2307020-PTS-GPUREVIEW1

RTX 4080 16GB PNY REVIEW

Compare your own system(s) to this result file with the Phoronix Test Suite by running the command: phoronix-test-suite benchmark 2307077-NE-2307020PT71
Jump To Table - Results

View

Do Not Show Noisy Results
Do Not Show Results With Incomplete Data
Do Not Show Results With Little Change/Spread
List Notable Results

Limit displaying results to tests within:

BLAS (Basic Linear Algebra Sub-Routine) Tests 3 Tests
C++ Boost Tests 2 Tests
CPU Massive 4 Tests
Creator Workloads 4 Tests
HPC - High Performance Computing 7 Tests
Machine Learning 4 Tests
Multi-Core 7 Tests
NVIDIA GPU Compute 25 Tests
OpenCL 6 Tests
OpenMPI Tests 2 Tests
Renderers 3 Tests
Scientific Computing 2 Tests
Server CPU Tests 2 Tests
Vulkan Compute 6 Tests
Common Workstation Benchmarks 2 Tests

Statistics

Show Overall Harmonic Mean(s)
Show Overall Geometric Mean
Show Geometric Means Per-Suite/Category
Show Wins / Losses Counts (Pie Chart)
Normalize Results
Remove Outliers Before Calculating Averages

Graph Settings

Force Line Graphs Where Applicable
Convert To Scalar Where Applicable
Disable Color Branding
Prefer Vertical Bar Graphs
No Box Plots
On Line Graphs With Missing Data, Connect The Line Gaps

Multi-Way Comparison

Condense Multi-Option Tests Into Single Result Graphs
Condense Test Profiles With Multiple Version Results Into Single Result Graphs

Table

Show Detailed System Result Table

Run Management

Highlight
Result
Hide
Result
Result
Identifier
Performance Per
Dollar
Date
Run
  Test
  Duration
RTX 3090 24GB -Zotac
July 02 2023
  11 Hours, 19 Minutes
RTX 2080 Ti 22GB -Dell
July 06 2023
  9 Hours, 18 Minutes
RTX 4090 24GB -Nvidia
July 06 2023
  7 Hours, 45 Minutes
RTX 4080 16GB -Pny
July 07 2023
  8 Hours, 28 Minutes
Invert Hiding All Results Option
  9 Hours, 12 Minutes

Only show results where is faster than
Only show results matching title/arguments (delimit multiple options with a comma):
Do not show results matching title/arguments (delimit multiple options with a comma):


2307020-PTS-GPUREVIEW1OpenBenchmarking.orgPhoronix Test SuiteIntel Xeon w9-3495X @ 4.80GHz (56 Cores / 112 Threads)ASUS Pro WS W790E-SAGE SE (0506 BIOS)Intel Device 7aa78 x 32 GB DDR5-4812MT/s Hynix HMCG88AEBRA115N6401GB Micron_9300_MTFDHAL6T4TDR + 0GB Virtual HDisk0NVIDIA GeForce RTX 3090 24GBNVIDIA GeForce RTX 2080 Ti 22GBNVIDIA GeForce RTX 4090 24GBNVIDIA GeForce RTX 4080 16GBRealtek ALC1220BenQ PD2720U2 x Intel X710 for 10GBASE-TUbuntu 22.046.3.0-060300-generic (x86_64)GNOME Shell 42.5X Server 1.21.1.4NVIDIA 530.41.034.6.0OpenCL 3.0 CUDA 12.1.981.3.236GCC 11.3.0 + CUDA 12.1ext43840x2160ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerDisplay DriverOpenGLOpenCLVulkanCompilerFile-SystemScreen Resolution2307020-PTS-GPUREVIEW1 BenchmarksSystem Logs- Transparent Huge Pages: madvise- --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-aYxV0E/gcc-11-11.3.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-aYxV0E/gcc-11-11.3.0/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 - Scaling Governor: intel_pstate powersave (EPP: performance) - CPU Microcode: 0x2b000390 - RTX 3090 24GB -Zotac: BAR1 / Visible vRAM Size: 32768 MiB - vBIOS Version: 94.02.26.48.65 - RTX 2080 Ti 22GB -Dell: BAR1 / Visible vRAM Size: 256 MiB - vBIOS Version: 90.02.30.40.4d - RTX 4090 24GB -Nvidia: BAR1 / Visible vRAM Size: 32768 MiB - vBIOS Version: 95.02.20.00.03 - RTX 4080 16GB -Pny: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 95.03.0e.00.67 - RTX 3090 24GB -Zotac: GPU Compute Cores: 10496- RTX 2080 Ti 22GB -Dell: GPU Compute Cores: 4352- RTX 4090 24GB -Nvidia: GPU Compute Cores: 16384- RTX 4080 16GB -Pny: GPU Compute Cores: 9728- Python 3.10.6- itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS IBPB: conditional RSB filling PBRSB-eIBRS: SW sequence + srbds: Not affected + tsx_async_abort: Not affected

RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -PnyResult OverviewPhoronix Test Suite100%183%267%350%434%Chaos Group V-RAYOctaneBenchHashcatclpeakFinanceBenchBlendervkpeakGROMACSIndigoBenchSHOC Scalable HeterOgeneous ComputingNCNNVkResampleRealSR-NCNNRodiniaMandelGPUNeatBenchNAMD CUDAArrayFireVkFFTWaifu2x-NCNN Vulkancl-memFAHBenchViennaCLLeelaChessZero

2307020-PTS-GPUREVIEW1ncnn: Vulkan GPU - squeezenet_ssdclpeak: Single-Precision Floatshoc: OpenCL - GEMM SGEMM_Nshoc: OpenCL - Max SP Flopsncnn: Vulkan GPU - googlenetncnn: Vulkan GPU - resnet18v-ray: NVIDIA CUDA GPUncnn: Vulkan GPU - alexnetv-ray: NVIDIA RTX GPUncnn: Vulkan GPU - mobilenetoctanebench: Total Scorevkpeak: fp32-vec4clpeak: Integer Compute INThashcat: 7-Zipblender: Pabellon Barcelona - NVIDIA OptiXhashcat: TrueCrypt RIPEMD160 + XTSindigobench: OpenCL GPU - Bedroomhashcat: SHA-512financebench: Black-Scholes OpenCLblender: Barbershop - NVIDIA OptiXblender: Classroom - NVIDIA OptiXhashcat: SHA1vkpeak: int16-vec4hashcat: MD5blender: Fishy Cat - NVIDIA OptiXviennacl: OpenCL BLAS - dGEMM-TTshoc: OpenCL - MD5 Hashvkpeak: int16-scalarvkpeak: fp16-vec4vkpeak: fp16-scalargromacs: NVIDIA CUDA GPU - water_GMX50_bareviennacl: OpenCL BLAS - dGEMM-TNvkpeak: int32-vec4shoc: OpenCL - Reductionvkpeak: fp64-scalarviennacl: OpenCL BLAS - dGEMM-NTvkpeak: fp32-scalarvkpeak: int32-scalarvkpeak: fp64-vec4clpeak: Double-Precision Doublevkresample: 2x - Doubleshoc: OpenCL - Texture Read Bandwidthrealsr-ncnn: 4x - Yesviennacl: OpenCL BLAS - dGEMM-NNindigobench: OpenCL GPU - Supercarshoc: OpenCL - S3Dmandelgpu: GPUshoc: OpenCL - Bus Speed Readbackshoc: OpenCL - Triadneatbench: GPUshoc: OpenCL - Bus Speed Downloadarrayfire: Conjugate Gradient OpenCLvkresample: 2x - Singleshoc: OpenCL - FFT SPrealsr-ncnn: 4x - Novkfft: cl-mem: Writewaifu2x-ncnn: 2x - 3 - Yesclpeak: Global Memory Bandwidthcl-mem: Readviennacl: OpenCL BLAS - dGEMV-Ncaffe: AlexNet - NVIDIA CUDA - 1000caffe: AlexNet - NVIDIA CUDA - 200viennacl: OpenCL BLAS - dAXPYncnn: Vulkan GPU - yolov4-tinycaffe: AlexNet - NVIDIA CUDA - 100caffe: GoogleNet - NVIDIA CUDA - 1000caffe: GoogleNet - NVIDIA CUDA - 200caffe: GoogleNet - NVIDIA CUDA - 100viennacl: OpenCL BLAS - sDOTviennacl: OpenCL BLAS - sCOPYfahbench: viennacl: OpenCL BLAS - sAXPYviennacl: OpenCL BLAS - dCOPYviennacl: OpenCL BLAS - dDOTcl-mem: Copyncnn: Vulkan GPU - vision_transformerviennacl: OpenCL BLAS - dGEMV-Tlczero: OpenCLviennacl: CPU BLAS - dCOPYviennacl: CPU BLAS - dDOTviennacl: CPU BLAS - dAXPYviennacl: CPU BLAS - sCOPYviennacl: CPU BLAS - sAXPYviennacl: CPU BLAS - dGEMV-Tviennacl: CPU BLAS - sDOTncnn: Vulkan GPU - FastestDetncnn: Vulkan GPU - regnety_400mncnn: Vulkan GPU - resnet50ncnn: Vulkan GPU - vgg16ncnn: Vulkan GPU - blazefacencnn: Vulkan GPU - efficientnet-b0ncnn: Vulkan GPU - mnasnetncnn: Vulkan GPU - shufflenet-v2ncnn: Vulkan GPU-v3-v3 - mobilenet-v3ncnn: Vulkan GPU-v2-v2 - mobilenet-v2blender: BMW27 - NVIDIA OptiXviennacl: CPU BLAS - dGEMM-TTviennacl: CPU BLAS - dGEMM-TNviennacl: CPU BLAS - dGEMM-NTviennacl: CPU BLAS - dGEMM-NNviennacl: CPU BLAS - dGEMV-Nrodinia: OpenCL Particle Filternamd-cuda: ATPase Simulation - 327,506 AtomsRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny57.2735000.447912.5937738.428.4421.45205221.31287837.93669.37102626392.0917864.54109595016.0773132720.77225952500005.79652.2814.252063405000016205.986370092666710.8458441.332413295.7240002.9320161.5923.74358720074.10392.804645.2458820403.3520305.75645.31640.15121.3532148.8631.17058551.701427.491568293355.526.397524.5486309025.19301.5839.3592349.396.50941547734.33.536814.92823.31856584.811321.3671649.10664.59724142.84830.592418.86372363316.5721495600658359.2324.88371143168346069841842227274976413.5225.046.354.405.3823.9719.1617.5718.7818.256.131431411201071723.7680.0674261.7414076.714858.1316106.428.2222.1395021.50130638.20349.49018515859.1211826.7684027527.4258026011.15220471833338.84092.8722.691630090000012975.655150325000016.8945832.518110270.2630649.1315521.1215.37746015752.98366.058500.9146115974.2215963.59503.53506.01153.1461147.2652.75545932.598270.801442866316.913.204612.6064208012.85301.67914.7901504.879.23933404452.24.307504.66543.829751852.48305306292.6940392475524318.1363.75370132289236581058175522617427476.8023.135.857.075.4926.097.036.1118.776.358.93145142119106.81794.5270.095158.9979463.3627175.988606.05.774.5042655.0854789.211326.23767458607.4640697.7327469388.42185780035.29662977666672.89030.557.474939336666739280.821543333333335.73134093.901929507.1787698.3944269.7943.470129344062.52967.1421394.44128044315.3544276.721396.261391.5155.9252980.1520.187115078.445644.603951815274.226.372425.0717409025.13110.88037.7992779.485.10959427785.82.490871.56886.02204379.70883.95377235.23447.34116499.33304.151660.01446444423.2069568661719410.3289.04443150979466711098187423607717845.085.467.2825.143.956.424.394.854.974.423.701441371151201882.0850.048509.1347802.8716952.655455.85.714.78312617.5740809.36977.5281436655.6924524.34174735010.41114656026.08039363500004.40437.979.323088153333324616.84968919666677.5984960.558118441.4654827.6627669.3732.04483127595.511020.235873.5279527771.8827758.93873.67869.2589.2373065.7624.37477466.030423.578772392077.326.393524.7590408025.10071.12211.5431831.865.63048201567.72.761611.67623.22245228.001053.8860835.87537.49116310.53279.171649.29417385424.3492487540597382.0290.15435153169566781085190124227877835.185.325.167.903.896.304.374.785.024.324.43137135122103.51872.6850.05427OpenBenchmarking.org

