ngc smoke run

ARMv8 Neoverse-V2 testing with a Quanta Cloud QuantaGrid S74G-2U 1S7GZ9Z0000 S7G MB (CG1) (3A06 BIOS) and NVIDIA GH200 480GB on Ubuntu 22.04 via the Phoronix Test Suite.

Compare your own system(s) to this result file with the Phoronix Test Suite by running the command: phoronix-test-suite benchmark 2403013-NE-NGCSMOKER54
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
Show Result Confidence Charts
Allow Limiting Results To Certain Suite(s)

Statistics

Show Overall Harmonic Mean(s)
Show Overall Geometric Mean
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
Prefer Vertical Bar Graphs

Multi-Way Comparison

Condense Multi-Option Tests Into Single Result Graphs

Table

Show Detailed System Result Table

Run Management

Highlight
Result
Toggle/Hide
Result
Result
Identifier
Performance Per
Dollar
Date
Run
  Test
  Duration
a
February 29
  1 Hour, 40 Minutes
b
March 01
  1 Hour, 44 Minutes
c
March 01
  1 Hour, 33 Minutes
d
March 01
  1 Hour, 59 Minutes
Invert Behavior (Only Show Selected Data)
  1 Hour, 44 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):

a

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Graphics Notes: BAR1 / Visible vRAM Size: N/A - vBIOS Version: 96.00.7e.00.02
Python Notes: Python 3.10.12
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

b

c

d

Processor: ARMv8 Neoverse-V2 @ 3.39GHz (72 Cores), Motherboard: Quanta Cloud QuantaGrid S74G-2U 1S7GZ9Z0000 S7G MB (CG1) (3A06 BIOS), Memory: 1 x 480GB DRAM-6400MT/s, Disk: 960GB SAMSUNG MZ1L2960HCJR-00A07 + 1920GB SAMSUNG MZTL21T9, Graphics: NVIDIA GH200 480GB, Network: 2 x Mellanox MT2910 + 2 x QLogic FastLinQ QL41000 10/25/40/50GbE

OS: Ubuntu 22.04, Kernel: 6.5.0-1007-NVIDIA-64k (aarch64), Display Driver: NVIDIA, OpenCL: OpenCL 3.0 CUDA 12.4.89, Compiler: GCC 11.4.0 + CUDA 11.5, File-System: ext4, Screen Resolution: 1920x1200

ngc smoke runOpenBenchmarking.orgPhoronix Test SuiteARMv8 Neoverse-V2 @ 3.39GHz (72 Cores)Quanta Cloud QuantaGrid S74G-2U 1S7GZ9Z0000 S7G MB (CG1) (3A06 BIOS)1 x 480GB DRAM-6400MT/s960GB SAMSUNG MZ1L2960HCJR-00A07 + 1920GB SAMSUNG MZTL21T9NVIDIA GH200 480GB2 x Mellanox MT2910 + 2 x QLogic FastLinQ QL41000 10/25/40/50GbEUbuntu 22.046.5.0-1007-NVIDIA-64k (aarch64)NVIDIAOpenCL 3.0 CUDA 12.4.89GCC 11.4.0 + CUDA 11.5ext41920x1200ProcessorMotherboardMemoryDiskGraphicsNetworkOSKernelDisplay DriverOpenCLCompilerFile-SystemScreen ResolutionNgc Smoke Run BenchmarksSystem Logs- Transparent Huge Pages: madvise- --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v - Scaling Governor: cppc_cpufreq performance (Boost: Disabled)- BAR1 / Visible vRAM Size: N/A - vBIOS Version: 96.00.7e.00.02- Python 3.10.12- gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

abcdResult OverviewPhoronix Test Suite100%100%101%101%VkFFTFinanceBenchNCNNArrayFirecl-memViennaCLclpeakVkResample

