NVIDIA AMD Linux GPU Compute December 2018

NVIDIA and AMD GPU Linux compute benchmarks December 2018 by Michael Larabel for a future article..

HTML result view exported from: https://openbenchmarking.org/result/1812085-SK-GPUCOMPUT99&grr.

NVIDIA AMD Linux GPU Compute December 2018ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerDisplay DriverOpenGLOpenCLVulkanCompilerFile-SystemScreen ResolutionGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 64Intel Core i9-9900K @ 5.00GHz (8 Cores / 16 Threads)ASUS PRIME Z390-A (0602 BIOS)Intel Cannon Lake PCH Shared SRAM16384MB2000GB SABRENT + Samsung SSD 970 EVO 250GBNVIDIA GeForce GTX 1070 8GB (1506/4006MHz)Realtek ALC1220Acer B286HKIntel ConnectionUbuntu 18.044.19.5-041905-generic (x86_64)GNOME Shell 3.28.3X Server 1.19.6NVIDIA 415.224.6.0OpenCL 1.2 CUDA 10.0.1321.1.84GCC 7.3.0 + CUDA 10.0ext43840x2160Zotac NVIDIA GeForce GTX 1070 Ti 8GB (1607/4006MHz)NVIDIA GeForce GTX 1080 8GB (1607/5005MHz)NVIDIA GeForce GTX 1080 Ti 11GB (1480/5508MHz)eVGA NVIDIA GeForce RTX 2070 8GB (1410/7000MHz)Zotac NVIDIA GeForce RTX 2080 8GB (1515/7000MHz)NVIDIA GeForce RTX 2080 Ti 11GB (1350/7000MHz)Sapphire AMD Radeon R9 FURY / NANO 4GB (1000/500MHz)4.5 Mesa 19.0.0-devel padoka PPA (LLVM 8.0.0)OpenCL 2.1 AMD-APP (2679.0)1.1.70GCC 7.3.0AMD Radeon RX Vega 8GB (1590/800MHz)AMD Radeon RX Vega 8GB (1630/945MHz)OpenBenchmarking.orgCompiler Details- --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --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: intel_pstate performanceOpenCL Details- GTX 1070: GPU Compute Cores: 1920- GTX 1070 Ti: GPU Compute Cores: 2432- GTX 1080: GPU Compute Cores: 2560- GTX 1080 Ti: GPU Compute Cores: 3584- RTX 2070: GPU Compute Cores: 2304- RTX 2080: GPU Compute Cores: 2944- RTX 2080 Ti: GPU Compute Cores: 4352Security Details- __user pointer sanitization + Full generic retpoline IBPB IBRS_FW + SSB disabled via prctl and seccomp

NVIDIA AMD Linux GPU Compute December 2018luxmark: GPU - Luxball HDRngc-tensorflow: Inception v4, FP16v-ray: CUDA GPUngc-tensorflow: Googlenet, FP16ngc-tensorflow: ResNet-50, FP32ngc-tensorflow: VGG-16, FP32ngc-tensorflow: ResNet-50, FP16ngc-tensorflow: VGG-16, FP16clpeak: Integer Compute INTclpeak: Kernel Latencyngc-tensorflow: AlexNet, FP32cuda-mini-nbody: Originalngc-tensorflow: AlexNet, FP16shoc: OpenCL - Texture Read Bandwidthcl-mem: Copyshoc: OpenCL - FFT SPshoc: OpenCL - MD5 HashGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 641728844.7790.4337512571.5716076.9316923.74151691.99156345118745210.721688649.2386.3041313376.4017584.4320823.661591112168343318849711.631382355.03102.0745814382.9019393.1324373.721764111187553020957514.172156274.6366.41629210119.50271131.3733213.802539186267459531797219.963009166.0230980073.522174244110133099819.2929641102.7772.07738205108.57335153.33100593.48241930431531119328108324.3742693135.9056.091015285152.40449206.97143853.57334442644321134454144335.862344814305.682502058469.203117919916.9838420392614.033281524916.94442221107016.53OpenBenchmarking.org