NCNN

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

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: squeezenet_ssdRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1428425670SE +/- 0.43, N = 3SE +/- 0.50, N = 3SE +/- 0.05, N = 3SE +/- 0.11, N = 357.2761.748.999.13MIN: 16.86 / MAX: 94.95MIN: 23.2 / MAX: 85.24MIN: 8.18 / MAX: 62.47MIN: 7.71 / MAX: 66.821. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: squeezenet_ssdRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1224364860Min: 56.73 / Avg: 57.27 / Max: 58.11Min: 60.75 / Avg: 61.74 / Max: 62.3Min: 8.89 / Avg: 8.99 / Max: 9.07Min: 8.94 / Avg: 9.13 / Max: 9.321. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

clpeak

Clpeak is designed to test the peak capabilities of OpenCL devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPS, More Is Betterclpeak 1.1.2OpenCL Test: Single-Precision FloatRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20K40K60K80K100KSE +/- 10.34, N = 15SE +/- 80.60, N = 11SE +/- 103.68, N = 14SE +/- 50.90, N = 1435000.4414076.7179463.3647802.871. (CXX) g++ options: -O3
OpenBenchmarking.orgGFLOPS, More Is Betterclpeak 1.1.2OpenCL Test: Single-Precision FloatRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny14K28K42K56K70KMin: 34873.95 / Avg: 35000.44 / Max: 35050.96Min: 13574.87 / Avg: 14076.71 / Max: 14357.38Min: 78695.43 / Avg: 79463.36 / Max: 79709.41Min: 47474.31 / Avg: 47802.87 / Max: 47919.431. (CXX) g++ options: -O3

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: GEMM SGEMM_NRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny6K12K18K24K30KSE +/- 16.79, N = 11SE +/- 33.05, N = 10SE +/- 164.66, N = 13SE +/- 224.54, N = 157912.594858.1327175.9016952.601. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: GEMM SGEMM_NRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny5K10K15K20K25KMin: 7828.3 / Avg: 7912.59 / Max: 8008.46Min: 4783.89 / Avg: 4858.13 / Max: 5089.71Min: 26678 / Avg: 27175.88 / Max: 27932.6Min: 16160.5 / Avg: 16952.57 / Max: 17860.51. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Max SP FlopsRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20K40K60K80K100KSE +/- 292.47, N = 3SE +/- 102.77, N = 3SE +/- 19.25, N = 3SE +/- 73.47, N = 337738.416106.488606.055455.81. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Max SP FlopsRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny15K30K45K60K75KMin: 37374.2 / Avg: 37738.4 / Max: 38316.9Min: 15983.5 / Avg: 16106.37 / Max: 16310.5Min: 88570.6 / Avg: 88606.03 / Max: 88636.8Min: 55361.4 / Avg: 55455.77 / Max: 55600.51. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

NCNN

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

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: googlenetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny714212835SE +/- 0.13, N = 3SE +/- 0.49, N = 3SE +/- 0.10, N = 3SE +/- 0.11, N = 328.4428.225.775.71MIN: 14.74 / MAX: 44.82MIN: 15.18 / MAX: 38.54MIN: 4.99 / MAX: 22.76MIN: 5.03 / MAX: 22.851. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: googlenetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 28.18 / Avg: 28.44 / Max: 28.59Min: 27.65 / Avg: 28.22 / Max: 29.19Min: 5.58 / Avg: 5.77 / Max: 5.94Min: 5.59 / Avg: 5.71 / Max: 5.941. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: resnet18RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025SE +/- 0.21, N = 3SE +/- 0.55, N = 3SE +/- 0.10, N = 3SE +/- 0.06, N = 321.4522.134.504.78MIN: 8.94 / MAX: 40.36MIN: 9.01 / MAX: 43.23MIN: 3.89 / MAX: 21.9MIN: 4.18 / MAX: 28.491. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: resnet18RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025Min: 21.06 / Avg: 21.45 / Max: 21.78Min: 21.38 / Avg: 22.13 / Max: 23.2Min: 4.35 / Avg: 4.5 / Max: 4.68Min: 4.66 / Avg: 4.78 / Max: 4.871. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

Chaos Group V-RAY

This is a test of Chaos Group's V-RAY benchmark. V-RAY is a commercial renderer that can integrate with various creator software products like SketchUp and 3ds Max. The V-RAY benchmark is standalone and supports CPU and NVIDIA CUDA/RTX based rendering. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgvpaths, More Is BetterChaos Group V-RAY 5.02Mode: NVIDIA CUDA GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9001800270036004500SE +/- 2.85, N = 3SE +/- 0.67, N = 3SE +/- 2.33, N = 3SE +/- 2.73, N = 3205295042653126
OpenBenchmarking.orgvpaths, More Is BetterChaos Group V-RAY 5.02Mode: NVIDIA CUDA GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny7001400210028003500Min: 2046 / Avg: 2051.67 / Max: 2055Min: 949 / Avg: 950.33 / Max: 951Min: 4261 / Avg: 4265.33 / Max: 4269Min: 3121 / Avg: 3126.33 / Max: 3130

NCNN

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

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: alexnetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025SE +/- 0.22, N = 3SE +/- 0.30, N = 3SE +/- 0.08, N = 3SE +/- 0.19, N = 321.3121.505.0817.57MIN: 7.31 / MAX: 35.87MIN: 3.33 / MAX: 40.17MIN: 3.13 / MAX: 27.91MIN: 7.8 / MAX: 33.451. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: alexnetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025Min: 20.89 / Avg: 21.31 / Max: 21.6Min: 20.91 / Avg: 21.5 / Max: 21.9Min: 4.92 / Avg: 5.08 / Max: 5.18Min: 17.28 / Avg: 17.57 / Max: 17.931. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

Chaos Group V-RAY

This is a test of Chaos Group's V-RAY benchmark. V-RAY is a commercial renderer that can integrate with various creator software products like SketchUp and 3ds Max. The V-RAY benchmark is standalone and supports CPU and NVIDIA CUDA/RTX based rendering. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgvrays, More Is BetterChaos Group V-RAY 5.02Mode: NVIDIA RTX GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny12002400360048006000SE +/- 18.12, N = 3SE +/- 10.17, N = 3SE +/- 63.00, N = 3SE +/- 30.20, N = 32878130654784080
OpenBenchmarking.orgvrays, More Is BetterChaos Group V-RAY 5.02Mode: NVIDIA RTX GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny10002000300040005000Min: 2843 / Avg: 2878.33 / Max: 2903Min: 1290 / Avg: 1306.33 / Max: 1325Min: 5359 / Avg: 5478.33 / Max: 5573Min: 4020 / Avg: 4080.33 / Max: 4113

NCNN

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

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: mobilenetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny918273645SE +/- 0.21, N = 3SE +/- 0.18, N = 3SE +/- 0.03, N = 3SE +/- 0.06, N = 337.9338.209.219.36MIN: 14.41 / MAX: 61.76MIN: 10.75 / MAX: 66.72MIN: 8.39 / MAX: 47.45MIN: 8.28 / MAX: 49.631. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: mobilenetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny816243240Min: 37.65 / Avg: 37.93 / Max: 38.33Min: 37.97 / Avg: 38.2 / Max: 38.56Min: 9.15 / Avg: 9.21 / Max: 9.25Min: 9.26 / Avg: 9.36 / Max: 9.471. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OctaneBench

OctaneBench is a test of the OctaneRender on the GPU and requires the use of NVIDIA CUDA. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgScore, More Is BetterOctaneBench 2020.1Total ScoreRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500669.37349.491326.24977.53

vkpeak

Vkpeak is a Vulkan compute benchmark inspired by OpenCL's clpeak. Vkpeak provides Vulkan compute performance measurements for FP16 / FP32 / FP64 / INT16 / INT32 scalar and vec4 performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp32-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny13K26K39K52K65KSE +/- 43.15, N = 3SE +/- 77.45, N = 3SE +/- 13.83, N = 3SE +/- 13.71, N = 326392.0915859.1258607.4636655.69
OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp32-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny10K20K30K40K50KMin: 26344.66 / Avg: 26392.09 / Max: 26478.25Min: 15765.59 / Avg: 15859.12 / Max: 16012.81Min: 58587.41 / Avg: 58607.46 / Max: 58633.99Min: 36638.97 / Avg: 36655.69 / Max: 36682.87

clpeak

Clpeak is designed to test the peak capabilities of OpenCL devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGIOPS, More Is Betterclpeak 1.1.2OpenCL Test: Integer Compute INTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9K18K27K36K45KSE +/- 36.29, N = 15SE +/- 86.60, N = 15SE +/- 49.03, N = 14SE +/- 20.64, N = 1417864.5411826.7640697.7324524.341. (CXX) g++ options: -O3
OpenBenchmarking.orgGIOPS, More Is Betterclpeak 1.1.2OpenCL Test: Integer Compute INTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny7K14K21K28K35KMin: 17400.56 / Avg: 17864.54 / Max: 17944.2Min: 11522.47 / Avg: 11826.76 / Max: 12534.81Min: 40335.04 / Avg: 40697.73 / Max: 40806.07Min: 24302.21 / Avg: 24524.34 / Max: 24558.671. (CXX) g++ options: -O3

