RTX 30 Compute September AMD Ryzen 9 5900X 12-Core testing with a ASUS ROG CROSSHAIR VIII HERO (3801 BIOS) and NVIDIA GeForce RTX 3090 24GB on Ubuntu 21.10 via the Phoronix Test Suite.
HTML result view exported from: https://openbenchmarking.org/result/2109061-TJ-RTX30COMP69&export=pdf&grs&rdt&rro .
RTX 30 Compute September Processor Motherboard Chipset Memory Disk Graphics Audio Monitor Network OS Kernel Desktop Display Server Display Driver OpenGL OpenCL Vulkan Compiler File-System Screen Resolution RTX 3070 RTX 3090 GeForce RTX 3090 AMD Ryzen 9 5900X 12-Core @ 3.70GHz (12 Cores / 24 Threads) ASUS ROG CROSSHAIR VIII HERO (3801 BIOS) AMD Starship/Matisse 16GB 1000GB Western Digital WDS100T1X0E-00AFY0 + 1000GB Western Digital WD_BLACK SN850 1TB + 2000GB NVIDIA GeForce RTX 3070 8GB NVIDIA GA104 HD Audio ASUS VP28U Realtek RTL8125 2.5GbE + Intel I211 Ubuntu 21.10 5.13.0-14-generic (x86_64) GNOME Shell 40.2 X Server 1.20.11 NVIDIA 470.63.01 4.6.0 OpenCL 3.0 CUDA 11.4.112 1.2.175 GCC 11.2.0 ext4 3840x2160 NVIDIA GeForce RTX 3090 24GB NVIDIA GA102 HD Audio OpenBenchmarking.org Kernel Details - Transparent Huge Pages: madvise Compiler Details - --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-targets=nvptx-none=/build/gcc-11-M6DaQn/gcc-11-11.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-M6DaQn/gcc-11-11.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v Processor Details - Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa201016 OpenCL Details - RTX 3070: GPU Compute Cores: 5888 - RTX 3090: GPU Compute Cores: 10496 - GeForce RTX 3090: GPU Compute Cores: 10496 Python Details - Python 3.9.6 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 IBRS_FW STIBP: always-on RSB filling + srbds: Not affected + tsx_async_abort: Not affected
RTX 30 Compute September shoc: OpenCL - GEMM SGEMM_N cl-mem: Read shoc: OpenCL - FFT SP shoc: OpenCL - S3D cl-mem: Write vkresample: 2x - Single vkpeak: fp32-scalar viennacl: OpenCL BLAS - dAXPY vkpeak: fp16-vec4 vkpeak: fp32-vec4 vkpeak: fp64-scalar vkpeak: int32-scalar vkpeak: fp16-scalar vkresample: 2x - Double vkpeak: int32-vec4 vkpeak: int16-scalar vkpeak: fp64-vec4 viennacl: OpenCL BLAS - dGEMM-TN viennacl: OpenCL BLAS - dGEMM-NT viennacl: OpenCL BLAS - dGEMM-NN blender: Fishy Cat - NVIDIA OptiX luxcorerender: LuxCore Benchmark - GPU shoc: OpenCL - MD5 Hash shoc: OpenCL - Max SP Flops hashcat: SHA1 luxcorerender: DLSC - GPU hashcat: SHA-512 hashcat: MD5 hashcat: TrueCrypt RIPEMD160 + XTS luxcorerender: Danish Mood - GPU hashcat: 7-Zip vkpeak: int16-vec4 viennacl: OpenCL BLAS - dDOT blender: BMW27 - NVIDIA OptiX realsr-ncnn: 4x - Yes octanebench: Total Score financebench: Black-Scholes OpenCL viennacl: OpenCL BLAS - dCOPY redshift: indigobench: OpenCL GPU - Bedroom rodinia: OpenCL Particle Filter blender: Classroom - NVIDIA OptiX blender: BMW27 - CUDA blender: Fishy Cat - CUDA luxcorerender: Rainbow Colors and Prism - GPU luxcorerender: Orange Juice - GPU blender: Classroom - CUDA indigobench: OpenCL GPU - Supercar lczero: OpenCL vkfft: arrayfire: Conjugate Gradient OpenCL realsr-ncnn: 4x - No viennacl: OpenCL BLAS - sAXPY blender: Barbershop - CUDA fahbench: waifu2x-ncnn: 2x - 3 - Yes viennacl: OpenCL BLAS - sCOPY ncnn: Vulkan GPU - vgg16 cl-mem: Copy shoc: OpenCL - Reduction ncnn: Vulkan GPU - yolov4-tiny viennacl: OpenCL BLAS - dGEMV-T viennacl: CPU BLAS - dDOT viennacl: CPU BLAS - sAXPY ncnn: Vulkan GPU - blazeface viennacl: CPU BLAS - sDOT ncnn: Vulkan GPU - resnet18 viennacl: OpenCL BLAS - sDOT viennacl: CPU BLAS - dCOPY ncnn: Vulkan GPU-v3-v3 - mobilenet-v3 viennacl: CPU BLAS - dGEMV-T ncnn: Vulkan GPU - googlenet viennacl: CPU BLAS - dGEMV-N ncnn: Vulkan GPU - mnasnet viennacl: OpenCL BLAS - dGEMV-N ncnn: Vulkan GPU - regnety_400m viennacl: CPU BLAS - sCOPY ncnn: Vulkan GPU - efficientnet-b0 viennacl: CPU BLAS - dAXPY namd-cuda: ATPase Simulation - 327,506 Atoms shoc: OpenCL - Texture Read Bandwidth viennacl: CPU BLAS - dGEMM-TT ncnn: Vulkan GPU - mobilenet viennacl: CPU BLAS - dGEMM-TN ncnn: Vulkan GPU - alexnet ncnn: Vulkan GPU-v2-v2 - mobilenet-v2 ncnn: Vulkan GPU - shufflenet-v2 shoc: OpenCL - Triad viennacl: CPU BLAS - dGEMM-NT ncnn: Vulkan GPU - resnet50 ncnn: Vulkan GPU - squeezenet_ssd viennacl: CPU BLAS - dGEMM-NN shoc: OpenCL - Bus Speed Readback shoc: OpenCL - Bus Speed Download viennacl: OpenCL BLAS - dGEMM-TT waifu2x-ncnn: 2x - 3 - No RTX 3070 RTX 3090 GeForce RTX 3090 3732.87 393.3 1134.82 219.195 382.0 17.529 11589.59 396 22503.16 15319.31 363.75 11571.35 11590.48 216.406 11521.48 7644.32 364.29 337 342 341 33.34 6.22 25.4087 23102.9 13194833333 6.68 1673633333 38925166667 503167 5.06 689667 10106.83 395 15.82 46.790 415.199251 10.457 370 226 12.941 5.920 45.77 29.00 54.69 21.27 6.99 76.06 37.318 37790 31554 2.089 7.856 359 507.39 265.5078 4.112 291 5.43 296.8 326.570 6.57 333 48.6 100.8 0.98 151 1.68 322 24.6 2.14 89.8 4.13 82.8 2.01 222 2.42 66.9 3.17 36.5 0.12555 2130.04 56.4 4.31 58.2 1.90 1.91 1.71 24.6671 53.6 3.74 4.43 53.1 27.1135 26.2837 8174.54 826.6 2349.19 427.935 745 9.254 21074.89 721 40868.96 27597.6 655.42 20836.32 20954 119.824 20743.8 13646.56 655.31 598 602 600 18.97 10.92 44.4267 40344.1 22853700000 11.53 2884600000 66739300000 857500 8.64 1154600 16889.26 663 9.52 28.415 683.595999 6.357 607 139 20.968 3.68 28.66 18.32 34.65 32.09 10.42 51.18 53.364 53653 43295 1.48 5.566 501 373.32 344.0195 3.276 364 4.35 364.3 391.211 7.2 376 43.2 89.6 1.07 136 1.66 351 22.6 2.31 83.8 4.41 77.6 2.14 236 2.52 63 3.34 34.8 0.13155 2228.01 54 4.47 56.1 1.95 1.98 1.77 25.5289 52 3.68 4.37 52.4 27.0785 26.2571 601 8105.73 2348.03 427.078 21146.01 40869.19 27807.75 658.56 20925.87 20953.65 20746.67 13752.19 651.2 44.1942 22798500000 2881300000 66589100000 859400 1161100 17006.61 28.421 44567 5.595 3.275 391.091 25.4496 OpenBenchmarking.org
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: GEMM SGEMM_N OpenBenchmarking.org GFLOPS, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: GEMM SGEMM_N GeForce RTX 3090 RTX 3090 RTX 3070 2K 4K 6K 8K 10K SE +/- 26.69, N = 3 8105.73 8174.54 3732.87 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
cl-mem Benchmark: Read OpenBenchmarking.org GB/s, More Is Better cl-mem 2017-01-13 Benchmark: Read RTX 3090 RTX 3070 200 400 600 800 1000 SE +/- 0.30, N = 3 826.6 393.3 1. (CC) gcc options: -O2 -flto -lOpenCL
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: FFT SP OpenBenchmarking.org GFLOPS, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: FFT SP GeForce RTX 3090 RTX 3090 RTX 3070 500 1000 1500 2000 2500 SE +/- 0.78, N = 3 2348.03 2349.19 1134.82 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: S3D OpenBenchmarking.org GFLOPS, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: S3D GeForce RTX 3090 RTX 3090 RTX 3070 90 180 270 360 450 SE +/- 0.17, N = 3 427.08 427.94 219.20 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