LuxMark

OpenCL Device: GPU - Scene: Luxball HDR

OpenBenchmarking.orgScore, More Is BetterLuxMark 3.1OpenCL Device: GPU - Scene: Luxball HDRGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 649K18K27K36K45KSE +/- 0.58, N = 3SE +/- 0.67, N = 3SE +/- 29.04, N = 3SE +/- 119.86, N = 3SE +/- 2.73, N = 3SE +/- 66.64, N = 3SE +/- 36.02, N = 3SE +/- 20.53, N = 3SE +/- 570.76, N = 317288168861382321562300912964142693234483117932815

NVIDIA GPU Cloud TensorFlow

Test: Inception v4, FP16

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: Inception v4, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti306090120150SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.07, N = 3SE +/- 0.07, N = 3SE +/- 0.10, N = 344.7749.2355.0374.63102.77135.90

Chaos Group V-RAY

Mode: CUDA GPU

OpenBenchmarking.orgSeconds, Fewer Is BetterChaos Group V-RAY 1.1.0Mode: CUDA GPUGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti20406080100SE +/- 0.03, N = 3SE +/- 0.04, N = 3SE +/- 3.08, N = 3SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.04, N = 3SE +/- 3.00, N = 390.4386.30102.0766.4166.0272.0756.09

NVIDIA GPU Cloud TensorFlow

Test: Googlenet, FP16

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: Googlenet, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti2004006008001000SE +/- 0.35, N = 3SE +/- 0.09, N = 3SE +/- 0.21, N = 3SE +/- 2.49, N = 3SE +/- 0.58, N = 3SE +/- 0.49, N = 33754134586297381015

NVIDIA GPU Cloud TensorFlow

Test: ResNet-50, FP32

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: ResNet-50, FP32GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti60120180240300SE +/- 0.09, N = 3SE +/- 0.06, N = 3SE +/- 0.07, N = 3SE +/- 0.19, N = 3SE +/- 0.28, N = 3SE +/- 0.44, N = 3125133143210205285

NVIDIA GPU Cloud TensorFlow

Test: VGG-16, FP32

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: VGG-16, FP32GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti306090120150SE +/- 0.13, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.12, N = 3SE +/- 0.07, N = 3SE +/- 0.12, N = 371.5776.4082.90119.50108.57152.40

NVIDIA GPU Cloud TensorFlow

Test: ResNet-50, FP16

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: ResNet-50, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 TiRTX 2070100200300400500SE +/- 0.03, N = 3SE +/- 0.07, N = 3SE +/- 0.10, N = 3SE +/- 0.15, N = 3SE +/- 0.85, N = 3SE +/- 0.28, N = 3160175193271335449309

NVIDIA GPU Cloud TensorFlow

Test: VGG-16, FP16

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: VGG-16, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti50100150200250SE +/- 0.07, N = 3SE +/- 0.03, N = 3SE +/- 0.24, N = 3SE +/- 0.12, N = 3SE +/- 0.09, N = 3SE +/- 0.07, N = 376.9384.4393.13131.37153.33206.97

clpeak

OpenCL Test: Integer Compute INT

OpenBenchmarking.orgGIOPS, More Is BetterclpeakOpenCL Test: Integer Compute INTGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 643K6K9K12K15KSE +/- 9.42, N = 3SE +/- 0.67, N = 3SE +/- 8.46, N = 3SE +/- 12.10, N = 3SE +/- 552.67, N = 3SE +/- 601.95, N = 3SE +/- 946.18, N = 3SE +/- 0.00, N = 3SE +/- 1.96, N = 3SE +/- 1.38, N = 3169220822437332180071005914385143019912491

clpeak

OpenCL Test: Kernel Latency

OpenBenchmarking.orgus, Fewer Is BetterclpeakOpenCL Test: Kernel LatencyGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 64246810SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.05, N = 3SE +/- 0.06, N = 3SE +/- 0.05, N = 3SE +/- 0.12, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 33.743.663.723.803.523.483.575.686.986.94