Hashcat

Hashcat is an open-source, advanced password recovery tool supporting GPU acceleration with OpenCL, NVIDIA CUDA, and Radeon ROCm. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: 7-ZipRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny600K1200K1800K2400K3000KSE +/- 1997.77, N = 8SE +/- 1608.10, N = 8SE +/- 1870.54, N = 8SE +/- 1869.68, N = 8109595084027527469381747350
OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: 7-ZipRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny500K1000K1500K2000K2500KMin: 1087000 / Avg: 1095950 / Max: 1102600Min: 834100 / Avg: 840275 / Max: 844700Min: 2737500 / Avg: 2746937.5 / Max: 2752400Min: 1737000 / Avg: 1747350 / Max: 1752500

Blender

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Pabellon Barcelona - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430SE +/- 0.01, N = 3SE +/- 0.04, N = 3SE +/- 0.02, N = 5SE +/- 0.01, N = 516.0727.428.4210.41
OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Pabellon Barcelona - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 16.05 / Avg: 16.07 / Max: 16.09Min: 27.38 / Avg: 27.42 / Max: 27.49Min: 8.37 / Avg: 8.42 / Max: 8.47Min: 10.39 / Avg: 10.41 / Max: 10.42

Hashcat

Hashcat is an open-source, advanced password recovery tool supporting GPU acceleration with OpenCL, NVIDIA CUDA, and Radeon ROCm. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: TrueCrypt RIPEMD160 + XTSRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny400K800K1200K1600K2000KSE +/- 5945.49, N = 15SE +/- 4392.30, N = 10SE +/- 13557.46, N = 7SE +/- 8051.58, N = 1573132758026018578001146560
OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: TrueCrypt RIPEMD160 + XTSRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny300K600K900K1200K1500KMin: 702200 / Avg: 731326.67 / Max: 766000Min: 550300 / Avg: 580260 / Max: 594700Min: 1784800 / Avg: 1857800 / Max: 1882900Min: 1111400 / Avg: 1146560 / Max: 1182700

IndigoBench

This is a test of Indigo Renderer's IndigoBench benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgM samples/s, More Is BetterIndigoBench 4.4Acceleration: OpenCL GPU - Scene: BedroomRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny816243240SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.03, N = 3SE +/- 0.02, N = 320.7711.1535.3026.08
OpenBenchmarking.orgM samples/s, More Is BetterIndigoBench 4.4Acceleration: OpenCL GPU - Scene: BedroomRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny816243240Min: 20.76 / Avg: 20.77 / Max: 20.79Min: 11.15 / Avg: 11.15 / Max: 11.16Min: 35.26 / Avg: 35.3 / Max: 35.37Min: 26.06 / Avg: 26.08 / Max: 26.11

Hashcat

Hashcat is an open-source, advanced password recovery tool supporting GPU acceleration with OpenCL, NVIDIA CUDA, and Radeon ROCm. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: SHA-512RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1300M2600M3900M5200M6500MSE +/- 2013247.79, N = 6SE +/- 6223365.47, N = 6SE +/- 2101057.93, N = 6SE +/- 1634982.16, N = 62595250000204718333362977666673936350000
OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: SHA-512RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1100M2200M3300M4400M5500MMin: 2589900000 / Avg: 2595250000 / Max: 2601800000Min: 2031700000 / Avg: 2047183333.33 / Max: 2075100000Min: 6288500000 / Avg: 6297766666.67 / Max: 6301700000Min: 3931200000 / Avg: 3936350000 / Max: 3942900000

FinanceBench

FinanceBench is a collection of financial program benchmarks with support for benchmarking on the GPU via OpenCL and CPU benchmarking with OpenMP. The FinanceBench test cases are focused on Black-Sholes-Merton Process with Analytic European Option engine, QMC (Sobol) Monte-Carlo method (Equity Option Example), Bonds Fixed-rate bond with flat forward curve, and Repo Securities repurchase agreement. FinanceBench was originally written by the Cavazos Lab at University of Delaware. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterFinanceBench 2016-07-25Benchmark: Black-Scholes OpenCLRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny246810SE +/- 0.002, N = 14SE +/- 0.110, N = 15SE +/- 0.005, N = 15SE +/- 0.013, N = 155.7968.8402.8904.4041. (CXX) g++ options: -O3 -march=native -fopenmp
OpenBenchmarking.orgms, Fewer Is BetterFinanceBench 2016-07-25Benchmark: Black-Scholes OpenCLRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3691215Min: 5.79 / Avg: 5.8 / Max: 5.81Min: 8.51 / Avg: 8.84 / Max: 10.2Min: 2.88 / Avg: 2.89 / Max: 2.96Min: 4.38 / Avg: 4.4 / Max: 4.591. (CXX) g++ options: -O3 -march=native -fopenmp

Blender

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Barbershop - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20406080100SE +/- 0.01, N = 3SE +/- 0.09, N = 3SE +/- 0.13, N = 3SE +/- 0.07, N = 352.2892.8730.5537.97
OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Barbershop - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20406080100Min: 52.26 / Avg: 52.28 / Max: 52.31Min: 92.73 / Avg: 92.87 / Max: 93.03Min: 30.33 / Avg: 30.55 / Max: 30.78Min: 37.89 / Avg: 37.97 / Max: 38.1

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Classroom - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025SE +/- 0.01, N = 4SE +/- 0.03, N = 3SE +/- 0.01, N = 6SE +/- 0.02, N = 514.2522.697.479.32
OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Classroom - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025Min: 14.22 / Avg: 14.25 / Max: 14.28Min: 22.65 / Avg: 22.69 / Max: 22.75Min: 7.43 / Avg: 7.47 / Max: 7.51Min: 9.26 / Avg: 9.32 / Max: 9.4

Hashcat

Hashcat is an open-source, advanced password recovery tool supporting GPU acceleration with OpenCL, NVIDIA CUDA, and Radeon ROCm. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: SHA1RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny11000M22000M33000M44000M55000MSE +/- 15661518.66, N = 6SE +/- 39682498.24, N = 6SE +/- 18148896.51, N = 6SE +/- 19028429.02, N = 620634050000163009000004939336666730881533333
OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: SHA1RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9000M18000M27000M36000M45000MMin: 20588800000 / Avg: 20634050000 / Max: 20688200000Min: 16215500000 / Avg: 16300900000 / Max: 16474700000Min: 49333100000 / Avg: 49393366666.67 / Max: 49453300000Min: 30793800000 / Avg: 30881533333.33 / Max: 30910600000

vkpeak

Vkpeak is a Vulkan compute benchmark inspired by OpenCL's clpeak. Vkpeak provides Vulkan compute performance measurements for FP16 / FP32 / FP64 / INT16 / INT32 scalar and vec4 performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int16-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny8K16K24K32K40KSE +/- 4.87, N = 3SE +/- 4.25, N = 3SE +/- 1.77, N = 3SE +/- 0.85, N = 316205.9812975.6539280.8224616.84
OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int16-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny7K14K21K28K35KMin: 16201 / Avg: 16205.98 / Max: 16215.73Min: 12967.16 / Avg: 12975.65 / Max: 12980.29Min: 39278.79 / Avg: 39280.82 / Max: 39284.36Min: 24615.27 / Avg: 24616.84 / Max: 24618.2

Hashcat

Hashcat is an open-source, advanced password recovery tool supporting GPU acceleration with OpenCL, NVIDIA CUDA, and Radeon ROCm. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: MD5RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30000M60000M90000M120000M150000MSE +/- 474267340.07, N = 15SE +/- 232076903.56, N = 6SE +/- 243127767.05, N = 6SE +/- 34207656.71, N = 6637009266675150325000015433333333396891966667
OpenBenchmarking.orgH/s, More Is BetterHashcat 6.2.4Benchmark: MD5RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30000M60000M90000M120000M150000MMin: 57828700000 / Avg: 63700926666.67 / Max: 64831500000Min: 51058400000 / Avg: 51503250000 / Max: 52492500000Min: 153400000000 / Avg: 154333333333.33 / Max: 154900000000Min: 96794200000 / Avg: 96891966666.67 / Max: 96995100000

Blender

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Fishy Cat - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620SE +/- 0.07, N = 15SE +/- 0.21, N = 4SE +/- 0.07, N = 15SE +/- 0.07, N = 1510.8416.895.737.59
OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: Fishy Cat - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620Min: 10.75 / Avg: 10.84 / Max: 11.82Min: 16.66 / Avg: 16.89 / Max: 17.52Min: 5.63 / Avg: 5.73 / Max: 6.67Min: 7.49 / Avg: 7.59 / Max: 8.52

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-TTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500SE +/- 1.45, N = 3SE +/- 0.00, N = 3SE +/- 1.20, N = 358445813408491. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-TTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 581 / Avg: 583.67 / Max: 586Min: 1340 / Avg: 1340 / Max: 1340Min: 847 / Avg: 848.67 / Max: 8511. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGHash/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: MD5 HashRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20406080100SE +/- 0.05, N = 14SE +/- 0.06, N = 14SE +/- 0.95, N = 15SE +/- 0.65, N = 1541.3332.5293.9060.561. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGHash/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: MD5 HashRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20406080100Min: 40.99 / Avg: 41.33 / Max: 41.72Min: 32.09 / Avg: 32.52 / Max: 32.98Min: 90.87 / Avg: 93.9 / Max: 98.56Min: 54.7 / Avg: 60.56 / Max: 61.891. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

vkpeak

Vkpeak is a Vulkan compute benchmark inspired by OpenCL's clpeak. Vkpeak provides Vulkan compute performance measurements for FP16 / FP32 / FP64 / INT16 / INT32 scalar and vec4 performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int16-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny6K12K18K24K30KSE +/- 0.25, N = 3SE +/- 0.89, N = 3SE +/- 4.42, N = 3SE +/- 0.34, N = 313295.7210270.2629507.1718441.46
OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int16-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny5K10K15K20K25KMin: 13295.47 / Avg: 13295.72 / Max: 13296.21Min: 10268.48 / Avg: 10270.26 / Max: 10271.26Min: 29502.52 / Avg: 29507.17 / Max: 29516Min: 18440.89 / Avg: 18441.46 / Max: 18442.05

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp16-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20K40K60K80K100KSE +/- 64.52, N = 3SE +/- 26.46, N = 3SE +/- 98.17, N = 3SE +/- 0.37, N = 340002.9330649.1387698.3954827.66
OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp16-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny15K30K45K60K75KMin: 39937.31 / Avg: 40002.93 / Max: 40131.96Min: 30611.63 / Avg: 30649.13 / Max: 30700.21Min: 87502.15 / Avg: 87698.39 / Max: 87802.18Min: 54827.04 / Avg: 54827.66 / Max: 54828.32

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp16-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9K18K27K36K45KSE +/- 30.30, N = 3SE +/- 61.99, N = 3SE +/- 7.21, N = 3SE +/- 0.86, N = 320161.5915521.1244269.7927669.37
OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp16-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny8K16K24K32K40KMin: 20101.14 / Avg: 20161.59 / Max: 20195.39Min: 15445.94 / Avg: 15521.12 / Max: 15644.08Min: 44258.8 / Avg: 44269.79 / Max: 44283.37Min: 27667.67 / Avg: 27669.37 / Max: 27670.44