cl-mem Benchmark: Write OpenBenchmarking.org GB/s, More Is Better cl-mem 2017-01-13 Benchmark: Write RTX 3090 RTX 3070 160 320 480 640 800 SE +/- 0.09, N = 3 745.0 382.0 1. (CC) gcc options: -O2 -flto -lOpenCL
VkResample Upscale: 2x - Precision: Single OpenBenchmarking.org ms, Fewer Is Better VkResample 1.0 Upscale: 2x - Precision: Single RTX 3090 RTX 3070 4 8 12 16 20 SE +/- 0.013, N = 3 9.254 17.529 1. (CXX) g++ options: -O3 -pthread
vkpeak fp32-scalar OpenBenchmarking.org GFLOPS, More Is Better vkpeak 20210424 fp32-scalar GeForce RTX 3090 RTX 3090 RTX 3070 5K 10K 15K 20K 25K SE +/- 35.78, N = 3 21146.01 21074.89 11589.59
ViennaCL Test: OpenCL BLAS - dAXPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dAXPY RTX 3090 RTX 3070 160 320 480 640 800 SE +/- 0.00, N = 3 721 396 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
vkpeak fp16-vec4 OpenBenchmarking.org GFLOPS, More Is Better vkpeak 20210424 fp16-vec4 GeForce RTX 3090 RTX 3090 RTX 3070 9K 18K 27K 36K 45K SE +/- 71.70, N = 3 40869.19 40868.96 22503.16
vkpeak fp32-vec4 OpenBenchmarking.org GFLOPS, More Is Better vkpeak 20210424 fp32-vec4 GeForce RTX 3090 RTX 3090 RTX 3070 6K 12K 18K 24K 30K SE +/- 35.81, N = 3 27807.75 27597.60 15319.31
vkpeak fp64-scalar OpenBenchmarking.org GFLOPS, More Is Better vkpeak 20210424 fp64-scalar GeForce RTX 3090 RTX 3090 RTX 3070 140 280 420 560 700 SE +/- 0.36, N = 3 658.56 655.42 363.75
vkpeak int32-scalar OpenBenchmarking.org GIOPS, More Is Better vkpeak 20210424 int32-scalar GeForce RTX 3090 RTX 3090 RTX 3070 4K 8K 12K 16K 20K SE +/- 16.33, N = 3 20925.87 20836.32 11571.35
vkpeak fp16-scalar OpenBenchmarking.org GFLOPS, More Is Better vkpeak 20210424 fp16-scalar GeForce RTX 3090 RTX 3090 RTX 3070 4K 8K 12K 16K 20K SE +/- 26.04, N = 3 20953.65 20954.00 11590.48
VkResample Upscale: 2x - Precision: Double OpenBenchmarking.org ms, Fewer Is Better VkResample 1.0 Upscale: 2x - Precision: Double RTX 3090 RTX 3070 50 100 150 200 250 SE +/- 0.03, N = 3 119.82 216.41 1. (CXX) g++ options: -O3 -pthread
vkpeak int32-vec4 OpenBenchmarking.org GIOPS, More Is Better vkpeak 20210424 int32-vec4 GeForce RTX 3090 RTX 3090 RTX 3070 4K 8K 12K 16K 20K SE +/- 18.17, N = 3 20746.67 20743.80 11521.48
vkpeak int16-scalar OpenBenchmarking.org GIOPS, More Is Better vkpeak 20210424 int16-scalar GeForce RTX 3090 RTX 3090 RTX 3070 3K 6K 9K 12K 15K SE +/- 11.41, N = 3 13752.19 13646.56 7644.32
vkpeak fp64-vec4 OpenBenchmarking.org GFLOPS, More Is Better vkpeak 20210424 fp64-vec4 GeForce RTX 3090 RTX 3090 RTX 3070 140 280 420 560 700 SE +/- 0.42, N = 3 651.20 655.31 364.29
ViennaCL Test: OpenCL BLAS - dGEMM-TN OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dGEMM-TN RTX 3090 RTX 3070 130 260 390 520 650 598 337 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
ViennaCL Test: OpenCL BLAS - dGEMM-NT OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dGEMM-NT RTX 3090 RTX 3070 130 260 390 520 650 SE +/- 1.33, N = 3 602 342 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
ViennaCL Test: OpenCL BLAS - dGEMM-NN OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dGEMM-NN RTX 3090 RTX 3070 130 260 390 520 650 SE +/- 1.33, N = 3 600 341 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
Blender Blend File: Fishy Cat - Compute: NVIDIA OptiX OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.92 Blend File: Fishy Cat - Compute: NVIDIA OptiX RTX 3090 RTX 3070 8 16 24 32 40 SE +/- 0.01, N = 3 18.97 33.34