NVIDIA GPU Cloud TensorFlow

Test: AlexNet, FP32

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: AlexNet, FP32GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti7001400210028003500SE +/- 0.44, N = 3SE +/- 0.32, N = 3SE +/- 1.62, N = 3SE +/- 1.83, N = 3SE +/- 1.84, N = 3SE +/- 6.99, N = 31516159117642539217424193344

CUDA Mini-Nbody

Test: Original

OpenBenchmarking.org(NBody^2)/s, More Is BetterCUDA Mini-Nbody 2015-11-10Test: OriginalGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti90180270360450SE +/- 0.16, N = 3SE +/- 0.07, N = 3SE +/- 0.17, N = 3SE +/- 0.19, N = 3SE +/- 0.61, N = 3SE +/- 0.92, N = 3SE +/- 0.80, N = 391.99112.00111.00186.00244.00304.00426.00

NVIDIA GPU Cloud TensorFlow

Test: AlexNet, FP16

OpenBenchmarking.orgImages Per Second, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: AlexNet, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti10002000300040005000SE +/- 2.26, N = 3SE +/- 0.50, N = 3SE +/- 2.60, N = 3SE +/- 1.32, N = 3SE +/- 3.35, N = 3SE +/- 2.27, N = 3156316831875267431534432

SHOC Scalable HeterOgeneous Computing

Target: OpenCL - Benchmark: Texture Read Bandwidth

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: OpenCL - Benchmark: Texture Read BandwidthGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 642004006008001000SE +/- 0.06, N = 3SE +/- 0.09, N = 3SE +/- 0.39, N = 3SE +/- 2.41, N = 3SE +/- 1.33, N = 3SE +/- 3.18, N = 3SE +/- 3.81, N = 3SE +/- 1.14, N = 3SE +/- 0.54, N = 3SE +/- 1.72, N = 3451433530595110111191134250384442-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-lSHOCCommonOpenCL -lOpenCL-lSHOCCommonOpenCL -lOpenCL-lSHOCCommonOpenCL -lOpenCL1. (CXX) g++ options: -O2 -lSHOCCommon -lrt

cl-mem

Benchmark: Copy

OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: CopyGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 64100200300400500SE +/- 0.07, N = 3SE +/- 0.07, N = 3SE +/- 0.07, N = 3SE +/- 0.07, N = 3SE +/- 0.12, N = 3SE +/- 0.09, N = 3SE +/- 0.39, N = 3SE +/- 0.47, N = 3SE +/- 0.21, N = 3SE +/- 0.07, N = 31871882093173303284542052032211. (CC) gcc options: -O2 -flto -lOpenCL

SHOC Scalable HeterOgeneous Computing

Target: OpenCL - Benchmark: FFT SP

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: OpenCL - Benchmark: FFT SPGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 6430060090012001500SE +/- 0.94, N = 3SE +/- 0.73, N = 3SE +/- 2.68, N = 3SE +/- 1.23, N = 3SE +/- 40.79, N = 3SE +/- 5.95, N = 3SE +/- 14.24, N = 3SE +/- 0.09, N = 3SE +/- 1.57, N = 3SE +/- 0.64, N = 3452497575972998108314438469261070-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-lSHOCCommonOpenCL -lOpenCL-lSHOCCommonOpenCL -lOpenCL-lSHOCCommonOpenCL -lOpenCL1. (CXX) g++ options: -O2 -lSHOCCommon -lrt

SHOC Scalable HeterOgeneous Computing

Target: OpenCL - Benchmark: MD5 Hash

OpenBenchmarking.orgGHash/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: OpenCL - Benchmark: MD5 HashGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 64816243240SE +/- 0.02, N = 3SE +/- 0.00, N = 3SE +/- 0.02, N = 3SE +/- 0.06, N = 3SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.34, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 310.7211.6314.1719.9619.2924.3735.869.2014.0316.53-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-std=c++14 -lcudadevrt -lcudart_static -lpthread -ldl -lcufft-lSHOCCommonOpenCL -lOpenCL-lSHOCCommonOpenCL -lOpenCL-lSHOCCommonOpenCL -lOpenCL1. (CXX) g++ options: -O2 -lSHOCCommon -lrt

