CUDA 2016 NVIDIA Linux Ubuntu

NVIDIA CUDA Linux 2016 compute benchmarks by Michael Larabel for a future article.

HTML result view exported from: https://openbenchmarking.org/result/1702015-TA-1612261TA53.

CUDA 2016 NVIDIA Linux UbuntuProcessorMotherboardChipsetMemoryDiskGraphicsAudioNetworkOSKernelDesktopDisplay ServerDisplay DriverOpenGLVulkanCompilerFile-SystemScreen ResolutionGeForce GTX 650GeForce GTX 680GeForce GTX 750GeForce GTX 760GeForce GTX 780 TiGeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 1080ubu_deepIntel Xeon E3-1280 v5 @ 4.00GHz (8 Cores)MSI C236A WORKSTATION (MS-7998) v1.0Intel Sky Lake16384MB256GB TOSHIBA-RD400MSI NVIDIA GeForce GTX 650 1024MB (1084/2500MHz)Realtek ALC1150Intel ConnectionUbuntu 16.044.4.0-57-generic (x86_64)Unity 7.4.0X Server 1.18.4NVIDIA 375.264.5.01.0.24GCC 5.4.0 20160609 + CUDA 8.0ext43840x2160NVIDIA GeForce GTX 680 2048MB (1006/3004MHz)eVGA NVIDIA GeForce GTX 750 1024MB (1019/2505MHz)NVIDIA GeForce GTX 760 2048MB (1124/3004MHz)NVIDIA GeForce GTX 780 Ti 3072MB (875/3500MHz)eVGA NVIDIA GeForce GTX 950 2048MB (1202/3304MHz)eVGA NVIDIA GeForce GTX 960 2048MB (1277/3505MHz)eVGA NVIDIA GeForce GTX 970 4096MB (1163/3505MHz)NVIDIA GeForce GTX 980 4096MB (1126/3505MHz)NVIDIA GeForce GTX 980 Ti 6144MB (999/3505MHz)Zotac NVIDIA GeForce GTX 1050 2048MB (1681/3504MHz)eVGA NVIDIA GeForce GTX 1050 Ti 4096MB (1341/3504MHz)NVIDIA GeForce GTX 1060 6GB 6144MB (557/4006MHz)NVIDIA GeForce GTX 1070 8192MB (1069/4006MHz)NVIDIA GeForce GTX 1080 8192MB (1538/5005MHz)2 x Intel 0000 @ 3.00GHz (48 Cores)Supermicro X10DRi-LN4+ v1.01Intel Xeon E7 v4/Xeon64512MB1000GB My Passport 0820NVIDIA Device 10f0Intel I350 Gigabit Connection4.4.0-59-generic (x86_64)1.0.81024x768OpenBenchmarking.orgCompiler Details- --build=x86_64-linux-gnu --disable-browser-plugin --disable-vtable-verify --disable-werror --enable-checking=release --enable-clocale=gnu --enable-gnu-unique-object --enable-gtk-cairo --enable-java-awt=gtk --enable-java-home --enable-languages=c,ada,c++,java,go,d,fortran,objc,obj-c++ --enable-libmpx --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-arch-directory=amd64 --with-default-libstdcxx-abi=new --with-multilib-list=m32,m64,mx32 --with-tune=generic -vProcessor Details- Scaling Governor: intel_pstate powersaveOpenCL Details- GeForce GTX 650: GPU Compute Cores: 384- GeForce GTX 680: GPU Compute Cores: 1536- GeForce GTX 750: GPU Compute Cores: 512- GeForce GTX 760: GPU Compute Cores: 1152- GeForce GTX 780 Ti: GPU Compute Cores: 2880- GeForce GTX 950: GPU Compute Cores: 768- GeForce GTX 960: GPU Compute Cores: 1024- GeForce GTX 970: GPU Compute Cores: 1664- GeForce GTX 980: GPU Compute Cores: 2048- GeForce GTX 980 Ti: GPU Compute Cores: 2816- GeForce GTX 1050: GPU Compute Cores: 640- GeForce GTX 1050 Ti: GPU Compute Cores: 768- GeForce GTX 1060: GPU Compute Cores: 1280- GeForce GTX 1070: GPU Compute Cores: 1920- GeForce GTX 1080: GPU Compute Cores: 2560System Details- GeForce GTX 650: GPU Compute Cores: 384.- GeForce GTX 680: GPU Compute Cores: 1536.- GeForce GTX 750: GPU Compute Cores: 512.- GeForce GTX 760: GPU Compute Cores: 1152.- GeForce GTX 780 Ti: GPU Compute Cores: 2880.- GeForce GTX 950: GPU Compute Cores: 768.- GeForce GTX 960: GPU Compute Cores: 1024.- GeForce GTX 970: GPU Compute Cores: 1664.- GeForce GTX 980: GPU Compute Cores: 2048.- GeForce GTX 980 Ti: GPU Compute Cores: 2816.- GeForce GTX 1050: GPU Compute Cores: 640.- GeForce GTX 1050 Ti: GPU Compute Cores: 768.- GeForce GTX 1060: GPU Compute Cores: 1280.- GeForce GTX 1070: GPU Compute Cores: 1920.- GeForce GTX 1080: GPU Compute Cores: 2560.