LuxCoreRender Scene: LuxCore Benchmark - Acceleration: GPU OpenBenchmarking.org M samples/sec, More Is Better LuxCoreRender 2.5 Scene: LuxCore Benchmark - Acceleration: GPU RTX 3090 RTX 3070 3 6 9 12 15 SE +/- 0.01, N = 3 10.92 6.22 MIN: 3.29 / MAX: 13.01 MIN: 2.05 / MAX: 7.38
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: MD5 Hash OpenBenchmarking.org GHash/s, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: MD5 Hash GeForce RTX 3090 RTX 3090 RTX 3070 10 20 30 40 50 SE +/- 0.03, N = 3 44.19 44.43 25.41 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: Max SP Flops OpenBenchmarking.org GFLOPS, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: Max SP Flops RTX 3090 RTX 3070 9K 18K 27K 36K 45K SE +/- 59.20, N = 3 40344.1 23102.9 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
Hashcat Benchmark: SHA1 OpenBenchmarking.org H/s, More Is Better Hashcat 6.1.1 Benchmark: SHA1 GeForce RTX 3090 RTX 3090 RTX 3070 5000M 10000M 15000M 20000M 25000M SE +/- 7750340.49, N = 3 22798500000 22853700000 13194833333
LuxCoreRender Scene: DLSC - Acceleration: GPU OpenBenchmarking.org M samples/sec, More Is Better LuxCoreRender 2.5 Scene: DLSC - Acceleration: GPU RTX 3090 RTX 3070 3 6 9 12 15 SE +/- 0.00, N = 3 11.53 6.68 MIN: 11.29 / MAX: 11.68 MIN: 6.25 / MAX: 6.94
Hashcat Benchmark: SHA-512 OpenBenchmarking.org H/s, More Is Better Hashcat 6.1.1 Benchmark: SHA-512 GeForce RTX 3090 RTX 3090 RTX 3070 600M 1200M 1800M 2400M 3000M SE +/- 735602.55, N = 3 2881300000 2884600000 1673633333
Hashcat Benchmark: MD5 OpenBenchmarking.org H/s, More Is Better Hashcat 6.1.1 Benchmark: MD5 GeForce RTX 3090 RTX 3090 RTX 3070 14000M 28000M 42000M 56000M 70000M SE +/- 41756290.28, N = 3 66589100000 66739300000 38925166667
Hashcat Benchmark: TrueCrypt RIPEMD160 + XTS OpenBenchmarking.org H/s, More Is Better Hashcat 6.1.1 Benchmark: TrueCrypt RIPEMD160 + XTS GeForce RTX 3090 RTX 3090 RTX 3070 200K 400K 600K 800K 1000K SE +/- 800.69, N = 3 859400 857500 503167
LuxCoreRender Scene: Danish Mood - Acceleration: GPU OpenBenchmarking.org M samples/sec, More Is Better LuxCoreRender 2.5 Scene: Danish Mood - Acceleration: GPU RTX 3090 RTX 3070 2 4 6 8 10 SE +/- 0.00, N = 3 8.64 5.06 MIN: 2.95 / MAX: 10.48 MIN: 1.5 / MAX: 6.12
Hashcat Benchmark: 7-Zip OpenBenchmarking.org H/s, More Is Better Hashcat 6.1.1 Benchmark: 7-Zip GeForce RTX 3090 RTX 3090 RTX 3070 200K 400K 600K 800K 1000K SE +/- 1128.91, N = 3 1161100 1154600 689667
vkpeak int16-vec4 OpenBenchmarking.org GIOPS, More Is Better vkpeak 20210424 int16-vec4 GeForce RTX 3090 RTX 3090 RTX 3070 4K 8K 12K 16K 20K SE +/- 20.90, N = 3 17006.61 16889.26 10106.83
ViennaCL Test: OpenCL BLAS - dDOT OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dDOT RTX 3090 RTX 3070 140 280 420 560 700 SE +/- 0.00, N = 3 663 395 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
Blender Blend File: BMW27 - Compute: NVIDIA OptiX OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.92 Blend File: BMW27 - Compute: NVIDIA OptiX RTX 3090 RTX 3070 4 8 12 16 20 SE +/- 0.01, N = 3 9.52 15.82
RealSR-NCNN Scale: 4x - TAA: Yes OpenBenchmarking.org Seconds, Fewer Is Better RealSR-NCNN 20200818 Scale: 4x - TAA: Yes GeForce RTX 3090 RTX 3090 RTX 3070 11 22 33 44 55 SE +/- 0.04, N = 3 28.42 28.42 46.79
OctaneBench Total Score OpenBenchmarking.org Score, More Is Better OctaneBench 2020.1 Total Score RTX 3090 RTX 3070 150 300 450 600 750 683.60 415.20
