent. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better TNN 0.3 Target: CPU - Model: SqueezeNet v1.1 1 2 3 50 100 150 200 250 SE +/- 0.31, N = 3 SE +/- 0.14, N = 3 SE +/- 0.10, N = 3 219.77 217.47 225.88 MIN: 218.71 / MAX: 229.87 MIN: 216.8 / MAX: 226.51 MIN: 224.92 / MAX: 233.1 1. (CXX) g++ options: -fopenmp -pthread -fvisibility=hidden -fvisibility=default -O3 -rdynamic -ldl
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: alexnet 1 2 3 4 8 12 16 20 SE +/- 0.03, N = 3 SE +/- 0.03, N = 3 SE +/- 0.38, N = 3 13.56 13.49 14.01 MIN: 13.39 / MAX: 13.79 MIN: 13.31 / MAX: 13.73 MIN: 13.5 / MAX: 25.5 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Apache Spark ALS 1 2 3 400 800 1200 1600 2000 SE +/- 5.11, N = 3 SE +/- 6.47, N = 3 SE +/- 23.56, N = 3 2052.5 2027.5 1977.4 MIN: 1946.3 / MAX: 2244.61 MIN: 1949.54 / MAX: 2258.66 MIN: 1850.21 / MAX: 2122.46
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: vgg16 1 2 3 15 30 45 60 75 SE +/- 0.69, N = 3 SE +/- 0.10, N = 3 SE +/- 0.63, N = 3 65.35 66.67 65.82 MIN: 64.12 / MAX: 85.11 MIN: 63.99 / MAX: 88.7 MIN: 64.27 / MAX: 85.2 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Genetic Algorithm Using Jenetics + Futures 1 2 3 400 800 1200 1600 2000 SE +/- 4.28, N = 3 SE +/- 18.80, N = 3 SE +/- 9.92, N = 3 1670.4 1692.6 1659.3 MIN: 1627.7 / MAX: 1711.42 MIN: 1624.36 / MAX: 1753.6 MIN: 1612.96 / MAX: 1705.59
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: resnet50 1 2 3 7 14 21 28 35 SE +/- 0.58, N = 3 SE +/- 0.03, N = 3 SE +/- 0.08, N = 3 28.11 27.64 27.57 MIN: 27.3 / MAX: 61.91 MIN: 27.42 / MAX: 27.99 MIN: 27.29 / MAX: 30.97 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Random Forest 1 2 3 150 300 450 600 750 SE +/- 9.40, N = 3 SE +/- 5.88, N = 3 SE +/- 2.20, N = 3 680.2 683.9 672.2 MIN: 608.7 / MAX: 840.54 MIN: 626.53 / MAX: 816.13 MIN: 616.75 / MAX: 830.59
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: blazeface 1 2 3 0.4185 0.837 1.2555 1.674 2.0925 SE +/- 0.03, N = 3 SE +/- 0.00, N = 3 SE +/- 0.00, N = 3 1.86 1.83 1.83 MIN: 1.81 / MAX: 4.44 MIN: 1.81 / MAX: 1.89 MIN: 1.8 / MAX: 2.81 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
OpenBenchmarking.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: mobilenet 1 2 3 4 8 12 16 20 SE +/- 0.08, N = 3 SE +/- 0.11, N = 3 SE +/- 0.18, N = 3 17.20 17.13 17.39 MIN: 16.61 / MAX: 18.04 MIN: 15.97 / MAX: 21.27 MIN: 16.67 / MAX: 22.42 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Mobile Neural Network MNN is the Mobile Neural Network as a highly efficient, lightweight deep learning framework developed by Alibaba. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Mobile Neural Network 1.2 Model: resnet-v2-50 1 2 3 6 12 18 24 30 SE +/- 0.13, N = 15 SE +/- 0.10, N = 15 SE +/- 0.51, N = 3 24.41 24.41 24.69 MIN: 23.21 / MAX: 41.39 MIN: 23.28 / MAX: 41.55 MIN: 23.73 / MAX: 71.67 1. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl
TNN TNN is an open-source deep learning reasoning framework developed by Tencent. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better TNN 0.3 Target: CPU - Model: SqueezeNet v2 1 2 3 14 28 42 56 70 SE +/- 0.09, N = 3 SE +/- 0.23, N = 3 SE +/- 0.42, N = 3 61.05 61.60 60.90 MIN: 60.42 / MAX: 61.71 MIN: 60.92 / MAX: 62.71 MIN: 59.77 / MAX: 61.85 1. (CXX) g++ options: -fopenmp -pthread -fvisibility=hidden -fvisibility=default -O3 -rdynamic -ldl