CUDA 2016 NVIDIA Linux Ubuntushoc: CUDA - FFT SPshoc: CUDA - MD5 Hashshoc: CUDA - Max SP Flopsshoc: CUDA - Texture Read Bandwidthaskap: Griddingaskap: Degriddingcuda-mini-nbody: Originalcaffe: CUDA AlexNetcaffe: CUDA GooglenetGeForce GTX 650GeForce GTX 680GeForce GTX 750GeForce GTX 760GeForce GTX 780 TiGeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 1080ubu_deep52573.77133624116.131.281160.99160.58182.0866411.8316400961.4428595.1065105.93178.562.692210.77364.333399.145625.68104.1530595.4768771.30194.103.832941.88379.433132.425325.1282.7027360.7359805.47266.735.444320.28349.905255.519399.8452.5117005.6740125.47292.366.475002.78335.276006.4510798.1346.6014977.2035955.47308.497.736145.80349.098320.5017010.8036.0611722.1031440.40171.272.492109.11433.2136985873.92115.2430845.0369616.53199.713.032688.23453.153715.365961.62101.8526985.9060253.57304.625.644765.98503.115625.689861.3358.4216266.2337468.77377.278.407096.44501.087607.3113312.8039.7011451.9027658.17462.6011.909385.11526.218236.4514273.0033.069738.6524039.579734.9425701.77OpenBenchmarking.org

SHOC Scalable HeterOgeneous Computing

Target: CUDA - Benchmark: FFT SP

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: CUDA - Benchmark: FFT SPGeForce GTX 750GeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 1080100200300400500SE +/- 0.20, N = 3SE +/- 0.50, N = 3SE +/- 1.44, N = 3SE +/- 1.20, N = 3SE +/- 0.74, N = 3SE +/- 0.35, N = 3SE +/- 0.73, N = 3SE +/- 1.12, N = 3SE +/- 0.78, N = 3SE +/- 2.67, N = 3SE +/- 1.33, N = 3116.13178.56194.10266.73292.36308.49171.27199.71304.62377.27462.601. (CXX) g++ options: -O2 -lSHOCCommon -lcudadevrt -lcudart_static -lrt -lpthread -ldl -lcufft

SHOC Scalable HeterOgeneous Computing

Target: CUDA - Benchmark: MD5 Hash

OpenBenchmarking.orgGHash/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: CUDA - Benchmark: MD5 HashGeForce GTX 750GeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 10803691215SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 31.282.693.835.446.477.732.493.035.648.4011.901. (CXX) g++ options: -O2 -lSHOCCommon -lcudadevrt -lcudart_static -lrt -lpthread -ldl -lcufft

SHOC Scalable HeterOgeneous Computing

Target: CUDA - Benchmark: Max SP Flops

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: CUDA - Benchmark: Max SP FlopsGeForce GTX 750GeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 10802K4K6K8K10KSE +/- 0.08, N = 3SE +/- 6.61, N = 3SE +/- 8.34, N = 3SE +/- 3.47, N = 3SE +/- 9.75, N = 3SE +/- 20.77, N = 3SE +/- 0.30, N = 3SE +/- 5.43, N = 3SE +/- 21.65, N = 3SE +/- 50.45, N = 3SE +/- 64.36, N = 31160.992210.772941.884320.285002.786145.802109.112688.234765.987096.449385.111. (CXX) g++ options: -O2 -lSHOCCommon -lcudadevrt -lcudart_static -lrt -lpthread -ldl -lcufft

SHOC Scalable HeterOgeneous Computing

Target: CUDA - Benchmark: Texture Read Bandwidth

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: CUDA - Benchmark: Texture Read BandwidthGeForce GTX 750GeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 1080110220330440550SE +/- 0.46, N = 3SE +/- 0.34, N = 3SE +/- 0.09, N = 3SE +/- 0.05, N = 3SE +/- 0.52, N = 3SE +/- 0.22, N = 3SE +/- 1.01, N = 3SE +/- 1.17, N = 3SE +/- 0.09, N = 3SE +/- 1.65, N = 3SE +/- 1.23, N = 3160.58364.33379.43349.90335.27349.09433.21453.15503.11501.08526.211. (CXX) g++ options: -O2 -lSHOCCommon -lcudadevrt -lcudart_static -lrt -lpthread -ldl -lcufft