FinanceBench Benchmark: Black-Scholes OpenCL OpenBenchmarking.org ms, Fewer Is Better FinanceBench 2016-07-25 Benchmark: Black-Scholes OpenCL RTX 3090 RTX 3070 3 6 9 12 15 SE +/- 0.003, N = 3 6.357 10.457 1. (CXX) g++ options: -O3 -march=native -fopenmp
ViennaCL Test: OpenCL BLAS - dCOPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dCOPY RTX 3090 RTX 3070 130 260 390 520 650 SE +/- 0.33, N = 3 607 370 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
RedShift Demo OpenBenchmarking.org Seconds, Fewer Is Better RedShift Demo 3.0 RTX 3090 RTX 3070 50 100 150 200 250 SE +/- 0.33, N = 3 139 226
IndigoBench Acceleration: OpenCL GPU - Scene: Bedroom OpenBenchmarking.org M samples/s, More Is Better IndigoBench 4.4 Acceleration: OpenCL GPU - Scene: Bedroom RTX 3090 RTX 3070 5 10 15 20 25 SE +/- 0.02, N = 3 20.97 12.94
Rodinia Test: OpenCL Particle Filter OpenBenchmarking.org Seconds, Fewer Is Better Rodinia 3.1 Test: OpenCL Particle Filter RTX 3090 RTX 3070 1.332 2.664 3.996 5.328 6.66 SE +/- 0.007, N = 3 3.680 5.920 1. (CXX) g++ options: -O2 -lOpenCL
Blender Blend File: Classroom - Compute: NVIDIA OptiX OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.92 Blend File: Classroom - Compute: NVIDIA OptiX RTX 3090 RTX 3070 10 20 30 40 50 SE +/- 0.02, N = 3 28.66 45.77
Blender Blend File: BMW27 - Compute: CUDA OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.92 Blend File: BMW27 - Compute: CUDA RTX 3090 RTX 3070 7 14 21 28 35 SE +/- 0.04, N = 3 18.32 29.00
Blender Blend File: Fishy Cat - Compute: CUDA OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.92 Blend File: Fishy Cat - Compute: CUDA RTX 3090 RTX 3070 12 24 36 48 60 SE +/- 0.04, N = 3 34.65 54.69
LuxCoreRender Scene: Rainbow Colors and Prism - Acceleration: GPU OpenBenchmarking.org M samples/sec, More Is Better LuxCoreRender 2.5 Scene: Rainbow Colors and Prism - Acceleration: GPU RTX 3090 RTX 3070 7 14 21 28 35 SE +/- 0.03, N = 3 32.09 21.27 MIN: 28.96 / MAX: 34.45 MIN: 20.13 / MAX: 22.99
LuxCoreRender Scene: Orange Juice - Acceleration: GPU OpenBenchmarking.org M samples/sec, More Is Better LuxCoreRender 2.5 Scene: Orange Juice - Acceleration: GPU RTX 3090 RTX 3070 3 6 9 12 15 SE +/- 0.01, N = 3 10.42 6.99 MIN: 8.45 / MAX: 13.57 MIN: 5.61 / MAX: 8.46
Blender Blend File: Classroom - Compute: CUDA OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.92 Blend File: Classroom - Compute: CUDA RTX 3090 RTX 3070 20 40 60 80 100 SE +/- 0.01, N = 3 51.18 76.06
IndigoBench Acceleration: OpenCL GPU - Scene: Supercar OpenBenchmarking.org M samples/s, More Is Better IndigoBench 4.4 Acceleration: OpenCL GPU - Scene: Supercar RTX 3090 RTX 3070 12 24 36 48 60 SE +/- 0.04, N = 3 53.36 37.32
LeelaChessZero Backend: OpenCL OpenBenchmarking.org Nodes Per Second, More Is Better LeelaChessZero 0.28 Backend: OpenCL RTX 3090 RTX 3070 11K 22K 33K 44K 55K SE +/- 178.94, N = 3 53653 37790 1. (CXX) g++ options: -flto -pthread
VkFFT OpenBenchmarking.org Benchmark Score, More Is Better VkFFT 1.1.1 GeForce RTX 3090 RTX 3090 RTX 3070 10K 20K 30K 40K 50K SE +/- 172.10, N = 3 44567 43295 31554 1. (CXX) g++ options: -O3 -pthread
ArrayFire Test: Conjugate Gradient OpenCL OpenBenchmarking.org ms, Fewer Is Better ArrayFire 3.7 Test: Conjugate Gradient OpenCL RTX 3090 RTX 3070 0.47 0.94 1.41 1.88 2.35 SE +/- 0.001, N = 3 1.480 2.089 1. (CXX) g++ options: -rdynamic
RealSR-NCNN Scale: 4x - TAA: No OpenBenchmarking.org Seconds, Fewer Is Better RealSR-NCNN 20200818 Scale: 4x - TAA: No GeForce RTX 3090 RTX 3090 RTX 3070 2 4 6 8 10 SE +/- 0.002, N = 3 5.595 5.566 7.856