Mobile Neural Network MNN is the Mobile Neural Network as a highly efficient, lightweight deep learning framework developed by Alibaba. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Mobile Neural Network 1.2 Model: squeezenetv1.1 1 2 3 0.9772 1.9544 2.9316 3.9088 4.886 SE +/- 0.030, N = 15 SE +/- 0.057, N = 15 SE +/- 0.020, N = 3 4.308 4.343 4.298 MIN: 4.07 / MAX: 14.99 MIN: 4.03 / MAX: 14.8 MIN: 4.22 / MAX: 6.35 1. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl
OpenBenchmarking.org ms, Fewer Is Better Mobile Neural Network 1.2 Model: SqueezeNetV1.0 1 2 3 2 4 6 8 10 SE +/- 0.059, N = 15 SE +/- 0.072, N = 15 SE +/- 0.081, N = 3 6.721 6.791 6.725 MIN: 6.32 / MAX: 17.35 MIN: 6.39 / MAX: 19.92 MIN: 6.52 / MAX: 8.79 1. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Savina Reactors.IO 1 2 3 2K 4K 6K 8K 10K SE +/- 90.50, N = 15 SE +/- 107.95, N = 15 SE +/- 113.52, N = 15 8181.2 8109.3 8110.2 MIN: 7515.49 / MAX: 13584.35 MIN: 7362.22 / MAX: 14764.62 MIN: 7457.43 / MAX: 13392.01
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: yolov4-tiny 1 2 3 6 12 18 24 30 SE +/- 0.05, N = 3 SE +/- 0.41, N = 3 SE +/- 0.03, N = 3 26.96 27.18 26.95 MIN: 26.7 / MAX: 29.67 MIN: 26.55 / MAX: 40.87 MIN: 26.74 / MAX: 27.34 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Apache Spark Bayes 1 2 3 400 800 1200 1600 2000 SE +/- 5.21, N = 3 SE +/- 8.42, N = 3 SE +/- 2.43, N = 3 1925.6 1941.5 1937.6 MIN: 1469.38 / MAX: 1935.6 MIN: 1474.04 / MAX: 1957.52 MIN: 1469.32 / MAX: 1940.02
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Finagle HTTP Requests 1 2 3 400 800 1200 1600 2000 SE +/- 3.61, N = 3 SE +/- 4.82, N = 3 SE +/- 11.72, N = 3 1729.8 1742.2 1736.4 MIN: 1579.93 / MAX: 1789.42 MIN: 1568.53 / MAX: 1823.17 MIN: 1555.2 / MAX: 1904.32
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: mnasnet 1 2 3 0.9653 1.9306 2.8959 3.8612 4.8265 SE +/- 0.03, N = 3 SE +/- 0.01, N = 3 SE +/- 0.01, N = 3 4.29 4.26 4.27 MIN: 4.22 / MAX: 4.53 MIN: 4.21 / MAX: 4.42 MIN: 4.22 / MAX: 4.38 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
TNN TNN is an open-source deep learning reasoning framework developed by Tencent. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better TNN 0.3 Target: CPU - Model: MobileNet v2 1 2 3 50 100 150 200 250 SE +/- 0.73, N = 3 SE +/- 1.15, N = 3 SE +/- 1.57, N = 3 247.99 248.41 249.55 MIN: 240.12 / MAX: 277.2 MIN: 241.17 / MAX: 260.2 MIN: 241.36 / MAX: 275 1. (CXX) g++ options: -fopenmp -pthread -fvisibility=hidden -fvisibility=default -O3 -rdynamic -ldl
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: Akka Unbalanced Cobwebbed Tree 1 2 3 2K 4K 6K 8K 10K SE +/- 95.45, N = 3 SE +/- 22.31, N = 3 SE +/- 97.37, N = 3 10511.5 10514.2 10574.6 MIN: 8095.7 / MAX: 10698.04 MIN: 8025.58 / MAX: 10557.75 MIN: 8116.36 / MAX: 10766.94
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: squeezenet_ssd 1 2 3 5 10 15 20 25 SE +/- 0.04, N = 3 SE +/- 0.05, N = 3 SE +/- 0.08, N = 3 22.32 22.19 22.29 MIN: 22.02 / MAX: 26.98 MIN: 21.93 / MAX: 23.3 MIN: 21.96 / MAX: 22.91 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Renaissance Renaissance is a suite of benchmarks designed to test the Java JVM from Apache Spark to a Twitter-like service to Scala and other features. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Renaissance 0.12 Test: ALS Movie Lens 1 2 3 1500 3000 4500 6000 7500 SE +/- 16.48, N = 3 SE +/- 23.24, N = 3 SE +/- 7.01, N = 3 7176.6 7141.8 7153.2 MIN: 7154.04 / MAX: 7879.66 MIN: 7102.96 / MAX: 7924.37 MIN: 7139.38 / MAX: 7831.25