clpeak

Performance / Cost - OpenCL Test: Integer Compute INT

OpenBenchmarking.orgGIOPS Per Dollar, More Is BetterclpeakPerformance / Cost - OpenCL Test: Integer Compute INTGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiRX Vega 56RX Vega 6436912154.244.644.444.7513.3712.6112.004.875.241. GTX 1070: $399 reported cost.2. GTX 1070 Ti: $449 reported cost.3. GTX 1080: $549 reported cost.4. GTX 1080 Ti: $699 reported cost.5. RTX 2070: $599 reported cost.6. RTX 2080: $798 reported cost.7. RTX 2080 Ti: $1199 reported cost.8. RX Vega 56: $409 reported cost.9. RX Vega 64: $475 reported cost.

clpeak

Performance / Cost - OpenCL Test: Kernel Latency

OpenBenchmarking.orgus x Dollar, Fewer Is BetterclpeakPerformance / Cost - OpenCL Test: Kernel LatencyGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiRX Vega 56RX Vega 6490018002700360045001492.261643.342042.282656.202108.482777.044280.432854.823296.501. GTX 1070: $399 reported cost.2. GTX 1070 Ti: $449 reported cost.3. GTX 1080: $549 reported cost.4. GTX 1080 Ti: $699 reported cost.5. RTX 2070: $599 reported cost.6. RTX 2080: $798 reported cost.7. RTX 2080 Ti: $1199 reported cost.8. RX Vega 56: $409 reported cost.9. RX Vega 64: $475 reported cost.

LuxMark

Performance / Cost - OpenCL Device: GPU - Scene: Luxball HDR

OpenBenchmarking.orgScore Per Dollar, More Is BetterLuxMark 3.1Performance / Cost - OpenCL Device: GPU - Scene: Luxball HDRGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiRX Vega 56RX Vega 642040608010043.3337.6125.1830.8550.2437.1435.6176.2369.081. GTX 1070: $399 reported cost.2. GTX 1070 Ti: $449 reported cost.3. GTX 1080: $549 reported cost.4. GTX 1080 Ti: $699 reported cost.5. RTX 2070: $599 reported cost.6. RTX 2080: $798 reported cost.7. RTX 2080 Ti: $1199 reported cost.8. RX Vega 56: $409 reported cost.9. RX Vega 64: $475 reported cost.

cl-mem

Performance / Cost - Benchmark: Copy

OpenBenchmarking.orgGB/s Per Dollar, More Is Bettercl-mem 2017-01-13Performance / Cost - Benchmark: CopyGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiRX Vega 56RX Vega 640.12380.24760.37140.49520.6190.470.420.380.450.550.410.380.500.471. GTX 1070: $399 reported cost.2. GTX 1070 Ti: $449 reported cost.3. GTX 1080: $549 reported cost.4. GTX 1080 Ti: $699 reported cost.5. RTX 2070: $599 reported cost.6. RTX 2080: $798 reported cost.7. RTX 2080 Ti: $1199 reported cost.8. RX Vega 56: $409 reported cost.9. RX Vega 64: $475 reported cost.

SHOC Scalable HeterOgeneous Computing

Performance / Cost - Target: OpenCL - Benchmark: Texture Read Bandwidth

OpenBenchmarking.orgGB/s Per Dollar, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Performance / Cost - Target: OpenCL - Benchmark: Texture Read BandwidthGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiRX Vega 56RX Vega 640.4140.8281.2421.6562.071.130.960.970.851.841.400.950.940.931. GTX 1070: $399 reported cost.2. GTX 1070 Ti: $449 reported cost.3. GTX 1080: $549 reported cost.4. GTX 1080 Ti: $699 reported cost.5. RTX 2070: $599 reported cost.6. RTX 2080: $798 reported cost.7. RTX 2080 Ti: $1199 reported cost.8. RX Vega 56: $409 reported cost.9. RX Vega 64: $475 reported cost.