GROMACS

The GROMACS (GROningen MAchine for Chemical Simulations) molecular dynamics package testing with the water_GMX50 data. This test profile allows selecting between CPU and GPU-based GROMACS builds. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgNs Per Day, More Is BetterGROMACS 2023Implementation: NVIDIA CUDA GPU - Input: water_GMX50_bareRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1020304050SE +/- 0.05, N = 3SE +/- 0.00, N = 3SE +/- 0.02, N = 3SE +/- 0.02, N = 323.7415.3843.4732.041. (CXX) g++ options: -O3
OpenBenchmarking.orgNs Per Day, More Is BetterGROMACS 2023Implementation: NVIDIA CUDA GPU - Input: water_GMX50_bareRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny918273645Min: 23.65 / Avg: 23.74 / Max: 23.81Min: 15.37 / Avg: 15.38 / Max: 15.39Min: 43.45 / Avg: 43.47 / Max: 43.5Min: 32.01 / Avg: 32.04 / Max: 32.091. (CXX) g++ options: -O3

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-TNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500SE +/- 1.67, N = 3SE +/- 0.50, N = 2SE +/- 3.33, N = 3SE +/- 0.67, N = 358746012938311. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-TNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 585 / Avg: 586.67 / Max: 590Min: 459 / Avg: 459.5 / Max: 460Min: 1290 / Avg: 1293.33 / Max: 1300Min: 830 / Avg: 831.33 / Max: 8321. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

vkpeak

Vkpeak is a Vulkan compute benchmark inspired by OpenCL's clpeak. Vkpeak provides Vulkan compute performance measurements for FP16 / FP32 / FP64 / INT16 / INT32 scalar and vec4 performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int32-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9K18K27K36K45KSE +/- 27.49, N = 3SE +/- 15.45, N = 3SE +/- 3.11, N = 3SE +/- 37.47, N = 320074.1015752.9844062.5227595.51
OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int32-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny8K16K24K32K40KMin: 20046.24 / Avg: 20074.1 / Max: 20129.07Min: 15729.55 / Avg: 15752.98 / Max: 15782.13Min: 44056.29 / Avg: 44062.52 / Max: 44065.74Min: 27520.58 / Avg: 27595.51 / Max: 27633.03

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: ReductionRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 0.11, N = 13SE +/- 0.06, N = 12SE +/- 9.75, N = 15SE +/- 13.90, N = 15392.80366.06967.141020.241. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: ReductionRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 392.29 / Avg: 392.8 / Max: 393.74Min: 365.65 / Avg: 366.06 / Max: 366.32Min: 922.16 / Avg: 967.14 / Max: 999.14Min: 939.68 / Avg: 1020.23 / Max: 1057.041. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

vkpeak

Vkpeak is a Vulkan compute benchmark inspired by OpenCL's clpeak. Vkpeak provides Vulkan compute performance measurements for FP16 / FP32 / FP64 / INT16 / INT32 scalar and vec4 performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp64-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500SE +/- 1.62, N = 3SE +/- 1.28, N = 3SE +/- 0.73, N = 3SE +/- 0.08, N = 3645.24500.911394.44873.52
OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp64-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 643.62 / Avg: 645.24 / Max: 648.47Min: 499.5 / Avg: 500.91 / Max: 503.47Min: 1393.64 / Avg: 1394.44 / Max: 1395.9Min: 873.44 / Avg: 873.52 / Max: 873.67

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-NTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500SE +/- 1.67, N = 3SE +/- 0.58, N = 3SE +/- 0.00, N = 3SE +/- 1.00, N = 358846112807951. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-NTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 586 / Avg: 587.67 / Max: 591Min: 460 / Avg: 461 / Max: 462Min: 1280 / Avg: 1280 / Max: 1280Min: 794 / Avg: 795 / Max: 7971. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

vkpeak

Vkpeak is a Vulkan compute benchmark inspired by OpenCL's clpeak. Vkpeak provides Vulkan compute performance measurements for FP16 / FP32 / FP64 / INT16 / INT32 scalar and vec4 performance. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp32-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9K18K27K36K45KSE +/- 34.37, N = 3SE +/- 136.44, N = 3SE +/- 10.09, N = 3SE +/- 35.38, N = 320403.3515974.2244315.3527771.88
OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp32-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny8K16K24K32K40KMin: 20355.27 / Avg: 20403.35 / Max: 20469.93Min: 15772.59 / Avg: 15974.22 / Max: 16234.26Min: 44303.49 / Avg: 44315.35 / Max: 44335.43Min: 27701.33 / Avg: 27771.88 / Max: 27811.8

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int32-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9K18K27K36K45KSE +/- 10.90, N = 3SE +/- 39.60, N = 3SE +/- 25.10, N = 3SE +/- 2.25, N = 320305.7515963.5944276.7227758.93
OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20210424int32-scalarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny8K16K24K32K40KMin: 20294.31 / Avg: 20305.75 / Max: 20327.54Min: 15905.24 / Avg: 15963.59 / Max: 16039.15Min: 44240.32 / Avg: 44276.72 / Max: 44324.87Min: 27754.84 / Avg: 27758.93 / Max: 27762.6

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp64-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500SE +/- 1.70, N = 3SE +/- 0.03, N = 3SE +/- 1.25, N = 3SE +/- 0.01, N = 3645.31503.531396.26873.67
OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20210424fp64-vec4RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 643.55 / Avg: 645.31 / Max: 648.72Min: 503.47 / Avg: 503.53 / Max: 503.57Min: 1393.77 / Avg: 1396.26 / Max: 1397.55Min: 873.64 / Avg: 873.67 / Max: 873.68

clpeak

Clpeak is designed to test the peak capabilities of OpenCL devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPS, More Is Betterclpeak 1.1.2OpenCL Test: Double-Precision DoubleRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500SE +/- 0.46, N = 5SE +/- 1.50, N = 4SE +/- 1.85, N = 6SE +/- 1.29, N = 6640.15506.011391.51869.251. (CXX) g++ options: -O3
OpenBenchmarking.orgGFLOPS, More Is Betterclpeak 1.1.2OpenCL Test: Double-Precision DoubleRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 639.27 / Avg: 640.15 / Max: 641.82Min: 502.28 / Avg: 506.01 / Max: 508.86Min: 1385.74 / Avg: 1391.51 / Max: 1396.76Min: 864.96 / Avg: 869.25 / Max: 873.621. (CXX) g++ options: -O3

VkResample

VkResample is a Vulkan-based image upscaling library based on VkFFT. The sample input file is upscaling a 4K image to 8K using Vulkan-based GPU acceleration. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterVkResample 1.0Upscale: 2x - Precision: DoubleRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150SE +/- 0.06, N = 3SE +/- 0.26, N = 3SE +/- 0.05, N = 3SE +/- 0.06, N = 3121.35153.1555.9389.241. (CXX) g++ options: -O3
OpenBenchmarking.orgms, Fewer Is BetterVkResample 1.0Upscale: 2x - Precision: DoubleRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150Min: 121.24 / Avg: 121.35 / Max: 121.44Min: 152.62 / Avg: 153.15 / Max: 153.41Min: 55.84 / Avg: 55.93 / Max: 56.03Min: 89.13 / Avg: 89.24 / Max: 89.321. (CXX) g++ options: -O3

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Texture Read BandwidthRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny7001400210028003500SE +/- 0.38, N = 3SE +/- 0.54, N = 3SE +/- 0.95, N = 6SE +/- 2.39, N = 72148.861147.262980.153065.761. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Texture Read BandwidthRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny5001000150020002500Min: 2148.29 / Avg: 2148.86 / Max: 2149.58Min: 1146.24 / Avg: 1147.26 / Max: 1148.09Min: 2977.32 / Avg: 2980.15 / Max: 2983.15Min: 3055.67 / Avg: 3065.76 / Max: 3076.431. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

RealSR-NCNN

RealSR-NCNN is an NCNN neural network implementation of the RealSR project and accelerated using the Vulkan API. RealSR is the Real-World Super Resolution via Kernel Estimation and Noise Injection. NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. This test profile times how long it takes to increase the resolution of a sample image by a scale of 4x with Vulkan. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterRealSR-NCNN 20200818Scale: 4x - TAA: YesRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1224364860SE +/- 0.18, N = 3SE +/- 0.07, N = 3SE +/- 0.04, N = 3SE +/- 0.04, N = 331.1752.7620.1924.37
OpenBenchmarking.orgSeconds, Fewer Is BetterRealSR-NCNN 20200818Scale: 4x - TAA: YesRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1122334455Min: 30.97 / Avg: 31.17 / Max: 31.53Min: 52.61 / Avg: 52.76 / Max: 52.84Min: 20.11 / Avg: 20.19 / Max: 20.24Min: 24.31 / Avg: 24.37 / Max: 24.43

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-NNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 1.67, N = 3SE +/- 0.88, N = 3SE +/- 0.00, N = 3SE +/- 0.88, N = 358545911507741. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-NNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 582 / Avg: 585.33 / Max: 587Min: 457 / Avg: 458.67 / Max: 460Min: 1150 / Avg: 1150 / Max: 1150Min: 773 / Avg: 774.33 / Max: 7761. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

IndigoBench

This is a test of Indigo Renderer's IndigoBench benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgM samples/s, More Is BetterIndigoBench 4.4Acceleration: OpenCL GPU - Scene: SupercarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20406080100SE +/- 0.02, N = 3SE +/- 0.05, N = 3SE +/- 0.01, N = 3SE +/- 0.07, N = 351.7032.6078.4566.03
OpenBenchmarking.orgM samples/s, More Is BetterIndigoBench 4.4Acceleration: OpenCL GPU - Scene: SupercarRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1530456075Min: 51.68 / Avg: 51.7 / Max: 51.75Min: 32.54 / Avg: 32.6 / Max: 32.69Min: 78.43 / Avg: 78.45 / Max: 78.46Min: 65.89 / Avg: 66.03 / Max: 66.12

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: S3DRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny140280420560700SE +/- 0.35, N = 12SE +/- 0.15, N = 12SE +/- 0.22, N = 12SE +/- 0.38, N = 13427.49270.80644.60423.581. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: S3DRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny110220330440550Min: 425.04 / Avg: 427.49 / Max: 429.27Min: 269.87 / Avg: 270.8 / Max: 272.06Min: 643.24 / Avg: 644.6 / Max: 645.96Min: 421.55 / Avg: 423.58 / Max: 425.081. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