ViennaCL Test: OpenCL BLAS - sAXPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - sAXPY RTX 3090 RTX 3070 110 220 330 440 550 SE +/- 0.33, N = 3 501 359 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
Blender Blend File: Barbershop - Compute: CUDA OpenBenchmarking.org Seconds, Fewer Is Better Blender 2.92 Blend File: Barbershop - Compute: CUDA RTX 3090 RTX 3070 110 220 330 440 550 SE +/- 0.17, N = 3 373.32 507.39
FAHBench OpenBenchmarking.org Ns Per Day, More Is Better FAHBench 2.3.2 RTX 3090 RTX 3070 70 140 210 280 350 SE +/- 0.05, N = 3 344.02 265.51
Waifu2x-NCNN Vulkan Scale: 2x - Denoise: 3 - TAA: Yes OpenBenchmarking.org Seconds, Fewer Is Better Waifu2x-NCNN Vulkan 20200818 Scale: 2x - Denoise: 3 - TAA: Yes GeForce RTX 3090 RTX 3090 RTX 3070 0.9252 1.8504 2.7756 3.7008 4.626 SE +/- 0.006, N = 3 3.275 3.276 4.112
ViennaCL Test: OpenCL BLAS - sCOPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - sCOPY RTX 3090 RTX 3070 80 160 240 320 400 SE +/- 0.58, N = 3 364 291 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: vgg16 OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: vgg16 RTX 3090 RTX 3070 1.2218 2.4436 3.6654 4.8872 6.109 SE +/- 0.01, N = 3 4.35 5.43 MIN: 4.02 / MAX: 18.18 MIN: 5.38 / MAX: 9.58 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
cl-mem Benchmark: Copy OpenBenchmarking.org GB/s, More Is Better cl-mem 2017-01-13 Benchmark: Copy RTX 3090 RTX 3070 80 160 240 320 400 SE +/- 0.45, N = 3 364.3 296.8 1. (CC) gcc options: -O2 -flto -lOpenCL
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: Reduction OpenBenchmarking.org GB/s, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: Reduction GeForce RTX 3090 RTX 3090 RTX 3070 80 160 240 320 400 SE +/- 0.70, N = 3 391.09 391.21 326.57 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
NCNN Target: Vulkan GPU - Model: yolov4-tiny OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: yolov4-tiny RTX 3090 RTX 3070 2 4 6 8 10 SE +/- 0.01, N = 3 7.20 6.57 MIN: 6.44 / MAX: 31.2 MIN: 6.33 / MAX: 6.98 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: OpenCL BLAS - dGEMV-T OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dGEMV-T RTX 3090 RTX 3070 80 160 240 320 400 SE +/- 0.67, N = 3 376 333 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
ViennaCL Test: CPU BLAS - dDOT OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dDOT RTX 3090 RTX 3070 11 22 33 44 55 SE +/- 0.50, N = 3 43.2 48.6 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
ViennaCL Test: CPU BLAS - sAXPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - sAXPY RTX 3090 RTX 3070 20 40 60 80 100 SE +/- 0.73, N = 3 89.6 100.8 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: blazeface OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: blazeface RTX 3090 RTX 3070 0.2408 0.4816 0.7224 0.9632 1.204 SE +/- 0.00, N = 3 1.07 0.98 MIN: 0.89 / MAX: 21.68 MIN: 0.95 / MAX: 2.27 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: CPU BLAS - sDOT OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - sDOT RTX 3090 RTX 3070 30 60 90 120 150 SE +/- 0.88, N = 3 136 151 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: resnet18 OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: resnet18 RTX 3090 RTX 3070 0.4073 0.8146 1.2219 1.6292 2.0365 SE +/- 0.01, N = 3 1.66 1.68 MIN: 1.53 / MAX: 9.86 MIN: 1.66 / MAX: 1.97 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: OpenCL BLAS - sDOT OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - sDOT RTX 3090 RTX 3070 80 160 240 320 400 SE +/- 0.67, N = 3 351 322 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