SHOC Scalable HeterOgeneous Computing

Performance / Cost - Target: OpenCL - Benchmark: MD5 Hash

OpenBenchmarking.orgGHash/s Per Dollar, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Performance / Cost - Target: OpenCL - Benchmark: MD5 HashGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiRX Vega 56RX Vega 640.00680.01360.02040.02720.0340.030.030.030.030.030.030.030.030.031. GTX 1070: $399 reported cost.2. GTX 1070 Ti: $449 reported cost.3. GTX 1080: $549 reported cost.4. GTX 1080 Ti: $699 reported cost.5. RTX 2070: $599 reported cost.6. RTX 2080: $798 reported cost.7. RTX 2080 Ti: $1199 reported cost.8. RX Vega 56: $409 reported cost.9. RX Vega 64: $475 reported cost.

SHOC Scalable HeterOgeneous Computing

Performance / Cost - Target: OpenCL - Benchmark: FFT SP

OpenBenchmarking.orgGFLOPS Per Dollar, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Performance / Cost - Target: OpenCL - Benchmark: FFT SPGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiRX Vega 56RX Vega 640.50851.0171.52552.0342.54251.131.111.051.391.671.361.202.262.251. GTX 1070: $399 reported cost.2. GTX 1070 Ti: $449 reported cost.3. GTX 1080: $549 reported cost.4. GTX 1080 Ti: $699 reported cost.5. RTX 2070: $599 reported cost.6. RTX 2080: $798 reported cost.7. RTX 2080 Ti: $1199 reported cost.8. RX Vega 56: $409 reported cost.9. RX Vega 64: $475 reported cost.

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti60120180240300Min: 54.1 / Avg: 170.77 / Max: 203.5Min: 58.5 / Avg: 162.02 / Max: 191Min: 46.3 / Avg: 198.19 / Max: 236.2Min: 51.2 / Avg: 241.65 / Max: 309.7Min: 55.7 / Avg: 239.26 / Max: 287.8Min: 50.2 / Avg: 270.5 / Max: 342.3

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti1530456075Min: 58 / Avg: 66.1 / Max: 70Min: 46 / Avg: 52.21 / Max: 55Min: 60 / Avg: 69.33 / Max: 74Min: 61 / Avg: 70.55 / Max: 77Min: 61 / Avg: 74.1 / Max: 80Min: 56 / Avg: 65.52 / Max: 71

NVIDIA GPU Cloud TensorFlow

Test: VGG-16, FP32

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: VGG-16, FP32GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti0.1260.2520.3780.5040.630.420.470.420.490.450.56

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti60120180240300Min: 43.7 / Avg: 168.53 / Max: 201.8Min: 46.8 / Avg: 160.44 / Max: 192Min: 90.8 / Avg: 194.86 / Max: 239.8Min: 51.7 / Avg: 242.51 / Max: 313.1Min: 48.1 / Avg: 223.76 / Max: 283.3Min: 51 / Avg: 219.66 / Max: 332.2

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti1530456075Min: 59 / Avg: 66.13 / Max: 70Min: 45 / Avg: 52.18 / Max: 55Min: 50 / Avg: 65.84 / Max: 74Min: 61 / Avg: 70.1 / Max: 76Min: 59 / Avg: 69.86 / Max: 77Min: 54 / Avg: 61.56 / Max: 68

NVIDIA GPU Cloud TensorFlow

Test: VGG-16, FP16

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: VGG-16, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti0.21150.4230.63450.8461.05750.460.530.480.540.690.94

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti60120180240300Min: 44.1 / Avg: 157.87 / Max: 210.2Min: 47.2 / Avg: 146.22 / Max: 188.6Min: 71.3 / Avg: 185.89 / Max: 249.7Min: 52.2 / Avg: 211.84 / Max: 320.9Min: 47.6 / Avg: 177.11 / Max: 285.3Min: 50 / Avg: 183.11 / Max: 341.1

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti1530456075Min: 55 / Avg: 64.07 / Max: 72Min: 43 / Avg: 49.56 / Max: 55Min: 51 / Avg: 64.33 / Max: 76Min: 56 / Avg: 66.24 / Max: 77Min: 57 / Avg: 64.7 / Max: 74Min: 53 / Avg: 57.76 / Max: 65

NVIDIA GPU Cloud TensorFlow

Test: Inception v4, FP16

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: Inception v4, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti0.16650.3330.49950.6660.83250.280.340.300.350.580.74