MandelGPU

MandelGPU is an OpenCL benchmark and this test runs with the OpenCL rendering float4 kernel with a maximum of 4096 iterations. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSamples/sec, More Is BetterMandelGPU 1.3pts1OpenCL Device: GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny200M400M600M800M1000MSE +/- 959338.71, N = 11SE +/- 1540493.49, N = 10SE +/- 3825360.26, N = 12SE +/- 1340958.21, N = 12568293355.5442866316.9951815274.2772392077.31. (CC) gcc options: -O3 -lm -ftree-vectorize -funroll-loops -lglut -lOpenCL -lGL
OpenBenchmarking.orgSamples/sec, More Is BetterMandelGPU 1.3pts1OpenCL Device: GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny160M320M480M640M800MMin: 562888522.9 / Avg: 568293355.45 / Max: 575316699Min: 433765928.8 / Avg: 442866316.9 / Max: 448235681.6Min: 919190495.8 / Avg: 951815274.2 / Max: 966366353.1Min: 759995838.3 / Avg: 772392077.27 / Max: 778372516.81. (CC) gcc options: -O3 -lm -ftree-vectorize -funroll-loops -lglut -lOpenCL -lGL

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Bus Speed ReadbackRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430SE +/- 0.00, N = 13SE +/- 0.00, N = 12SE +/- 0.00, N = 13SE +/- 0.00, N = 1426.4013.2026.3726.391. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Bus Speed ReadbackRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 26.4 / Avg: 26.4 / Max: 26.4Min: 13.2 / Avg: 13.2 / Max: 13.2Min: 26.37 / Avg: 26.37 / Max: 26.38Min: 26.39 / Avg: 26.39 / Max: 26.41. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: TriadRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430SE +/- 0.00, N = 12SE +/- 0.00, N = 12SE +/- 0.01, N = 12SE +/- 0.01, N = 1224.5512.6125.0724.761. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: TriadRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 24.53 / Avg: 24.55 / Max: 24.58Min: 12.59 / Avg: 12.61 / Max: 12.62Min: 24.98 / Avg: 25.07 / Max: 25.1Min: 24.69 / Avg: 24.76 / Max: 24.81. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

NeatBench

NeatBench is a benchmark of the cross-platform Neat Video software on the CPU and optional GPU (OpenCL / CUDA) support. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterNeatBench 5Acceleration: GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny9001800270036004500SE +/- 0.00, N = 13SE +/- 0.00, N = 13SE +/- 0.00, N = 14SE +/- 0.00, N = 143090208040904080
OpenBenchmarking.orgFPS, More Is BetterNeatBench 5Acceleration: GPURTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny7001400210028003500Min: 3090 / Avg: 3090 / Max: 3090Min: 2080 / Avg: 2080 / Max: 2080Min: 4090 / Avg: 4090 / Max: 4090Min: 4080 / Avg: 4080 / Max: 4080

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Bus Speed DownloadRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430SE +/- 0.02, N = 13SE +/- 0.00, N = 12SE +/- 0.02, N = 13SE +/- 0.02, N = 1325.1912.8525.1325.101. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGB/s, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: Bus Speed DownloadRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 25.11 / Avg: 25.19 / Max: 25.27Min: 12.84 / Avg: 12.85 / Max: 12.87Min: 25.03 / Avg: 25.13 / Max: 25.22Min: 25.03 / Avg: 25.1 / Max: 25.211. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

ArrayFire

ArrayFire is an GPU and CPU numeric processing library, this test uses the built-in CPU and OpenCL ArrayFire benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterArrayFire 3.7Test: Conjugate Gradient OpenCLRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny0.37780.75561.13341.51121.889SE +/- 0.0020, N = 9SE +/- 0.0055, N = 9SE +/- 0.0014, N = 9SE +/- 0.0013, N = 101.58301.67900.88031.12201. (CXX) g++ options: -rdynamic
OpenBenchmarking.orgms, Fewer Is BetterArrayFire 3.7Test: Conjugate Gradient OpenCLRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny246810Min: 1.58 / Avg: 1.58 / Max: 1.59Min: 1.66 / Avg: 1.68 / Max: 1.71Min: 0.87 / Avg: 0.88 / Max: 0.89Min: 1.12 / Avg: 1.12 / Max: 1.131. (CXX) g++ options: -rdynamic

VkResample

VkResample is a Vulkan-based image upscaling library based on VkFFT. The sample input file is upscaling a 4K image to 8K using Vulkan-based GPU acceleration. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterVkResample 1.0Upscale: 2x - Precision: SingleRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620SE +/- 0.008, N = 5SE +/- 0.023, N = 5SE +/- 0.003, N = 5SE +/- 0.003, N = 59.35914.7907.79911.5431. (CXX) g++ options: -O3
OpenBenchmarking.orgms, Fewer Is BetterVkResample 1.0Upscale: 2x - Precision: SingleRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620Min: 9.34 / Avg: 9.36 / Max: 9.38Min: 14.73 / Avg: 14.79 / Max: 14.85Min: 7.79 / Avg: 7.8 / Max: 7.81Min: 11.54 / Avg: 11.54 / Max: 11.551. (CXX) g++ options: -O3

SHOC Scalable HeterOgeneous Computing

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

OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: FFT SPRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny6001200180024003000SE +/- 0.59, N = 12SE +/- 0.70, N = 11SE +/- 1.36, N = 12SE +/- 2.58, N = 122349.391504.872779.481831.861. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi
OpenBenchmarking.orgGFLOPS, More Is BetterSHOC Scalable HeterOgeneous Computing 2020-04-17Target: OpenCL - Benchmark: FFT SPRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny5001000150020002500Min: 2345.34 / Avg: 2349.39 / Max: 2352.71Min: 1501.88 / Avg: 1504.87 / Max: 1509.32Min: 2769.78 / Avg: 2779.48 / Max: 2785.64Min: 1817.43 / Avg: 1831.86 / Max: 18451. (CXX) g++ options: -O2 -lSHOCCommonMPI -lSHOCCommonOpenCL -lSHOCCommon -lOpenCL -lrt -lmpi_cxx -lmpi

RealSR-NCNN

RealSR-NCNN is an NCNN neural network implementation of the RealSR project and accelerated using the Vulkan API. RealSR is the Real-World Super Resolution via Kernel Estimation and Noise Injection. NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. This test profile times how long it takes to increase the resolution of a sample image by a scale of 4x with Vulkan. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterRealSR-NCNN 20200818Scale: 4x - TAA: NoRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3691215SE +/- 0.015, N = 6SE +/- 0.025, N = 5SE +/- 0.018, N = 7SE +/- 0.015, N = 76.5099.2395.1095.630
OpenBenchmarking.orgSeconds, Fewer Is BetterRealSR-NCNN 20200818Scale: 4x - TAA: NoRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3691215Min: 6.48 / Avg: 6.51 / Max: 6.57Min: 9.19 / Avg: 9.24 / Max: 9.31Min: 5.04 / Avg: 5.11 / Max: 5.16Min: 5.58 / Avg: 5.63 / Max: 5.68

VkFFT

VkFFT is a Fast Fourier Transform (FFT) Library that is GPU accelerated by means of the Vulkan API. The VkFFT benchmark runs FFT performance differences of many different sizes before returning an overall benchmark score. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.1.1RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny13K26K39K52K65KSE +/- 475.08, N = 9SE +/- 163.05, N = 3SE +/- 72.97, N = 3SE +/- 213.00, N = 3415473340459427482011. (CXX) g++ options: -O3
OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.1.1RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny10K20K30K40K50KMin: 39801 / Avg: 41546.89 / Max: 44483Min: 33164 / Avg: 33403.67 / Max: 33715Min: 59295 / Avg: 59426.67 / Max: 59547Min: 47987 / Avg: 48201 / Max: 486271. (CXX) g++ options: -O3

cl-mem

A basic OpenCL memory benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: WriteRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 0.49, N = 10SE +/- 0.85, N = 8SE +/- 0.36, N = 10SE +/- 0.54, N = 9734.3452.2785.8567.71. (CC) gcc options: -O2 -flto -lOpenCL
OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: WriteRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny140280420560700Min: 730.1 / Avg: 734.32 / Max: 735.7Min: 449.4 / Avg: 452.23 / Max: 456.4Min: 784.3 / Avg: 785.75 / Max: 788Min: 566.4 / Avg: 567.74 / Max: 571.51. (CC) gcc options: -O2 -flto -lOpenCL

Waifu2x-NCNN Vulkan

Waifu2x-NCNN is an NCNN neural network implementation of the Waifu2x converter project and accelerated using the Vulkan API. NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. This test profile times how long it takes to increase the resolution of a sample image with Vulkan. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterWaifu2x-NCNN Vulkan 20200818Scale: 2x - Denoise: 3 - TAA: YesRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny0.96911.93822.90733.87644.8455SE +/- 0.005, N = 9SE +/- 0.007, N = 8SE +/- 0.003, N = 10SE +/- 0.004, N = 103.5364.3072.4902.761
OpenBenchmarking.orgSeconds, Fewer Is BetterWaifu2x-NCNN Vulkan 20200818Scale: 2x - Denoise: 3 - TAA: YesRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny246810Min: 3.52 / Avg: 3.54 / Max: 3.57Min: 4.28 / Avg: 4.31 / Max: 4.34Min: 2.48 / Avg: 2.49 / Max: 2.51Min: 2.74 / Avg: 2.76 / Max: 2.78

clpeak

Clpeak is designed to test the peak capabilities of OpenCL devices. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGBPS, More Is Betterclpeak 1.1.2OpenCL Test: Global Memory BandwidthRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 0.01, N = 10SE +/- 0.36, N = 9SE +/- 0.04, N = 10SE +/- 0.17, N = 10814.92504.66871.56611.671. (CXX) g++ options: -O3
OpenBenchmarking.orgGBPS, More Is Betterclpeak 1.1.2OpenCL Test: Global Memory BandwidthRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny150300450600750Min: 814.88 / Avg: 814.92 / Max: 814.95Min: 502.35 / Avg: 504.66 / Max: 505.53Min: 871.27 / Avg: 871.56 / Max: 871.67Min: 610.18 / Avg: 611.67 / Max: 611.891. (CXX) g++ options: -O3

cl-mem

A basic OpenCL memory benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: ReadRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 0.81, N = 9SE +/- 0.15, N = 8SE +/- 0.31, N = 10SE +/- 0.71, N = 9823.3543.8886.0623.21. (CC) gcc options: -O2 -flto -lOpenCL
OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: ReadRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny160320480640800Min: 819.6 / Avg: 823.33 / Max: 825.1Min: 543.3 / Avg: 543.79 / Max: 544.4Min: 883.5 / Avg: 885.99 / Max: 887.4Min: 617.6 / Avg: 623.21 / Max: 624.31. (CC) gcc options: -O2 -flto -lOpenCL

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMV-NRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny60120180240300SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 31852972202241. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMV-NRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny50100150200250Min: 185 / Avg: 185.33 / Max: 186Min: 296 / Avg: 296.67 / Max: 297Min: 220 / Avg: 220 / Max: 220Min: 224 / Avg: 224 / Max: 2241. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

Caffe

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

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 1000RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny14002800420056007000SE +/- 2.76, N = 6SE +/- 2.74, N = 7SE +/- 3.07, N = 66584.814379.705228.001. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas
OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 1000RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny11002200330044005500Min: 6575.72 / Avg: 6584.81 / Max: 6593.17Min: 4368.12 / Avg: 4379.7 / Max: 4387.53Min: 5220.71 / Avg: 5228 / Max: 5240.841. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas

Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 1000

RTX 2080 Ti 22GB -Dell: The test quit with a non-zero exit status. The test quit with a non-zero exit status. The test quit with a non-zero exit status. E: @ 0x7f3f7313f78f google::LogMessageFatal::~LogMessageFatal()

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 200RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500SE +/- 1.00, N = 10SE +/- 0.87, N = 11SE +/- 1.43, N = 101321.36883.951053.881. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas
OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 200RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 1313.3 / Avg: 1321.36 / Max: 1323.94Min: 878.89 / Avg: 883.95 / Max: 888.12Min: 1048.25 / Avg: 1053.88 / Max: 1062.031. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas

Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 200

RTX 2080 Ti 22GB -Dell: The test quit with a non-zero exit status. The test quit with a non-zero exit status. The test quit with a non-zero exit status. E: @ 0x7efdd9bea78f google::LogMessageFatal::~LogMessageFatal()

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny170340510680850SE +/- 1.00, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 37165187726081. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny140280420560700Min: 715 / Avg: 716 / Max: 718Min: 518 / Avg: 518.33 / Max: 519Min: 772 / Avg: 772.33 / Max: 773Min: 607 / Avg: 607.67 / Max: 6081. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

NCNN

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

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: yolov4-tinyRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1224364860SE +/- 0.40, N = 3SE +/- 0.13, N = 3SE +/- 0.10, N = 3SE +/- 0.08, N = 349.1052.4835.2335.87MIN: 17.59 / MAX: 76.07MIN: 20.11 / MAX: 75.46MIN: 11.4 / MAX: 59.2MIN: 11.56 / MAX: 61.961. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: yolov4-tinyRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1122334455Min: 48.6 / Avg: 49.1 / Max: 49.89Min: 52.27 / Avg: 52.48 / Max: 52.72Min: 35.05 / Avg: 35.23 / Max: 35.37Min: 35.78 / Avg: 35.87 / Max: 36.021. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

Caffe

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

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 100RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny140280420560700SE +/- 1.37, N = 11SE +/- 1.11, N = 11SE +/- 1.06, N = 11664.60447.34537.491. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas
OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 100RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny120240360480600Min: 658.38 / Avg: 664.6 / Max: 671.94Min: 440.34 / Avg: 447.34 / Max: 451.04Min: 532.93 / Avg: 537.49 / Max: 5421. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas

Model: AlexNet - Acceleration: NVIDIA CUDA - Iterations: 100

RTX 2080 Ti 22GB -Dell: The test quit with a non-zero exit status. The test quit with a non-zero exit status. The test quit with a non-zero exit status. E: @ 0x7f50d65e678f google::LogMessageFatal::~LogMessageFatal()

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 1000RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny5K10K15K20K25KSE +/- 30.41, N = 3SE +/- 24.30, N = 3SE +/- 18.97, N = 324142.816499.316310.51. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas
OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 1000RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny4K8K12K16K20KMin: 24105.8 / Avg: 24142.8 / Max: 24203.1Min: 16451.5 / Avg: 16499.33 / Max: 16530.7Min: 16287.9 / Avg: 16310.5 / Max: 16348.21. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas

Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 1000

RTX 2080 Ti 22GB -Dell: The test quit with a non-zero exit status. The test quit with a non-zero exit status. The test quit with a non-zero exit status. E: @ 0x7f92f815c78f google::LogMessageFatal::~LogMessageFatal()

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 200RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny10002000300040005000SE +/- 2.81, N = 7SE +/- 3.64, N = 8SE +/- 3.25, N = 84830.593304.153279.171. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas
OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 200RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny8001600240032004000Min: 4816.22 / Avg: 4830.59 / Max: 4838.06Min: 3294.38 / Avg: 3304.15 / Max: 3320.68Min: 3266.13 / Avg: 3279.17 / Max: 3290.81. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas

Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 200

RTX 2080 Ti 22GB -Dell: The test quit with a non-zero exit status. The test quit with a non-zero exit status. The test quit with a non-zero exit status. E: @ 0x7f1cb2fce78f google::LogMessageFatal::~LogMessageFatal()

OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 100RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny5001000150020002500SE +/- 1.69, N = 9SE +/- 1.34, N = 10SE +/- 2.33, N = 102418.861660.011649.291. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas
OpenBenchmarking.orgMilli-Seconds, Fewer Is BetterCaffe 2020-02-13Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 100RTX 3090 24GB -ZotacRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny400800120016002000Min: 2410.23 / Avg: 2418.86 / Max: 2426.92Min: 1655.07 / Avg: 1660.01 / Max: 1667.29Min: 1636.49 / Avg: 1649.29 / Max: 1659.081. (CXX) g++ options: -fPIC -O3 -rdynamic -lglog -lgflags -lprotobuf -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -llmdb -lopenblas

Model: GoogleNet - Acceleration: NVIDIA CUDA - Iterations: 100

RTX 2080 Ti 22GB -Dell: The test quit with a non-zero exit status. The test quit with a non-zero exit status. The test quit with a non-zero exit status. E: @ 0x7f4bd97dc78f google::LogMessageFatal::~LogMessageFatal()

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny100200300400500SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 33723054464171. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny80160240320400Min: 372 / Avg: 372.33 / Max: 373Min: 305 / Avg: 305.33 / Max: 306Min: 445 / Avg: 445.67 / Max: 446Min: 416 / Avg: 416.67 / Max: 4171. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny100200300400500SE +/- 1.20, N = 3SE +/- 1.73, N = 3SE +/- 0.88, N = 3SE +/- 0.33, N = 33633064443851. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny80160240320400Min: 361 / Avg: 363.33 / Max: 365Min: 303 / Avg: 306 / Max: 309Min: 442 / Avg: 443.67 / Max: 445Min: 384 / Avg: 384.67 / Max: 3851. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

FAHBench

FAHBench is a Folding@Home benchmark on the GPU. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgNs Per Day, More Is BetterFAHBench 2.3.2RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny90180270360450SE +/- 0.41, N = 3SE +/- 0.79, N = 3SE +/- 0.24, N = 3SE +/- 0.63, N = 3316.57292.69423.21424.35
OpenBenchmarking.orgNs Per Day, More Is BetterFAHBench 2.3.2RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny80160240320400Min: 315.97 / Avg: 316.57 / Max: 317.35Min: 291.19 / Avg: 292.69 / Max: 293.84Min: 422.87 / Avg: 423.21 / Max: 423.66Min: 423.28 / Avg: 424.35 / Max: 425.46

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny120240360480600SE +/- 0.67, N = 3SE +/- 0.33, N = 3SE +/- 0.00, N = 3SE +/- 0.33, N = 34953925684871. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny100200300400500Min: 494 / Avg: 495.33 / Max: 496Min: 392 / Avg: 392.33 / Max: 393Min: 568 / Avg: 568 / Max: 568Min: 486 / Avg: 486.67 / Max: 4871. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny140280420560700SE +/- 1.20, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 36004756615401. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny120240360480600Min: 598 / Avg: 600.33 / Max: 602Min: 474 / Avg: 474.67 / Max: 475Min: 660 / Avg: 660.67 / Max: 661Min: 540 / Avg: 540.33 / Max: 5411. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny160320480640800SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.67, N = 3SE +/- 0.33, N = 36585247195971. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny130260390520650Min: 658 / Avg: 658.33 / Max: 659Min: 524 / Avg: 524.33 / Max: 525Min: 718 / Avg: 718.67 / Max: 720Min: 597 / Avg: 597.33 / Max: 5981. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

cl-mem

A basic OpenCL memory benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: CopyRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny90180270360450SE +/- 0.21, N = 10SE +/- 0.27, N = 8SE +/- 0.05, N = 10SE +/- 0.03, N = 9359.2318.1410.3382.01. (CC) gcc options: -O2 -flto -lOpenCL
OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: CopyRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny70140210280350Min: 358.4 / Avg: 359.22 / Max: 360.3Min: 317.2 / Avg: 318.09 / Max: 319Min: 410 / Avg: 410.31 / Max: 410.5Min: 381.9 / Avg: 382.04 / Max: 382.21. (CC) gcc options: -O2 -flto -lOpenCL

NCNN

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

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: vision_transformerRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny80160240320400SE +/- 0.55, N = 3SE +/- 0.57, N = 3SE +/- 1.34, N = 3SE +/- 3.05, N = 3324.88363.75289.04290.15MIN: 283.26 / MAX: 423.43MIN: 316.04 / MAX: 471.43MIN: 260.04 / MAX: 661.43MIN: 247.6 / MAX: 837.911. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: vision_transformerRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny60120180240300Min: 323.93 / Avg: 324.88 / Max: 325.84Min: 362.65 / Avg: 363.75 / Max: 364.55Min: 286.38 / Avg: 289.04 / Max: 290.64Min: 285.76 / Avg: 290.15 / Max: 296.021. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMV-TRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny100200300400500SE +/- 0.58, N = 3SE +/- 0.58, N = 3SE +/- 0.00, N = 2SE +/- 0.67, N = 33713704434351. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMV-TRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny80160240320400Min: 370 / Avg: 371 / Max: 372Min: 369 / Avg: 370 / Max: 371Min: 443 / Avg: 443 / Max: 443Min: 434 / Avg: 435.33 / Max: 4361. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

LeelaChessZero

LeelaChessZero (lc0 / lczero) is a chess engine automated vian neural networks. This test profile can be used for OpenCL, CUDA + cuDNN, and BLAS (CPU-based) benchmarking. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgNodes Per Second, More Is BetterLeelaChessZero 0.28Backend: OpenCLRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3K6K9K12K15KSE +/- 52.37, N = 3SE +/- 135.58, N = 4SE +/- 198.64, N = 3SE +/- 126.87, N = 3143161322815097153161. (CXX) g++ options: -flto -pthread
OpenBenchmarking.orgNodes Per Second, More Is BetterLeelaChessZero 0.28Backend: OpenCLRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3K6K9K12K15KMin: 14248 / Avg: 14316 / Max: 14419Min: 12846 / Avg: 13227.5 / Max: 13448Min: 14710 / Avg: 15096.67 / Max: 15369Min: 15084 / Avg: 15316 / Max: 155211. (CXX) g++ options: -flto -pthread