ASKAP tConvolveCuda

Processing: Gridding

OpenBenchmarking.orgMillion Grid Points Per Second, More Is BetterASKAP tConvolveCuda 2015-11-10Processing: GriddingGeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 10802K4K6K8K10KSE +/- 14.40, N = 3SE +/- 0.00, N = 3SE +/- 34.80, N = 3SE +/- 44.82, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 17.36, N = 3SE +/- 39.34, N = 3SE +/- 0.00, N = 3SE +/- 84.05, N = 33399.143132.425255.516006.458320.503698.003715.365625.687607.318236.451. (CXX) g++ options: -fPIC -O3 -m64 -lcudadevrt -lcudart_static -lrt -lpthread -ldl

ASKAP tConvolveCuda

Processing: Degridding

OpenBenchmarking.orgMillion Grid Points Per Second, More Is BetterASKAP tConvolveCuda 2015-11-10Processing: DegriddingGeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 10804K8K12K16K20KSE +/- 39.34, N = 3SE +/- 0.00, N = 3SE +/- 109.30, N = 3SE +/- 147.93, N = 3SE +/- 369.80, N = 3SE +/- 42.88, N = 3SE +/- 44.82, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 259.50, N = 35625.685325.129399.8410798.1317010.805873.925961.629861.3313312.8014273.001. (CXX) g++ options: -fPIC -O3 -m64 -lcudadevrt -lcudart_static -lrt -lpthread -ldl

CUDA Mini-Nbody

Test: Original

OpenBenchmarking.orgSeconds, Fewer Is BetterCUDA Mini-Nbody 2015-11-10Test: OriginalGeForce GTX 750GeForce GTX 780 TiGeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 10804080120160200SE +/- 0.15, N = 3SE +/- 0.16, N = 3SE +/- 0.40, N = 3SE +/- 0.31, N = 3SE +/- 0.14, N = 3SE +/- 0.23, N = 3SE +/- 0.48, N = 3SE +/- 0.09, N = 3SE +/- 0.14, N = 3SE +/- 0.11, N = 3SE +/- 0.05, N = 3SE +/- 0.09, N = 3182.0861.44104.1582.7052.5146.6036.06115.24101.8558.4239.7033.06

Caffe AlexNet

Build: CUDA AlexNet

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe AlexNet 2016-06-11Build: CUDA AlexNetGeForce GTX 680GeForce GTX 760GeForce GTX 780 TiGeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 1080ubu_deep14K28K42K56K70KSE +/- 76.71, N = 3SE +/- 12.08, N = 3SE +/- 3.70, N = 3SE +/- 34.72, N = 3SE +/- 5.41, N = 3SE +/- 8.15, N = 3SE +/- 2.22, N = 3SE +/- 12.80, N = 3SE +/- 2.73, N = 3SE +/- 10.92, N = 3SE +/- 7.27, N = 3SE +/- 2.23, N = 3SE +/- 11.97, N = 3SE +/- 9.00, N = 352573.7766411.8328595.1030595.4727360.7317005.6714977.2011722.1030845.0326985.9016266.2311451.909738.659734.941. (CXX) g++ options: -pthread -fPIC -O2 -lcaffe -lglog -lgflags -lprotobuf -lboost_system -lboost_filesystem -lm -lhdf5_hl -lhdf5 -lleveldb -lsnappy -llmdb -lopencv_core -lopencv_highgui -lopencv_imgproc -lboost_thread -lstdc++ -lcblas -latlas

Caffe AlexNet

Build: CUDA Googlenet

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe AlexNet 2016-06-11Build: CUDA GooglenetGeForce GTX 680GeForce GTX 760GeForce GTX 780 TiGeForce GTX 950GeForce GTX 960GeForce GTX 970GeForce GTX 980GeForce GTX 980 TiGeForce GTX 1050GeForce GTX 1050 TiGeForce GTX 1060GeForce GTX 1070GeForce GTX 1080ubu_deep40K80K120K160K200KSE +/- 31.51, N = 3SE +/- 93.86, N = 3SE +/- 20.21, N = 3SE +/- 9.66, N = 3SE +/- 13.56, N = 3SE +/- 15.33, N = 3SE +/- 77.16, N = 3SE +/- 95.74, N = 3SE +/- 7.85, N = 3SE +/- 33.10, N = 3SE +/- 4.75, N = 3SE +/- 66.90, N = 3SE +/- 5.57, N = 3SE +/- 93.86, N = 3133624.00164009.0065105.9368771.3059805.4740125.4735955.4731440.4069616.5360253.5737468.7727658.1724039.5725701.771. (CXX) g++ options: -pthread -fPIC -O2 -lcaffe -lglog -lgflags -lprotobuf -lboost_system -lboost_filesystem -lm -lhdf5_hl -lhdf5 -lleveldb -lsnappy -llmdb -lopencv_core -lopencv_highgui -lopencv_imgproc -lboost_thread -lstdc++ -lcblas -latlas

Phoronix Test Suite v10.8.4