ViennaCL Test: CPU BLAS - dCOPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dCOPY RTX 3090 RTX 3070 6 12 18 24 30 SE +/- 0.03, N = 3 22.6 24.6 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3 OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3 RTX 3090 RTX 3070 0.5198 1.0396 1.5594 2.0792 2.599 SE +/- 0.01, N = 3 2.31 2.14 MIN: 2.09 / MAX: 11.02 MIN: 2.11 / MAX: 3.93 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: CPU BLAS - dGEMV-T OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dGEMV-T RTX 3090 RTX 3070 20 40 60 80 100 SE +/- 0.10, N = 3 83.8 89.8 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: googlenet OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: googlenet RTX 3090 RTX 3070 0.9923 1.9846 2.9769 3.9692 4.9615 SE +/- 0.13, N = 2 4.41 4.13 MIN: 3.6 / MAX: 14.7 MIN: 3.77 / MAX: 13.51 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: CPU BLAS - dGEMV-N OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dGEMV-N RTX 3090 RTX 3070 20 40 60 80 100 SE +/- 0.09, N = 3 77.6 82.8 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: mnasnet OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: mnasnet RTX 3090 RTX 3070 0.4815 0.963 1.4445 1.926 2.4075 SE +/- 0.01, N = 3 2.14 2.01 MIN: 1.94 / MAX: 8.72 MIN: 1.98 / MAX: 4.82 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: OpenCL BLAS - dGEMV-N OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dGEMV-N RTX 3090 RTX 3070 50 100 150 200 250 SE +/- 0.58, N = 3 236 222 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: regnety_400m OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: regnety_400m RTX 3090 RTX 3070 0.5783 1.1566 1.7349 2.3132 2.8915 SE +/- 0.01, N = 3 2.52 2.42 MIN: 2.38 / MAX: 10.9 MIN: 2.4 / MAX: 3.84 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: CPU BLAS - sCOPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - sCOPY RTX 3090 RTX 3070 15 30 45 60 75 SE +/- 0.58, N = 3 63.0 66.9 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: efficientnet-b0 OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: efficientnet-b0 RTX 3090 RTX 3070 0.7538 1.5076 2.2614 3.0152 3.769 SE +/- 0.01, N = 3 3.34 3.17 MIN: 3.09 / MAX: 9.48 MIN: 3.14 / MAX: 5.06 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: CPU BLAS - dAXPY OpenBenchmarking.org GB/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dAXPY RTX 3090 RTX 3070 8 16 24 32 40 SE +/- 0.32, N = 3 34.8 36.5 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NAMD CUDA ATPase Simulation - 327,506 Atoms OpenBenchmarking.org days/ns, Fewer Is Better NAMD CUDA 2.14 ATPase Simulation - 327,506 Atoms RTX 3090 RTX 3070 0.0296 0.0592 0.0888 0.1184 0.148 SE +/- 0.00049, N = 3 0.13155 0.12555
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: Texture Read Bandwidth OpenBenchmarking.org GB/s, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: Texture Read Bandwidth RTX 3090 RTX 3070 500 1000 1500 2000 2500 SE +/- 6.58, N = 3 2228.01 2130.04 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