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: shufflenet-v2 1 2 3 0.972 1.944 2.916 3.888 4.86 SE +/- 0.01, N = 3 SE +/- 0.02, N = 3 SE +/- 0.00, N = 3 4.31 4.32 4.30 MIN: 4.25 / MAX: 4.42 MIN: 4.24 / MAX: 4.46 MIN: 4.24 / MAX: 4.6 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Mobile Neural Network MNN is the Mobile Neural Network as a highly efficient, lightweight deep learning framework developed by Alibaba. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Mobile Neural Network 1.2 Model: inception-v3 1 2 3 8 16 24 32 40 SE +/- 0.14, N = 15 SE +/- 0.10, N = 15 SE +/- 0.26, N = 3 32.81 32.92 32.80 MIN: 31.66 / MAX: 55.39 MIN: 31.72 / MAX: 54.26 MIN: 32.33 / MAX: 54 1. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl
TNN TNN is an open-source deep learning reasoning framework developed by Tencent. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better TNN 0.3 Target: CPU - Model: DenseNet 1 2 3 600 1200 1800 2400 3000 SE +/- 8.10, N = 3 SE +/- 8.27, N = 3 SE +/- 3.19, N = 3 2877.78 2881.72 2888.97 MIN: 2766.93 / MAX: 2988.44 MIN: 2773.43 / MAX: 2985.5 MIN: 2783.25 / MAX: 2990.51 1. (CXX) g++ options: -fopenmp -pthread -fvisibility=hidden -fvisibility=default -O3 -rdynamic -ldl
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU-v2-v2 - Model: mobilenet-v2 1 2 3 1.2195 2.439 3.6585 4.878 6.0975 SE +/- 0.01, N = 3 SE +/- 0.01, N = 3 SE +/- 0.00, N = 3 5.41 5.40 5.42 MIN: 5.32 / MAX: 5.53 MIN: 5.34 / MAX: 5.59 MIN: 5.36 / MAX: 5.52 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
OpenBenchmarking.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: googlenet 1 2 3 4 8 12 16 20 SE +/- 0.11, N = 3 SE +/- 0.01, N = 3 SE +/- 0.01, N = 3 15.48 15.43 15.45 MIN: 15.24 / MAX: 15.96 MIN: 15.28 / MAX: 17.42 MIN: 15.3 / MAX: 15.77 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
OpenBenchmarking.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: efficientnet-b0 1 2 3 2 4 6 8 10 SE +/- 0.04, N = 3 SE +/- 0.01, N = 3 SE +/- 0.02, N = 3 7.24 7.22 7.23 MIN: 7.14 / MAX: 12.02 MIN: 7.13 / MAX: 7.51 MIN: 7.13 / MAX: 7.37 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Mobile Neural Network MNN is the Mobile Neural Network as a highly efficient, lightweight deep learning framework developed by Alibaba. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Mobile Neural Network 1.2 Model: mobilenetV3 1 2 3 0.6437 1.2874 1.9311 2.5748 3.2185 SE +/- 0.044, N = 15 SE +/- 0.031, N = 15 SE +/- 0.005, N = 3 2.861 2.861 2.856 MIN: 2.61 / MAX: 13.16 MIN: 2.62 / MAX: 13.43 MIN: 2.82 / MAX: 3.04 1. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl
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.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: resnet18 1 2 3 4 8 12 16 20 SE +/- 0.02, N = 3 SE +/- 0.08, N = 3 SE +/- 0.11, N = 3 17.10 17.08 17.10 MIN: 16.98 / MAX: 17.38 MIN: 16.84 / MAX: 17.5 MIN: 16.86 / MAX: 22.21 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
OpenBenchmarking.org ms, Fewer Is Better NCNN 20210525 Target: CPU-v3-v3 - Model: mobilenet-v3 1 2 3 1.0328 2.0656 3.0984 4.1312 5.164 SE +/- 0.02, N = 3 SE +/- 0.00, N = 3 SE +/- 0.01, N = 3 4.59 4.59 4.59 MIN: 4.52 / MAX: 7.44 MIN: 4.53 / MAX: 4.94 MIN: 4.52 / MAX: 7.52 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
OpenBenchmarking.org ms, Fewer Is Better NCNN 20210525 Target: CPU - Model: regnety_400m 1 2 3 3 6 9 12 15 SE +/- 0.37, N = 3 SE +/- 0.00, N = 3 SE +/- 0.01, N = 3 10.54 10.19 10.18 MIN: 10.08 / MAX: 33.92 MIN: 10.1 / MAX: 10.46 MIN: 10.1 / MAX: 10.37 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
Mobile Neural Network MNN is the Mobile Neural Network as a highly efficient, lightweight deep learning framework developed by Alibaba. Learn more via the OpenBenchmarking.org test page.