ViennaCL

ViennaCL is an open-source linear algebra library written in C++ and with support for OpenCL and OpenMP. This test profile makes use of ViennaCL's built-in benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 7.74, N = 15SE +/- 8.86, N = 15SE +/- 11.59, N = 5SE +/- 9.35, N = 128349239469561. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 804 / Avg: 833.67 / Max: 909Min: 879 / Avg: 923.33 / Max: 990Min: 913 / Avg: 946.4 / Max: 984Min: 926 / Avg: 955.67 / Max: 10201. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny150300450600750SE +/- 1.68, N = 15SE +/- 4.70, N = 15SE +/- 6.71, N = 5SE +/- 8.14, N = 126066586716781. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny120240360480600Min: 589 / Avg: 605.93 / Max: 613Min: 631 / Avg: 658.27 / Max: 692Min: 655 / Avg: 670.8 / Max: 695Min: 660 / Avg: 677.92 / Max: 7651. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 6.89, N = 15SE +/- 5.45, N = 15SE +/- 22.23, N = 5SE +/- 7.44, N = 129841058109810851. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000Min: 959 / Avg: 983.8 / Max: 1060Min: 1030 / Avg: 1058 / Max: 1100Min: 1060 / Avg: 1098 / Max: 1180Min: 1050 / Avg: 1085 / Max: 11301. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny400800120016002000SE +/- 12.27, N = 15SE +/- 17.80, N = 15SE +/- 19.65, N = 5SE +/- 13.45, N = 1218421755187419011. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sCOPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny30060090012001500Min: 1800 / Avg: 1842 / Max: 2000Min: 1650 / Avg: 1755.33 / Max: 1910Min: 1810 / Avg: 1874 / Max: 1920Min: 1800 / Avg: 1900.83 / Max: 19601. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny5001000150020002500SE +/- 18.29, N = 15SE +/- 16.08, N = 15SE +/- 13.04, N = 5SE +/- 31.16, N = 1222722261236024221. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sAXPYRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny400800120016002000Min: 2150 / Avg: 2272 / Max: 2450Min: 2190 / Avg: 2260.67 / Max: 2430Min: 2310 / Avg: 2360 / Max: 2380Min: 2280 / Avg: 2421.67 / Max: 25801. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMV-TRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 4.88, N = 15SE +/- 6.17, N = 14SE +/- 2.58, N = 5SE +/- 7.65, N = 127497427717871. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMV-TRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny140280420560700Min: 727 / Avg: 748.6 / Max: 794Min: 723 / Avg: 741.5 / Max: 807Min: 763 / Avg: 771.4 / Max: 777Min: 761 / Avg: 786.75 / Max: 8381. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny2004006008001000SE +/- 6.38, N = 15SE +/- 4.44, N = 15SE +/- 7.38, N = 5SE +/- 5.06, N = 127647477847831. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sDOTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny140280420560700Min: 733 / Avg: 763.6 / Max: 798Min: 728 / Avg: 747.4 / Max: 786Min: 767 / Avg: 783.6 / Max: 811Min: 731 / Avg: 783 / Max: 7991. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

GPU Temperature Monitor

OpenBenchmarking.orgCelsiusGPU Temperature MonitorPhoronix Test Suite System MonitoringRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1632486480Min: 41 / Avg: 65.62 / Max: 84Min: 44 / Avg: 69.38 / Max: 83Min: 32 / Avg: 43.13 / Max: 66Min: 31 / Avg: 44.05 / Max: 73

GPU Power Consumption Monitor

OpenBenchmarking.orgWattsGPU Power Consumption MonitorPhoronix Test Suite System MonitoringRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny80160240320400Min: 18.42 / Avg: 170.36 / Max: 367.21Min: 18.41 / Avg: 148.08 / Max: 347.3Min: 6 / Avg: 110.92 / Max: 456.05Min: 4 / Avg: 92.31 / Max: 331.21

NCNN

MinAvgMaxRTX 3090 24GB -Zotac42.051.664.0RTX 2080 Ti 22GB -Dell46.049.861.0RTX 4090 24GB -Nvidia33.037.445.0RTX 4080 16GB -Pny35.038.243.0OpenBenchmarking.orgCelsius, Fewer Is BetterNCNN 20220729GPU Temperature Monitor20406080100

MinAvgMaxRTX 3090 24GB -Zotac18.836.1148.3RTX 2080 Ti 22GB -Dell19.231.7108.8RTX 4090 24GB -Nvidia6.413.659.6RTX 4080 16GB -Pny4.218.566.4OpenBenchmarking.orgWatts, Fewer Is BetterNCNN 20220729GPU Power Consumption Monitor4080120160200

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: FastestDetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3691215SE +/- 3.64, N = 3SE +/- 0.10, N = 3SE +/- 0.04, N = 3SE +/- 0.16, N = 313.526.805.085.18MIN: 5.33 / MAX: 35.07MIN: 5.02 / MAX: 32.78MIN: 4.3 / MAX: 23.08MIN: 4.28 / MAX: 25.381. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: FastestDetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620Min: 6.24 / Avg: 13.52 / Max: 17.35Min: 6.65 / Avg: 6.8 / Max: 6.98Min: 5.01 / Avg: 5.08 / Max: 5.14Min: 4.91 / Avg: 5.18 / Max: 5.471. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: regnety_400mRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430SE +/- 0.25, N = 3SE +/- 2.01, N = 3SE +/- 0.04, N = 3SE +/- 0.02, N = 325.0423.135.465.32MIN: 11.25 / MAX: 44.6MIN: 8.74 / MAX: 39.68MIN: 4.78 / MAX: 28.36MIN: 4.68 / MAX: 22.841. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: regnety_400mRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 24.59 / Avg: 25.04 / Max: 25.45Min: 21.09 / Avg: 23.13 / Max: 27.15Min: 5.39 / Avg: 5.46 / Max: 5.52Min: 5.28 / Avg: 5.32 / Max: 5.351. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: resnet50RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny246810SE +/- 0.14, N = 3SE +/- 0.14, N = 3SE +/- 0.77, N = 3SE +/- 0.04, N = 36.355.857.285.16MIN: 5 / MAX: 21.95MIN: 4.75 / MAX: 26.01MIN: 3.88 / MAX: 30.89MIN: 4.78 / MAX: 21.851. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: resnet50RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3691215Min: 6.11 / Avg: 6.35 / Max: 6.58Min: 5.64 / Avg: 5.85 / Max: 6.12Min: 6.09 / Avg: 7.28 / Max: 8.73Min: 5.07 / Avg: 5.16 / Max: 5.221. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: vgg16RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430SE +/- 0.11, N = 3SE +/- 1.00, N = 3SE +/- 0.10, N = 3SE +/- 3.46, N = 34.407.0725.147.90MIN: 3.72 / MAX: 24.39MIN: 4.46 / MAX: 48.35MIN: 12.85 / MAX: 33.3MIN: 3.84 / MAX: 36.491. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: vgg16RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 4.18 / Avg: 4.4 / Max: 4.51Min: 5.18 / Avg: 7.07 / Max: 8.58Min: 25.03 / Avg: 25.14 / Max: 25.34Min: 4.43 / Avg: 7.9 / Max: 14.811. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: blazefaceRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1.23532.47063.70594.94126.1765SE +/- 0.22, N = 3SE +/- 0.07, N = 3SE +/- 0.08, N = 3SE +/- 0.10, N = 35.385.493.953.89MIN: 4.06 / MAX: 22.14MIN: 3.72 / MAX: 33.58MIN: 3.22 / MAX: 30.17MIN: 3.33 / MAX: 36.651. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: blazefaceRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny246810Min: 5.16 / Avg: 5.38 / Max: 5.81Min: 5.39 / Avg: 5.49 / Max: 5.62Min: 3.83 / Avg: 3.95 / Max: 4.1Min: 3.7 / Avg: 3.89 / Max: 4.031. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: efficientnet-b0RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430SE +/- 0.63, N = 3SE +/- 0.20, N = 3SE +/- 0.36, N = 3SE +/- 0.10, N = 323.9726.096.426.30MIN: 10.15 / MAX: 37.01MIN: 12.37 / MAX: 40.64MIN: 5.35 / MAX: 28.33MIN: 5.32 / MAX: 26.261. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: efficientnet-b0RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny612182430Min: 23.01 / Avg: 23.97 / Max: 25.15Min: 25.72 / Avg: 26.09 / Max: 26.4Min: 6.05 / Avg: 6.42 / Max: 7.14Min: 6.12 / Avg: 6.3 / Max: 6.481. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: mnasnetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025SE +/- 0.13, N = 3SE +/- 0.73, N = 3SE +/- 0.10, N = 2SE +/- 0.02, N = 319.167.034.394.37MIN: 7.92 / MAX: 33.04MIN: 4.55 / MAX: 25.62MIN: 3.79 / MAX: 22MIN: 3.71 / MAX: 23.071. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: mnasnetRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025Min: 18.91 / Avg: 19.16 / Max: 19.29Min: 5.78 / Avg: 7.03 / Max: 8.3Min: 4.29 / Avg: 4.39 / Max: 4.49Min: 4.33 / Avg: 4.37 / Max: 4.411. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: shufflenet-v2RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620SE +/- 0.06, N = 3SE +/- 0.28, N = 3SE +/- 0.04, N = 3SE +/- 0.01, N = 317.576.114.854.78MIN: 8.24 / MAX: 39.36MIN: 4.77 / MAX: 33.2MIN: 4.28 / MAX: 25.38MIN: 4.03 / MAX: 23.061. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU - Model: shufflenet-v2RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620Min: 17.46 / Avg: 17.57 / Max: 17.65Min: 5.83 / Avg: 6.11 / Max: 6.66Min: 4.78 / Avg: 4.85 / Max: 4.93Min: 4.76 / Avg: 4.78 / Max: 4.81. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025SE +/- 0.18, N = 3SE +/- 0.83, N = 3SE +/- 0.06, N = 3SE +/- 0.11, N = 318.7818.774.975.02MIN: 8.92 / MAX: 34.42MIN: 7.68 / MAX: 35.81MIN: 4.23 / MAX: 23.03MIN: 4.12 / MAX: 23.741. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025Min: 18.47 / Avg: 18.78 / Max: 19.08Min: 17.91 / Avg: 18.77 / Max: 20.43Min: 4.84 / Avg: 4.97 / Max: 5.04Min: 4.89 / Avg: 5.02 / Max: 5.231. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny48121620SE +/- 0.28, N = 3SE +/- 0.39, N = 3SE +/- 0.01, N = 3SE +/- 0.10, N = 318.256.354.424.32MIN: 7.4 / MAX: 39.06MIN: 4.5 / MAX: 27.11MIN: 3.68 / MAX: 20.27MIN: 3.68 / MAX: 23.771. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20220729Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2RTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny510152025Min: 17.69 / Avg: 18.25 / Max: 18.55Min: 5.84 / Avg: 6.35 / Max: 7.11Min: 4.4 / Avg: 4.42 / Max: 4.45Min: 4.12 / Avg: 4.32 / Max: 4.471. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

Blender

MinAvgMaxRTX 3090 24GB -Zotac46.067.379.0RTX 2080 Ti 22GB -Dell60.072.678.0RTX 4090 24GB -Nvidia42.044.850.0RTX 4080 16GB -Pny41.045.351.0OpenBenchmarking.orgCelsius, Fewer Is BetterBlender 3.6GPU Temperature Monitor20406080100