ngc smoke runvkfft: FFT + iFFT R2C / C2Rvkfft: FFT + iFFT C2C 1D batched in half precisionvkfft: FFT + iFFT C2C Bluestein in single precisionvkfft: FFT + iFFT C2C 1D batched in double precisionvkfft: FFT + iFFT C2C 1D batched in single precisionvkfft: FFT + iFFT C2C multidimensional in single precisionvkfft: FFT + iFFT C2C Bluestein benchmark in double precisionvkfft: FFT + iFFT C2C 1D batched in single precision, no reshufflingcl-mem: Copycl-mem: Readcl-mem: Writevkresample: 2x - Doublevkresample: 2x - Singleclpeak: Integer Compute INTclpeak: Single-Precision Floatclpeak: Double-Precision Doubleclpeak: Global Memory Bandwidtharrayfire: Conjugate Gradient OpenCLfinancebench: Black-Scholes OpenCLviennacl: CPU BLAS - sCOPYviennacl: CPU BLAS - sAXPYviennacl: CPU BLAS - sDOTviennacl: CPU BLAS - dCOPYviennacl: CPU BLAS - dAXPYviennacl: CPU BLAS - dDOTviennacl: CPU BLAS - dGEMV-Nviennacl: CPU BLAS - dGEMV-Tviennacl: CPU BLAS - dGEMM-NNviennacl: CPU BLAS - dGEMM-NTviennacl: CPU BLAS - dGEMM-TNviennacl: CPU BLAS - dGEMM-TTviennacl: OpenCL BLAS - sCOPYviennacl: OpenCL BLAS - sAXPYviennacl: OpenCL BLAS - sDOTviennacl: OpenCL BLAS - dCOPYviennacl: OpenCL BLAS - dAXPYviennacl: OpenCL BLAS - dDOTviennacl: OpenCL BLAS - dGEMV-Nviennacl: OpenCL BLAS - dGEMV-Tviennacl: OpenCL BLAS - dGEMM-NNviennacl: OpenCL BLAS - dGEMM-NTviennacl: OpenCL BLAS - dGEMM-TNviennacl: OpenCL BLAS - dGEMM-TTncnn: Vulkan GPU - mobilenetncnn: Vulkan GPU-v2-v2 - mobilenet-v2ncnn: Vulkan GPU-v3-v3 - mobilenet-v3ncnn: Vulkan GPU - shufflenet-v2ncnn: Vulkan GPU - mnasnetncnn: Vulkan GPU - efficientnet-b0ncnn: Vulkan GPU - blazefacencnn: Vulkan GPU - googlenetncnn: Vulkan GPU - vgg16ncnn: Vulkan GPU - resnet18ncnn: Vulkan GPU - alexnetncnn: Vulkan GPU - resnet50ncnn: Vulkan GPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3ncnn: Vulkan GPU - yolov4-tinyncnn: Vulkan GPU - squeezenet_ssdncnn: Vulkan GPU - regnety_400mncnn: Vulkan GPU - vision_transformerncnn: Vulkan GPU - FastestDetabcd4239715191217867584051857744448920810194497308.61045.92354.924.2965.23033119.1064545.6232959.173483.992.9974.3472920394366720271803124741168613512514113731642028260379955081.230870577527702770704.892.132.262.292.043.491.754.165.262.161.634.274.896.795.4314.7831.523.094180915191017967582531860824373121000190037308.51045.92353.424.2945.23133144.7464547.7432961.213484.062.9834.3732892392466419481806123840869913712514013831642728260479855281.530870937537706770704.922.162.272.292.033.551.784.235.252.181.634.284.926.805.4714.7432.323.104258115286617886582561899444507121094190909308.61046.12354.924.2905.23033146.1264547.2532941.993483.952.9984.3512907391766619201837124340569113912414114031642728360479955281.230870377537700070574.922.122.302.272.043.521.744.235.252.171.654.324.926.815.4814.7731.923.084304815196917942582991903104500721320192507308.51046.02352.124.2975.23033129.3464520.9732933.633484.322.9974.3392857392066319171830124741869614112414013631642628360479955381.430770537540707070704.912.122.272.272.043.531.774.215.262.201.624.324.916.825.4315.2231.133.12OpenBenchmarking.org