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti60120180240300Min: 43.4 / Avg: 181.99 / Max: 210.5Min: 46.8 / Avg: 163.6 / Max: 187.8Min: 76.2 / Avg: 215.22 / Max: 247Min: 64 / Avg: 257.4 / Max: 319.4Min: 47.7 / Avg: 229.8 / Max: 284.5Min: 49.5 / Avg: 249.29 / Max: 343.1

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti1530456075Min: 56 / Avg: 69 / Max: 74Min: 43 / Avg: 52.93 / Max: 56Min: 58 / Avg: 72.23 / Max: 79Min: 60 / Avg: 72.84 / Max: 80Min: 60 / Avg: 73.17 / Max: 80Min: 55 / Avg: 64.78 / Max: 71

NVIDIA GPU Cloud TensorFlow

Test: Googlenet, FP16

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: Googlenet, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti0.91581.83162.74743.66324.5792.062.532.132.453.214.07

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti60120180240300Min: 87.1 / Avg: 157.65 / Max: 201.5Min: 48.2 / Avg: 144.76 / Max: 185.9Min: 100.9 / Avg: 178.22 / Max: 229Min: 126 / Avg: 208.21 / Max: 307.4Min: 64.2 / Avg: 185.45 / Max: 243.7Min: 48 / Avg: 211.46 / Max: 284.1Min: 48.9 / Avg: 257.45 / Max: 338.7

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti1530456075Min: 61 / Avg: 64.2 / Max: 68Min: 44 / Avg: 49.07 / Max: 52Min: 63 / Avg: 66.8 / Max: 71Min: 64 / Avg: 67.58 / Max: 71Min: 46 / Avg: 55.88 / Max: 61Min: 59 / Avg: 69.81 / Max: 76Min: 53 / Avg: 61.36 / Max: 67

NVIDIA GPU Cloud TensorFlow

Test: AlexNet, FP32

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: AlexNet, FP32GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti36912159.6110.999.9012.1911.7211.4412.99

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti60120180240300Min: 68.9 / Avg: 158.49 / Max: 192.3Min: 46.9 / Avg: 135.79 / Max: 175Min: 44.1 / Avg: 167.55 / Max: 234.8Min: 51.3 / Avg: 213.37 / Max: 303.2Min: 47.3 / Avg: 190.7 / Max: 270.1Min: 49.3 / Avg: 177.62 / Max: 332.6

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti1428425670Min: 59 / Avg: 64.06 / Max: 67Min: 46 / Avg: 49.93 / Max: 53Min: 62 / Avg: 66.46 / Max: 70Min: 63 / Avg: 67.2 / Max: 70Min: 63 / Avg: 67.57 / Max: 72Min: 57 / Avg: 60.36 / Max: 64

NVIDIA GPU Cloud TensorFlow

Test: AlexNet, FP16

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: AlexNet, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti6121824309.8612.3911.1912.5316.5324.95

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti60120180240300Min: 45.9 / Avg: 172.02 / Max: 213.8Min: 48 / Avg: 157.29 / Max: 187.8Min: 44.1 / Avg: 196.38 / Max: 249.7Min: 79.7 / Avg: 234.81 / Max: 318.6Min: 48 / Avg: 206.72 / Max: 285Min: 50.5 / Avg: 248.25 / Max: 346.7

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti1530456075Min: 57 / Avg: 65.61 / Max: 71Min: 44 / Avg: 50.85 / Max: 54Min: 60 / Avg: 68.77 / Max: 75Min: 62 / Avg: 70.04 / Max: 77Min: 59 / Avg: 70.41 / Max: 77Min: 54 / Avg: 61.16 / Max: 67

NVIDIA GPU Cloud TensorFlow

Test: ResNet-50, FP32

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: ResNet-50, FP32GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 Ti0.25880.51760.77641.03521.2940.730.840.730.890.991.15