OpenBenchmarking.org ms, Fewer Is Better Mobile Neural Network 1.2 Model: mobilenet-v1-1.0 1 2 3 0.7259 1.4518 2.1777 2.9036 3.6295 SE +/- 0.029, N = 15 SE +/- 0.057, N = 14 SE +/- 0.005, N = 3 3.197 3.226 3.196 MIN: 2.98 / MAX: 13.17 MIN: 3.03 / MAX: 13.31 MIN: 3.16 / MAX: 3.6 1. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl
OpenBenchmarking.org ms, Fewer Is Better Mobile Neural Network 1.2 Model: MobileNetV2_224 1 2 3 0.9146 1.8292 2.7438 3.6584 4.573 SE +/- 0.028, N = 15 SE +/- 0.038, N = 15 SE +/- 0.173, N = 3 3.822 3.898 4.065 MIN: 3.57 / MAX: 4.34 MIN: 3.65 / MAX: 14.74 MIN: 3.82 / MAX: 13.8 1. (CXX) g++ options: -std=c++11 -O3 -fvisibility=hidden -fomit-frame-pointer -fstrict-aliasing -ffunction-sections -fdata-sections -ffast-math -fno-rtti -fno-exceptions -rdynamic -pthread -ldl
1 Kernel Notes: Transparent Huge Pages: alwaysCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none --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 --with-tune=generic --without-cuda-driver -vProcessor Notes: Scaling Governor: acpi-cpufreq ondemand (Boost: Enabled) - CPU Microcode: 0x8701021Java Notes: OpenJDK Runtime Environment (build 11.0.11+9-post-Debian-1deb10u1)Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 9 July 2021 08:24 by user phoronix.
2 Kernel Notes: Transparent Huge Pages: alwaysCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none --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 --with-tune=generic --without-cuda-driver -vProcessor Notes: Scaling Governor: acpi-cpufreq ondemand (Boost: Enabled) - CPU Microcode: 0x8701021Java Notes: OpenJDK Runtime Environment (build 11.0.11+9-post-Debian-1deb10u1)Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 9 July 2021 10:36 by user phoronix.
3 Processor: AMD Ryzen 7 3800XT 8-Core @ 3.90GHz (8 Cores / 16 Threads), Motherboard: MSI X370 XPOWER GAMING TITANIUM (MS-7A31) v1.0 (1.MS BIOS), Chipset: AMD Starship/Matisse, Memory: 16GB, Disk: 128GB INTEL SSDPEKKW128G7, Graphics: Sapphire AMD Radeon HD 4650, Audio: AMD RV710/730, Network: Intel I211
OS: Debian 10, Kernel: 4.19.0-17-amd64 (x86_64), Display Server: X Server 1.20.4, Compiler: GCC 8.3.0, File-System: ext4, Screen Resolution: 1024x768
Kernel Notes: Transparent Huge Pages: alwaysCompiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none --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 --with-tune=generic --without-cuda-driver -vProcessor Notes: Scaling Governor: acpi-cpufreq ondemand (Boost: Enabled) - CPU Microcode: 0x8701021Java Notes: OpenJDK Runtime Environment (build 11.0.11+9-post-Debian-1deb10u1)Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full AMD retpoline IBPB: conditional STIBP: conditional RSB filling + srbds: Not affected + tsx_async_abort: Not affected
Testing initiated at 9 July 2021 12:46 by user phoronix.