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 1.3.4Test: FFT + iFFT R2C / C2Rabcd9K18K27K36K45KSE +/- 298.99, N = 3SE +/- 460.34, N = 3SE +/- 552.29, N = 3SE +/- 289.59, N = 15423974180942581430481. (CXX) g++ options: -O3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.3.4Test: FFT + iFFT C2C 1D batched in half precisionabcd30K60K90K120K150KSE +/- 190.55, N = 3SE +/- 506.81, N = 3SE +/- 136.79, N = 3SE +/- 377.55, N = 31519121519101528661519691. (CXX) g++ options: -O3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.3.4Test: FFT + iFFT C2C Bluestein in single precisionabcd4K8K12K16K20KSE +/- 131.79, N = 3SE +/- 196.01, N = 5SE +/- 147.24, N = 3SE +/- 168.45, N = 7178671796717886179421. (CXX) g++ options: -O3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.3.4Test: FFT + iFFT C2C 1D batched in double precisionabcd13K26K39K52K65KSE +/- 150.19, N = 3SE +/- 46.92, N = 3SE +/- 17.34, N = 3SE +/- 21.83, N = 3584055825358256582991. (CXX) g++ options: -O3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.3.4Test: FFT + iFFT C2C 1D batched in single precisionabcd40K80K120K160K200KSE +/- 1557.78, N = 3SE +/- 1095.80, N = 3SE +/- 1666.00, N = 3SE +/- 479.86, N = 31857741860821899441903101. (CXX) g++ options: -O3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.3.4Test: FFT + iFFT C2C multidimensional in single precisionabcd10K20K30K40K50KSE +/- 479.16, N = 15SE +/- 441.36, N = 3SE +/- 475.72, N = 3SE +/- 571.37, N = 3444894373145071450071. (CXX) g++ options: -O3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.3.4Test: FFT + iFFT C2C Bluestein benchmark in double precisionabcd5K10K15K20K25KSE +/- 188.78, N = 3SE +/- 195.51, N = 3SE +/- 282.81, N = 3SE +/- 152.14, N = 15208102100021094213201. (CXX) g++ options: -O3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.3.4Test: FFT + iFFT C2C 1D batched in single precision, no reshufflingabcd40K80K120K160K200KSE +/- 2261.14, N = 3SE +/- 521.00, N = 3SE +/- 583.76, N = 3SE +/- 720.67, N = 31944971900371909091925071. (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: Copyabcd70140210280350SE +/- 0.03, N = 3SE +/- 0.12, N = 3SE +/- 0.07, N = 3SE +/- 0.03, N = 3308.6308.5308.6308.51. (CC) gcc options: -O2 -flto -lOpenCL

OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: Readabcd2004006008001000SE +/- 0.00, N = 3SE +/- 0.20, N = 3SE +/- 0.03, N = 3SE +/- 0.03, N = 31045.91045.91046.11046.01. (CC) gcc options: -O2 -flto -lOpenCL

OpenBenchmarking.orgGB/s, More Is Bettercl-mem 2017-01-13Benchmark: Writeabcd5001000150020002500SE +/- 1.31, N = 3SE +/- 0.88, N = 3SE +/- 0.99, N = 3SE +/- 3.80, N = 32354.92353.42354.92352.11. (CC) gcc options: -O2 -flto -lOpenCL

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: Doubleabcd612182430SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 324.3024.2924.2924.301. (CXX) g++ options: -O3

OpenBenchmarking.orgms, Fewer Is BetterVkResample 1.0Upscale: 2x - Precision: Singleabcd1.1772.3543.5314.7085.885SE +/- 0.000, N = 3SE +/- 0.001, N = 3SE +/- 0.001, N = 3SE +/- 0.002, N = 35.2305.2315.2305.2301. (CXX) g++ options: -O3

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 INTabcd7K14K21K28K35KSE +/- 2.54, N = 3SE +/- 0.09, N = 3SE +/- 0.26, N = 3SE +/- 8.15, N = 333119.1033144.7433146.1233129.341. (CXX) g++ options: -O3

OpenBenchmarking.orgGFLOPS, More Is Betterclpeak 1.1.2OpenCL Test: Single-Precision Floatabcd14K28K42K56K70KSE +/- 0.43, N = 3SE +/- 0.85, N = 3SE +/- 0.56, N = 3SE +/- 0.91, N = 364545.6264547.7464547.2564520.971. (CXX) g++ options: -O3

OpenBenchmarking.orgGFLOPS, More Is Betterclpeak 1.1.2OpenCL Test: Double-Precision Doubleabcd7K14K21K28K35KSE +/- 0.74, N = 3SE +/- 0.74, N = 3SE +/- 18.62, N = 3SE +/- 1.51, N = 332959.1732961.2132941.9932933.631. (CXX) g++ options: -O3