NVIDIA GPU Cloud TensorFlow

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 TiRTX 207060120180240300Min: 42.7 / Avg: 174.04 / Max: 212.6Min: 47 / Avg: 153.76 / Max: 192.3Min: 44.7 / Avg: 189.85 / Max: 248.8Min: 54.6 / Avg: 222.3 / Max: 319.6Min: 47.3 / Avg: 179.59 / Max: 284.7Min: 56.1 / Avg: 213.62 / Max: 342.3Min: 71.1 / Avg: 162.36 / Max: 246.8

NVIDIA GPU Cloud TensorFlow

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterNVIDIA GPU Cloud TensorFlow 18.09GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 TiRTX 20701428425670Min: 58 / Avg: 65.55 / Max: 70Min: 46 / Avg: 50.94 / Max: 55Min: 61 / Avg: 68.3 / Max: 75Min: 61 / Avg: 68.17 / Max: 75Min: 60 / Avg: 67.05 / Max: 75Min: 55 / Avg: 59.52 / Max: 65Min: 32 / Avg: 45.47 / Max: 58

NVIDIA GPU Cloud TensorFlow

Test: ResNet-50, FP16

OpenBenchmarking.orgImages Per Second Per Watt, More Is BetterNVIDIA GPU Cloud TensorFlow 18.09Test: ResNet-50, FP16GTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2080RTX 2080 TiRTX 20700.47250.9451.41751.892.36250.921.141.021.221.862.101.90

CUDA Mini-Nbody

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterCUDA Mini-Nbody 2015-11-10System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti60120180240300Min: 43.7 / Avg: 194.09 / Max: 210.9Min: 47 / Avg: 185.31 / Max: 213.8Min: 43.7 / Avg: 223.02 / Max: 251.7Min: 50.5 / Avg: 249.72 / Max: 312.9Min: 54.3 / Avg: 215.38 / Max: 246.1Min: 48.7 / Avg: 233.61 / Max: 284Min: 49.1 / Avg: 234.64 / Max: 336

CUDA Mini-Nbody

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterCUDA Mini-Nbody 2015-11-10GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti1530456075Min: 61 / Avg: 70.86 / Max: 75Min: 45 / Avg: 56.11 / Max: 60Min: 64 / Avg: 73.06 / Max: 79Min: 65 / Avg: 74.42 / Max: 80Min: 55 / Avg: 60.44 / Max: 64Min: 65 / Avg: 73.5 / Max: 78Min: 58 / Avg: 64 / Max: 67

Chaos Group V-RAY

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterChaos Group V-RAY 1.1.0System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti50100150200250Min: 43.4 / Avg: 151.17 / Max: 164.2Min: 46.1 / Avg: 130.78 / Max: 135.6Min: 43.6 / Avg: 153.26 / Max: 165.9Min: 120.4 / Avg: 218.99 / Max: 231.8Min: 53.7 / Avg: 185.77 / Max: 199.3Min: 63 / Avg: 191.71 / Max: 202.8Min: 69.5 / Avg: 242.72 / Max: 272

Chaos Group V-RAY

GPU Temperature Monitor

OpenBenchmarking.orgCelsius, Fewer Is BetterChaos Group V-RAY 1.1.0GPU Temperature MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 Ti1428425670Min: 46 / Avg: 62.55 / Max: 68Min: 38 / Avg: 47.77 / Max: 50Min: 48 / Avg: 62.93 / Max: 67Min: 54 / Avg: 68.11 / Max: 73Min: 53 / Avg: 60.72 / Max: 62Min: 50 / Avg: 68.2 / Max: 74Min: 46 / Avg: 60.03 / Max: 68

System Power Consumption Monitor

Phoronix Test Suite System Monitoring

OpenBenchmarking.orgWattsSystem Power Consumption MonitorPhoronix Test Suite System MonitoringGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 6470140210280350Min: 41.9 / Avg: 143.92 / Max: 213.8Min: 45.9 / Avg: 127.84 / Max: 213.8Min: 42.2 / Avg: 154.56 / Max: 251.7Min: 43.4 / Avg: 200.91 / Max: 320.9Min: 42.1 / Avg: 146.13 / Max: 247.2Min: 43.9 / Avg: 175.16 / Max: 287.8Min: 45.5 / Avg: 208.03 / Max: 346.7Min: 79.9 / Avg: 160.12 / Max: 365.3Min: 48.8 / Avg: 143.52 / Max: 284.1Min: 48.4 / Avg: 169.45 / Max: 358.6