ViennaCL Test: CPU BLAS - dGEMM-TT OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dGEMM-TT RTX 3090 RTX 3070 13 26 39 52 65 SE +/- 0.09, N = 3 54.0 56.4 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: mobilenet OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: mobilenet RTX 3090 RTX 3070 1.008 2.016 3.024 4.032 5.04 SE +/- 0.01, N = 3 4.47 4.31 MIN: 4.1 / MAX: 24.65 MIN: 4.25 / MAX: 5.94 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: CPU BLAS - dGEMM-TN OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dGEMM-TN RTX 3090 RTX 3070 13 26 39 52 65 SE +/- 0.00, N = 3 56.1 58.2 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: alexnet OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: alexnet RTX 3090 RTX 3070 0.4388 0.8776 1.3164 1.7552 2.194 SE +/- 0.01, N = 3 1.95 1.90 MIN: 1.76 / MAX: 18.4 MIN: 1.86 / MAX: 5.12 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
NCNN Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2 OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2 RTX 3090 RTX 3070 0.4455 0.891 1.3365 1.782 2.2275 SE +/- 0.01, N = 3 1.98 1.91 MIN: 1.84 / MAX: 7.62 MIN: 1.88 / MAX: 3.42 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
NCNN Target: Vulkan GPU - Model: shufflenet-v2 OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: shufflenet-v2 RTX 3090 RTX 3070 0.3983 0.7966 1.1949 1.5932 1.9915 SE +/- 0.01, N = 3 1.77 1.71 MIN: 1.67 / MAX: 9.47 MIN: 1.68 / MAX: 3.55 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: Triad OpenBenchmarking.org GB/s, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: Triad GeForce RTX 3090 RTX 3090 RTX 3070 6 12 18 24 30 SE +/- 0.01, N = 3 25.45 25.53 24.67 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
ViennaCL Test: CPU BLAS - dGEMM-NT OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dGEMM-NT RTX 3090 RTX 3070 12 24 36 48 60 SE +/- 0.12, N = 3 52.0 53.6 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
NCNN Target: Vulkan GPU - Model: resnet50 OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: resnet50 RTX 3090 RTX 3070 0.8415 1.683 2.5245 3.366 4.2075 SE +/- 0.01, N = 3 3.68 3.74 MIN: 3.44 / MAX: 13.64 MIN: 3.71 / MAX: 4.06 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
NCNN Target: Vulkan GPU - Model: squeezenet_ssd OpenBenchmarking.org ms, Fewer Is Better NCNN 20210720 Target: Vulkan GPU - Model: squeezenet_ssd RTX 3090 RTX 3070 0.9968 1.9936 2.9904 3.9872 4.984 SE +/- 0.13, N = 3 4.37 4.43 MIN: 3.93 / MAX: 23.52 MIN: 4.1 / MAX: 7.62 1. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread -pthread
ViennaCL Test: CPU BLAS - dGEMM-NN OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: CPU BLAS - dGEMM-NN RTX 3090 RTX 3070 12 24 36 48 60 SE +/- 0.26, N = 3 52.4 53.1 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: Bus Speed Readback OpenBenchmarking.org GB/s, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: Bus Speed Readback RTX 3090 RTX 3070 6 12 18 24 30 SE +/- 0.01, N = 3 27.08 27.11 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
SHOC Scalable HeterOgeneous Computing Target: OpenCL - Benchmark: Bus Speed Download OpenBenchmarking.org GB/s, More Is Better SHOC Scalable HeterOgeneous Computing 2020-04-17 Target: OpenCL - Benchmark: Bus Speed Download RTX 3090 RTX 3070 6 12 18 24 30 SE +/- 0.02, N = 3 26.26 26.28 1. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -pthread -lmpi_cxx -lmpi
ViennaCL Test: OpenCL BLAS - dGEMM-TT OpenBenchmarking.org GFLOPs/s, More Is Better ViennaCL 1.7.1 Test: OpenCL BLAS - dGEMM-TT RTX 3090 130 260 390 520 650 601 1. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
Phoronix Test Suite v10.8.5