OpenBenchmarking.orgGBPS, More Is Betterclpeak 1.1.2OpenCL Test: Global Memory Bandwidthabcd7001400210028003500SE +/- 0.33, N = 3SE +/- 0.20, N = 3SE +/- 0.27, N = 3SE +/- 0.04, N = 33483.993484.063483.953484.321. (CXX) g++ options: -O3

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 3.9Test: Conjugate Gradient OpenCLabcd0.67461.34922.02382.69843.373SE +/- 0.003, N = 3SE +/- 0.005, N = 3SE +/- 0.005, N = 3SE +/- 0.003, N = 32.9972.9832.9982.9971. (CXX) g++ options: -O3

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 OpenCLabcd0.98391.96782.95173.93564.9195SE +/- 0.010, N = 3SE +/- 0.004, N = 3SE +/- 0.010, N = 3SE +/- 0.016, N = 34.3474.3734.3514.3391. (CXX) g++ options: -O3 -march=native -fopenmp

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 - sCOPYabcd6001200180024003000SE +/- 20.00, N = 3SE +/- 29.56, N = 5SE +/- 3.33, N = 3SE +/- 23.33, N = 329202892290728571. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sAXPYabcd8001600240032004000SE +/- 14.53, N = 3SE +/- 9.80, N = 5SE +/- 16.67, N = 3SE +/- 15.28, N = 339433924391739201. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - sDOTabcd140280420560700SE +/- 4.18, N = 3SE +/- 5.77, N = 5SE +/- 4.33, N = 3SE +/- 3.51, N = 36676646666631. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dCOPYabcd400800120016002000SE +/- 44.85, N = 3SE +/- 17.15, N = 5SE +/- 41.63, N = 3SE +/- 14.53, N = 320271948192019171. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dAXPYabcd400800120016002000SE +/- 23.33, N = 3SE +/- 29.93, N = 5SE +/- 3.33, N = 3SE +/- 10.00, N = 318031806183718301. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dDOTabcd30060090012001500SE +/- 3.33, N = 3SE +/- 2.00, N = 5SE +/- 3.33, N = 3SE +/- 3.33, N = 312471238124312471. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMV-Nabcd90180270360450SE +/- 0.33, N = 3SE +/- 2.90, N = 5SE +/- 1.86, N = 3SE +/- 8.51, N = 34114084054181. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMV-Tabcd150300450600750SE +/- 17.19, N = 3SE +/- 3.65, N = 5SE +/- 1.20, N = 3SE +/- 10.68, N = 36866996916961. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-NNabcd306090120150SE +/- 0.88, N = 3SE +/- 1.29, N = 5SE +/- 2.67, N = 3SE +/- 2.65, N = 31351371391411. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-NTabcd306090120150SE +/- 0.88, N = 3SE +/- 0.77, N = 5SE +/- 0.33, N = 3SE +/- 0.33, N = 31251251241241. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-TNabcd306090120150SE +/- 1.20, N = 3SE +/- 0.93, N = 5SE +/- 0.58, N = 3SE +/- 0.58, N = 31411401411401. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: CPU BLAS - dGEMM-TTabcd306090120150SE +/- 1.53, N = 3SE +/- 2.18, N = 5SE +/- 4.73, N = 3SE +/- 1.20, N = 31371381401361. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sCOPYabcd70140210280350SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 33163163163161. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sAXPYabcd90180270360450SE +/- 2.60, N = 3SE +/- 1.33, N = 3SE +/- 2.33, N = 3SE +/- 3.93, N = 34204274274261. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - sDOTabcd60120180240300SE +/- 0.00, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 0.67, N = 32822822832831. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dCOPYabcd130260390520650SE +/- 0.33, N = 3SE +/- 0.58, N = 3SE +/- 0.88, N = 3SE +/- 0.58, N = 36036046046041. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dAXPYabcd2004006008001000SE +/- 0.33, N = 3SE +/- 0.33, N = 3SE +/- 1.20, N = 3SE +/- 0.88, N = 37997987997991. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dDOTabcd120240360480600SE +/- 0.88, N = 3SE +/- 0.33, N = 3SE +/- 1.15, N = 3SE +/- 1.00, N = 35505525525531. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMV-Nabcd20406080100SE +/- 0.13, N = 3SE +/- 0.12, N = 3SE +/- 0.26, N = 3SE +/- 0.21, N = 381.281.581.281.41. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGB/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMV-Tabcd70140210280350SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.33, N = 3SE +/- 0.33, N = 33083083083071. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-NNabcd15003000450060007500SE +/- 31.80, N = 3SE +/- 58.97, N = 3SE +/- 31.80, N = 3SE +/- 35.28, N = 370577093703770531. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-NTabcd16003200480064008000SE +/- 3.33, N = 3SE +/- 8.82, N = 3SE +/- 3.33, N = 3SE +/- 0.00, N = 375277537753775401. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-TNabcd15003000450060007500SE +/- 17.64, N = 3SE +/- 46.31, N = 3SE +/- 15.28, N = 3SE +/- 45.09, N = 370277067700070701. (CXX) g++ options: -fopenmp -O3 -rdynamic -lOpenCL