GPU Temperature Monitor

Phoronix Test Suite System Monitoring

OpenBenchmarking.orgCelsiusGPU Temperature MonitorPhoronix Test Suite System MonitoringGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 641632486480Min: 31 / Avg: 61.13 / Max: 75Min: 36 / Avg: 46.92 / Max: 60Min: 45 / Avg: 63.06 / Max: 79Min: 33 / Avg: 66.46 / Max: 80Min: 33 / Avg: 54.57 / Max: 64Min: 35 / Avg: 66.5 / Max: 80Min: 34 / Avg: 60.3 / Max: 76Min: 33 / Avg: 66.96 / Max: 79Min: 31 / Avg: 51.02 / Max: 75Min: 28 / Avg: 52.98 / Max: 85

LuxMark

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterLuxMark 3.1System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 6460120180240300Min: 83.1 / Avg: 176.9 / Max: 181.3Min: 146.8 / Avg: 148.92 / Max: 149.4Min: 90.8 / Avg: 174.24 / Max: 176.2Min: 122.9 / Avg: 241.59 / Max: 247.8Min: 78 / Avg: 225.33 / Max: 232.6Min: 101.9 / Avg: 240.83 / Max: 247.3Min: 111.1 / Avg: 315.29 / Max: 328.4Min: 90.4 / Avg: 249.15 / Max: 256.6Min: 51.8 / Avg: 251.16 / Max: 263.3Min: 53 / Avg: 323.35 / Max: 337.9

cl-mem

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is Bettercl-mem 2017-01-13System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 6460120180240300Min: 82.6 / Avg: 137.43 / Max: 149.2Min: 46.1 / Avg: 105.92 / Max: 138.1Min: 115.3 / Avg: 149.23 / Max: 161.8Min: 120.7 / Avg: 179.53 / Max: 209Min: 43.3 / Avg: 85.23 / Max: 108.2Min: 46.4 / Avg: 92.77 / Max: 122Min: 47.1 / Avg: 180.3 / Max: 248.8Min: 90.3 / Avg: 165.77 / Max: 211.3Min: 49.6 / Avg: 196.84 / Max: 261.2Min: 52.4 / Avg: 216.5 / Max: 318.8

cl-mem

Benchmark: Copy

OpenBenchmarking.orgGB/s Per Watt, More Is Bettercl-mem 2017-01-13Benchmark: CopyGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 640.8731.7462.6193.4924.3651.361.771.401.773.883.532.521.241.031.02

SHOC Scalable HeterOgeneous Computing

System Power Consumption Monitor

OpenBenchmarking.orgWatts, Fewer Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10System Power Consumption MonitorGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 6450100150200250Min: 82.6 / Avg: 134.72 / Max: 155.3Min: 47 / Avg: 122.6 / Max: 141.5Min: 89.7 / Avg: 144.76 / Max: 164.6Min: 120.8 / Avg: 190.06 / Max: 216.8Min: 53.7 / Avg: 162.86 / Max: 198.7Min: 47.1 / Avg: 169.8 / Max: 202.9Min: 47.5 / Avg: 222.96 / Max: 282.7Min: 113.9 / Avg: 172.33 / Max: 220.8Min: 49.4 / Avg: 197.63 / Max: 234.4Min: 51.8 / Avg: 223.18 / Max: 249.9

SHOC Scalable HeterOgeneous Computing

Target: OpenCL - Benchmark: Texture Read Bandwidth

OpenBenchmarking.orgGB/s Per Watt, More Is BetterSHOC Scalable HeterOgeneous Computing 2015-11-10Target: OpenCL - Benchmark: Texture Read BandwidthGTX 1070GTX 1070 TiGTX 1080GTX 1080 TiRTX 2070RTX 2080RTX 2080 TiR9 FuryRX Vega 56RX Vega 642468103.353.533.663.136.766.595.081.451.941.98

Phoronix Test Suite v10.8.5