MinAvgMaxRTX 3090 24GB -Zotac19.2215.9349.9RTX 2080 Ti 22GB -Dell23.3173.8254.3RTX 4090 24GB -Nvidia7.0107.7242.2RTX 4080 16GB -Pny14.395.0198.6OpenBenchmarking.orgWatts, Fewer Is BetterBlender 3.6GPU Power Consumption Monitor100200300400500

OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: BMW27 - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny246810SE +/- 0.06, N = 15SE +/- 0.06, N = 15SE +/- 0.06, N = 15SE +/- 0.06, N = 156.138.933.704.43
OpenBenchmarking.orgSeconds, Fewer Is BetterBlender 3.6Blend File: BMW27 - Compute: NVIDIA OptiXRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny3691215Min: 6.05 / Avg: 6.13 / Max: 7.03Min: 8.85 / Avg: 8.93 / Max: 9.78Min: 3.61 / Avg: 3.7 / Max: 4.6Min: 4.35 / Avg: 4.43 / Max: 5.32

ViennaCL

MinAvgMaxRTX 3090 24GB -Zotac48.053.357.0RTX 2080 Ti 22GB -Dell44.049.258.0RTX 4090 24GB -Nvidia32.033.937.0RTX 4080 16GB -Pny34.035.738.0OpenBenchmarking.orgCelsius, Fewer Is BetterViennaCL 1.7.1GPU Temperature Monitor1632486480

MinAvgMaxRTX 3090 24GB -Zotac19.026.237.2RTX 2080 Ti 22GB -Dell18.421.643.4RTX 4090 24GB -Nvidia6.08.015.8RTX 4080 16GB -Pny4.012.118.8OpenBenchmarking.orgWatts, Fewer Is BetterViennaCL 1.7.1GPU Power Consumption Monitor1224364860

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-TTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150SE +/- 5.83, N = 15SE +/- 4.93, N = 15SE +/- 9.73, N = 5SE +/- 6.46, N = 111431451441371. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-TTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150Min: 115 / Avg: 143.47 / Max: 168Min: 119 / Avg: 144.73 / Max: 171Min: 123 / Avg: 144.2 / Max: 171Min: 119 / Avg: 137 / Max: 1721. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-TNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150SE +/- 5.51, N = 15SE +/- 4.47, N = 15SE +/- 9.50, N = 5SE +/- 4.72, N = 121411421371351. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-TNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150Min: 118 / Avg: 141 / Max: 173Min: 121 / Avg: 142.47 / Max: 175Min: 121 / Avg: 136.8 / Max: 174Min: 122 / Avg: 135 / Max: 1741. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-NTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150SE +/- 4.37, N = 15SE +/- 2.19, N = 15SE +/- 0.48, N = 4SE +/- 4.18, N = 121201191151221. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-NTRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20406080100Min: 105 / Avg: 120 / Max: 156Min: 112 / Avg: 119.13 / Max: 136Min: 114 / Avg: 115.25 / Max: 116Min: 108 / Avg: 121.75 / Max: 1621. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-NNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150SE +/- 1.05, N = 15SE +/- 1.59, N = 15SE +/- 10.59, N = 5SE +/- 1.29, N = 12107.0106.8120.0103.51. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-NNRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny20406080100Min: 101 / Avg: 106.93 / Max: 114Min: 98.8 / Avg: 106.78 / Max: 117Min: 106 / Avg: 120 / Max: 162Min: 98 / Avg: 103.52 / Max: 1101. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMV-NRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny4080120160200SE +/- 8.48, N = 15SE +/- 1.84, N = 15SE +/- 1.74, N = 5SE +/- 0.91, N = 121721791881871. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL
OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMV-NRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny306090120150Min: 55 / Avg: 171.87 / Max: 197Min: 159 / Avg: 178.67 / Max: 192Min: 184 / Avg: 188.2 / Max: 194Min: 181 / Avg: 186.75 / Max: 1931. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

Rodinia

MinAvgMaxRTX 3090 24GB -Zotac52.064.977.0RTX 2080 Ti 22GB -Dell62.069.274.0RTX 4090 24GB -Nvidia41.044.650.0RTX 4080 16GB -Pny39.043.849.0OpenBenchmarking.orgCelsius, Fewer Is BetterRodinia 3.1GPU Temperature Monitor20406080100

MinAvgMaxRTX 3090 24GB -Zotac19.9148.1254.9RTX 2080 Ti 22GB -Dell23.3133.0219.9RTX 4090 24GB -Nvidia6.867.5167.3RTX 4080 16GB -Pny14.563.1123.6OpenBenchmarking.orgWatts, Fewer Is BetterRodinia 3.1GPU Power Consumption Monitor70140210280350

OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 3.1Test: OpenCL Particle FilterRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny1.01862.03723.05584.07445.093SE +/- 0.029, N = 10SE +/- 0.031, N = 8SE +/- 0.080, N = 3SE +/- 0.005, N = 103.7684.5272.0852.6851. (CXX) g++ options: -m64 -lm -lcuda -lcudart -lcudadevrt -lcudart_static -lrt -lpthread -ldl
OpenBenchmarking.orgSeconds, Fewer Is BetterRodinia 3.1Test: OpenCL Particle FilterRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny246810Min: 3.7 / Avg: 3.77 / Max: 4.01Min: 4.46 / Avg: 4.53 / Max: 4.74Min: 2 / Avg: 2.09 / Max: 2.25Min: 2.66 / Avg: 2.69 / Max: 2.721. (CXX) g++ options: -m64 -lm -lcuda -lcudart -lcudadevrt -lcudart_static -lrt -lpthread -ldl

NAMD CUDA

MinAvgMaxRTX 3090 24GB -Zotac47.054.378.0RTX 2080 Ti 22GB -Dell60.067.374.0RTX 4090 24GB -Nvidia34.037.344.0RTX 4080 16GB -Pny36.039.552.0OpenBenchmarking.orgCelsius, Fewer Is BetterNAMD CUDA 2.14GPU Temperature Monitor20406080100

MinAvgMaxRTX 3090 24GB -Zotac19.068.9327.9RTX 2080 Ti 22GB -Dell23.6127.2297.8RTX 4090 24GB -Nvidia6.454.4281.7RTX 4080 16GB -Pny13.469.1263.4OpenBenchmarking.orgWatts, Fewer Is BetterNAMD CUDA 2.14GPU Power Consumption Monitor80160240320400

OpenBenchmarking.orgdays/ns, Fewer Is BetterNAMD CUDA 2.14ATPase Simulation - 327,506 AtomsRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny0.02140.04280.06420.08560.107SE +/- 0.00094, N = 3SE +/- 0.00074, N = 15SE +/- 0.00119, N = 15SE +/- 0.00113, N = 150.067420.095150.048500.05427
OpenBenchmarking.orgdays/ns, Fewer Is BetterNAMD CUDA 2.14ATPase Simulation - 327,506 AtomsRTX 3090 24GB -ZotacRTX 2080 Ti 22GB -DellRTX 4090 24GB -NvidiaRTX 4080 16GB -Pny12345Min: 0.07 / Avg: 0.07 / Max: 0.07Min: 0.09 / Avg: 0.1 / Max: 0.1Min: 0.05 / Avg: 0.05 / Max: 0.06Min: 0.05 / Avg: 0.05 / Max: 0.06

116 Results Shown

NCNN
clpeak
SHOC Scalable HeterOgeneous Computing:
  OpenCL - GEMM SGEMM_N
  OpenCL - Max SP Flops
NCNN:
  Vulkan GPU - googlenet
  Vulkan GPU - resnet18
Chaos Group V-RAY
NCNN
Chaos Group V-RAY
NCNN
OctaneBench
vkpeak
clpeak
Hashcat
Blender
Hashcat
IndigoBench
Hashcat
FinanceBench
Blender:
  Barbershop - NVIDIA OptiX
  Classroom - NVIDIA OptiX
Hashcat
vkpeak
Hashcat
Blender
ViennaCL
SHOC Scalable HeterOgeneous Computing
vkpeak:
  int16-scalar
  fp16-vec4
  fp16-scalar
GROMACS
ViennaCL
vkpeak
SHOC Scalable HeterOgeneous Computing
vkpeak
ViennaCL
vkpeak:
  fp32-scalar
  int32-scalar
  fp64-vec4
clpeak
VkResample
SHOC Scalable HeterOgeneous Computing
RealSR-NCNN
ViennaCL
IndigoBench
SHOC Scalable HeterOgeneous Computing
MandelGPU
SHOC Scalable HeterOgeneous Computing:
  OpenCL - Bus Speed Readback
  OpenCL - Triad
NeatBench
SHOC Scalable HeterOgeneous Computing
ArrayFire
VkResample
SHOC Scalable HeterOgeneous Computing
RealSR-NCNN
VkFFT
cl-mem
Waifu2x-NCNN Vulkan
clpeak
cl-mem
ViennaCL
Caffe:
  AlexNet - NVIDIA CUDA - 1000
  AlexNet - NVIDIA CUDA - 200
ViennaCL
NCNN
Caffe:
  AlexNet - NVIDIA CUDA - 100
  GoogleNet - NVIDIA CUDA - 1000
  GoogleNet - NVIDIA CUDA - 200
  GoogleNet - NVIDIA CUDA - 100
ViennaCL:
  OpenCL BLAS - sDOT
  OpenCL BLAS - sCOPY
FAHBench
ViennaCL:
  OpenCL BLAS - sAXPY
  OpenCL BLAS - dCOPY
  OpenCL BLAS - dDOT
cl-mem
NCNN
ViennaCL
LeelaChessZero
ViennaCL:
  CPU BLAS - dCOPY
  CPU BLAS - dDOT
  CPU BLAS - dAXPY
  CPU BLAS - sCOPY
  CPU BLAS - sAXPY
  CPU BLAS - dGEMV-T
  CPU BLAS - sDOT
GPU Temperature Monitor:
  Phoronix Test Suite System Monitoring:
    Celsius
    Watts
  GPU Temp Monitor:
    Celsius
  GPU Power Consumption Monitor:
    Watts
NCNN:
  Vulkan GPU - FastestDet
  Vulkan GPU - regnety_400m
  Vulkan GPU - resnet50
  Vulkan GPU - vgg16
  Vulkan GPU - blazeface
  Vulkan GPU - efficientnet-b0
  Vulkan GPU - mnasnet
  Vulkan GPU - shufflenet-v2
  Vulkan GPU-v3-v3 - mobilenet-v3
  Vulkan GPU-v2-v2 - mobilenet-v2
Blender:
  GPU Temp Monitor
  GPU Power Consumption Monitor
Blender
ViennaCL:
  GPU Temp Monitor
  GPU Power Consumption Monitor
ViennaCL:
  CPU BLAS - dGEMM-TT
  CPU BLAS - dGEMM-TN
  CPU BLAS - dGEMM-NT
  CPU BLAS - dGEMM-NN
  CPU BLAS - dGEMV-N
Rodinia:
  GPU Temp Monitor
  GPU Power Consumption Monitor
Rodinia
NAMD CUDA:
  GPU Temp Monitor
  GPU Power Consumption Monitor
NAMD CUDA