OpenBenchmarking.orgGFLOPs/s, More Is BetterViennaCL 1.7.1Test: OpenCL BLAS - dGEMM-TTabcd15003000450060007500SE +/- 0.00, N = 3SE +/- 11.55, N = 3SE +/- 3.33, N = 3SE +/- 0.00, N = 370707070705770701. (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 20230517Target: Vulkan GPU - Model: mobilenetabcd1.1072.2143.3214.4285.535SE +/- 0.02, N = 3SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 34.894.924.924.91MIN: 4.76 / MAX: 7.99MIN: 4.77 / MAX: 7.57MIN: 4.74 / MAX: 6.99MIN: 4.79 / MAX: 6.531. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2abcd0.4860.9721.4581.9442.43SE +/- 0.03, N = 3SE +/- 0.02, N = 3SE +/- 0.04, N = 3SE +/- 0.05, N = 32.132.162.122.12MIN: 1.99 / MAX: 3.7MIN: 2.04 / MAX: 8.11MIN: 1.96 / MAX: 3.59MIN: 1.96 / MAX: 5.51. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3abcd0.51751.0351.55252.072.5875SE +/- 0.03, N = 3SE +/- 0.01, N = 3SE +/- 0.05, N = 3SE +/- 0.04, N = 32.262.272.302.27MIN: 2.11 / MAX: 3.89MIN: 2.16 / MAX: 5.3MIN: 2.1 / MAX: 3.74MIN: 2.11 / MAX: 4.591. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: shufflenet-v2abcd0.51531.03061.54592.06122.5765SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.04, N = 32.292.292.272.27MIN: 2.13 / MAX: 3.94MIN: 2.12 / MAX: 3.59MIN: 2.13 / MAX: 5.55MIN: 2.1 / MAX: 5.571. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: mnasnetabcd0.4590.9181.3771.8362.295SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.04, N = 3SE +/- 0.03, N = 32.042.032.042.04MIN: 1.89 / MAX: 3.56MIN: 1.94 / MAX: 3.48MIN: 1.86 / MAX: 3.85MIN: 1.87 / MAX: 6.481. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: efficientnet-b0abcd0.79881.59762.39643.19523.994SE +/- 0.02, N = 3SE +/- 0.03, N = 3SE +/- 0.04, N = 3SE +/- 0.04, N = 33.493.553.523.53MIN: 3.27 / MAX: 5.07MIN: 3.33 / MAX: 8.63MIN: 3.22 / MAX: 6.65MIN: 3.25 / MAX: 6.521. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: blazefaceabcd0.40050.8011.20151.6022.0025SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.03, N = 3SE +/- 0.02, N = 31.751.781.741.77MIN: 1.68 / MAX: 3.09MIN: 1.67 / MAX: 7.1MIN: 1.6 / MAX: 2.93MIN: 1.64 / MAX: 3.011. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: googlenetabcd0.95181.90362.85543.80724.759SE +/- 0.02, N = 3SE +/- 0.05, N = 3SE +/- 0.06, N = 3SE +/- 0.01, N = 34.164.234.234.21MIN: 3.99 / MAX: 5.76MIN: 4.01 / MAX: 6.82MIN: 4 / MAX: 6.41MIN: 4.03 / MAX: 5.751. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: vgg16abcd1.18352.3673.55054.7345.9175SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.03, N = 3SE +/- 0.02, N = 35.265.255.255.26MIN: 5.08 / MAX: 8.11MIN: 5.07 / MAX: 7.45MIN: 4.94 / MAX: 11.68MIN: 5.07 / MAX: 7.081. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: resnet18abcd0.4950.991.4851.982.475SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.01, N = 32.162.182.172.20MIN: 2.04 / MAX: 3.59MIN: 2.05 / MAX: 5.44MIN: 2.04 / MAX: 3.54MIN: 2.08 / MAX: 3.651. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: alexnetabcd0.37130.74261.11391.48521.8565SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.04, N = 3SE +/- 0.00, N = 31.631.631.651.62MIN: 1.49 / MAX: 2.81MIN: 1.5 / MAX: 2.94MIN: 1.44 / MAX: 4.74MIN: 1.5 / MAX: 4.841. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: resnet50abcd0.9721.9442.9163.8884.86SE +/- 0.02, N = 3SE +/- 0.03, N = 3SE +/- 0.04, N = 3SE +/- 0.02, N = 34.274.284.324.32MIN: 4.05 / MAX: 6.65MIN: 4.05 / MAX: 7.65MIN: 4.05 / MAX: 8.1MIN: 4.1 / MAX: 7.581. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPUv2-yolov3v2-yolov3 - Model: mobilenetv2-yolov3abcd1.1072.2143.3214.4285.535SE +/- 0.02, N = 3SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 34.894.924.924.91MIN: 4.76 / MAX: 7.99MIN: 4.77 / MAX: 7.57MIN: 4.74 / MAX: 6.99MIN: 4.79 / MAX: 6.531. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: yolov4-tinyabcd246810SE +/- 0.01, N = 3SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.02, N = 36.796.806.816.82MIN: 6.66 / MAX: 8.49MIN: 6.64 / MAX: 8.25MIN: 6.42 / MAX: 12.5MIN: 6.69 / MAX: 8.11. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: squeezenet_ssdabcd1.2332.4663.6994.9326.165SE +/- 0.03, N = 3SE +/- 0.07, N = 3SE +/- 0.07, N = 3SE +/- 0.04, N = 35.435.475.485.43MIN: 5.18 / MAX: 8.53MIN: 5.22 / MAX: 11.66MIN: 5.16 / MAX: 11.04MIN: 5.17 / MAX: 7.391. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: regnety_400mabcd48121620SE +/- 0.13, N = 3SE +/- 0.15, N = 3SE +/- 0.24, N = 3SE +/- 0.19, N = 314.7814.7414.7715.22MIN: 13.74 / MAX: 17.76MIN: 14 / MAX: 20.37MIN: 13.51 / MAX: 18.11MIN: 14.15 / MAX: 21.541. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: vision_transformerabcd816243240SE +/- 0.28, N = 3SE +/- 0.98, N = 3SE +/- 0.60, N = 3SE +/- 0.08, N = 331.5232.3231.9231.13MIN: 30.23 / MAX: 62.77MIN: 30.14 / MAX: 67.5MIN: 30.27 / MAX: 57.34MIN: 30.31 / MAX: 64.211. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: FastestDetabcd0.7021.4042.1062.8083.51SE +/- 0.02, N = 3SE +/- 0.04, N = 3SE +/- 0.06, N = 3SE +/- 0.03, N = 33.093.103.083.12MIN: 2.95 / MAX: 4.64MIN: 2.92 / MAX: 4.71MIN: 2.91 / MAX: 4.57MIN: 2.97 / MAX: 4.561. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

61 Results Shown

VkFFT:
  FFT + iFFT R2C / C2R
  FFT + iFFT C2C 1D batched in half precision
  FFT + iFFT C2C Bluestein in single precision
  FFT + iFFT C2C 1D batched in double precision
  FFT + iFFT C2C 1D batched in single precision
  FFT + iFFT C2C multidimensional in single precision
  FFT + iFFT C2C Bluestein benchmark in double precision
  FFT + iFFT C2C 1D batched in single precision, no reshuffling
cl-mem:
  Copy
  Read
  Write
VkResample:
  2x - Double
  2x - Single
clpeak:
  Integer Compute INT
  Single-Precision Float
  Double-Precision Double
  Global Memory Bandwidth
ArrayFire
FinanceBench
ViennaCL:
  CPU BLAS - sCOPY
  CPU BLAS - sAXPY
  CPU BLAS - sDOT
  CPU BLAS - dCOPY
  CPU BLAS - dAXPY
  CPU BLAS - dDOT
  CPU BLAS - dGEMV-N
  CPU BLAS - dGEMV-T
  CPU BLAS - dGEMM-NN
  CPU BLAS - dGEMM-NT
  CPU BLAS - dGEMM-TN
  CPU BLAS - dGEMM-TT
  OpenCL BLAS - sCOPY
  OpenCL BLAS - sAXPY
  OpenCL BLAS - sDOT
  OpenCL BLAS - dCOPY
  OpenCL BLAS - dAXPY
  OpenCL BLAS - dDOT
  OpenCL BLAS - dGEMV-N
  OpenCL BLAS - dGEMV-T
  OpenCL BLAS - dGEMM-NN
  OpenCL BLAS - dGEMM-NT
  OpenCL BLAS - dGEMM-TN
  OpenCL BLAS - dGEMM-TT
NCNN:
  Vulkan GPU - mobilenet
  Vulkan GPU-v2-v2 - mobilenet-v2
  Vulkan GPU-v3-v3 - mobilenet-v3
  Vulkan GPU - shufflenet-v2
  Vulkan GPU - mnasnet
  Vulkan GPU - efficientnet-b0
  Vulkan GPU - blazeface
  Vulkan GPU - googlenet
  Vulkan GPU - vgg16
  Vulkan GPU - resnet18
  Vulkan GPU - alexnet
  Vulkan GPU - resnet50
  Vulkan GPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3
  Vulkan GPU - yolov4-tiny
  Vulkan GPU - squeezenet_ssd
  Vulkan GPU - regnety_400m
  Vulkan GPU - vision_transformer
  Vulkan GPU - FastestDet

a

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Graphics Notes: BAR1 / Visible vRAM Size: N/A - vBIOS Version: 96.00.7e.00.02
Python Notes: Python 3.10.12
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 29 February 2024 22:50 by user x.

b

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Graphics Notes: BAR1 / Visible vRAM Size: N/A - vBIOS Version: 96.00.7e.00.02
Python Notes: Python 3.10.12
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 1 March 2024 00:08 by user x.

c

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Graphics Notes: BAR1 / Visible vRAM Size: N/A - vBIOS Version: 96.00.7e.00.02
Python Notes: Python 3.10.12
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 1 March 2024 02:24 by user x.

d

Processor: ARMv8 Neoverse-V2 @ 3.39GHz (72 Cores), Motherboard: Quanta Cloud QuantaGrid S74G-2U 1S7GZ9Z0000 S7G MB (CG1) (3A06 BIOS), Memory: 1 x 480GB DRAM-6400MT/s, Disk: 960GB SAMSUNG MZ1L2960HCJR-00A07 + 1920GB SAMSUNG MZTL21T9, Graphics: NVIDIA GH200 480GB, Network: 2 x Mellanox MT2910 + 2 x QLogic FastLinQ QL41000 10/25/40/50GbE

OS: Ubuntu 22.04, Kernel: 6.5.0-1007-NVIDIA-64k (aarch64), Display Driver: NVIDIA, OpenCL: OpenCL 3.0 CUDA 12.4.89, Compiler: GCC 11.4.0 + CUDA 11.5, File-System: ext4, Screen Resolution: 1920x1200

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=aarch64-linux-gnu --disable-libquadmath --disable-libquadmath-support --disable-werror --enable-bootstrap --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-fix-cortex-a53-843419 --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-nls --enable-objc-gc=auto --enable-plugin --enable-shared --enable-threads=posix --host=aarch64-linux-gnu --program-prefix=aarch64-linux-gnu- --target=aarch64-linux-gnu --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-target-system-zlib=auto -v
Processor Notes: Scaling Governor: cppc_cpufreq performance (Boost: Disabled)
Graphics Notes: BAR1 / Visible vRAM Size: N/A - vBIOS Version: 96.00.7e.00.02
Python Notes: Python 3.10.12
Security Notes: gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of __user pointer sanitization + spectre_v2: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 1 March 2024 11:07 by user x.