vulkan-benchmarks

AMD Ryzen 9 7950X 16-Core testing with a ASUS ROG STRIX X670E-E GAMING WIFI (1416 BIOS) and NVIDIA GeForce RTX 4090 24GB on Ubuntu 23.04 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 2308069-PTS-VULKANBE16
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

Limit displaying results to tests within:

NVIDIA GPU Compute 4 Tests
Vulkan Compute 4 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

Multi-Way Comparison

Condense Multi-Option Tests 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
a
August 01 2023
  3 Hours, 11 Minutes
b
August 01 2023
  1 Hour, 30 Minutes
c
August 01 2023
  1 Hour, 32 Minutes
d
August 01 2023
  3 Hours, 45 Minutes
e
August 01 2023
  3 Hours, 16 Minutes
f
August 02 2023
  1 Hour, 53 Minutes
g
August 02 2023
  2 Hours, 9 Minutes
h
August 02 2023
  47 Minutes
i
August 02 2023
  1 Hour, 50 Minutes
4080
August 02 2023
  2 Hours, 4 Minutes
4080 rep
August 02 2023
  2 Hours, 7 Minutes
4080 xxx
August 02 2023
  2 Hours, 8 Minutes
4080 zzz
August 02 2023
  2 Hours, 9 Minutes
3090
August 03 2023
  2 Hours, 44 Minutes
3090 rep
August 03 2023
  2 Hours, 54 Minutes
3070
August 03 2023
  4 Hours, 56 Minutes
RTX 3070 Ti
August 04 2023
  1 Day, 7 Hours, 27 Minutes
4090
August 06 2023
  2 Hours, 52 Minutes
4090 rep
August 06 2023
  2 Hours, 54 Minutes
nv 4090
August 06 2023
  2 Hours, 52 Minutes
Invert Hiding All Results Option
  3 Hours, 57 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):


vulkan-benchmarks ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerOpenGLCompilerFile-SystemScreen ResolutionDisplay Driverabcdefghi40804080 rep4080 xxx4080 zzz30903090 rep3070RTX 3070 Ti40904090 repnv 4090AMD Ryzen 9 7950X 16-Core @ 4.50GHz (16 Cores / 32 Threads)ASUS ROG STRIX X670E-E GAMING WIFI (1416 BIOS)AMD Device 14d832GBWestern Digital WD_BLACK SN850X 1000GB + 4001GBAMD Radeon RX 6700 XT (2855/1000MHz)AMD Navi 21/23ASUS MG28UIntel I225-V + Intel Wi-Fi 6 AX210/AX211/AX411Ubuntu 23.046.4.6-060406-generic (x86_64)GNOME Shell 44.2X Server 1.21.1.7 + Wayland4.6 Mesa 23.3~git2307260600.87109c~oibaf~l (git-87109c3 2023-07-26 lunar-oibaf-ppa) (LLVM 15.0.7 DRM 3.52)GCC 12.2.0ext43840x2160MSI NVIDIA GeForce RTX 4060 8GBNVIDIA Device 22beX Server 1.21.1.7NVIDIA 535.86.054.6.0eVGA NVIDIA GeForce RTX 3060 12GBNVIDIA GA106 HD AudioNVIDIA GeForce RTX 3060 Ti 8GBNVIDIA GA104 HD Audio2560x1440NVIDIA GeForce RTX 4080 16GBNVIDIA Device 22bb3840x2160NVIDIA GeForce RTX 3090 24GBNVIDIA GA102 HD AudioNVIDIA GeForce RTX 3070 8GBNVIDIA GA104 HD Audio2560x1440NVIDIA GeForce RTX 3070 Ti 8GBNVIDIA GeForce RTX 4090 24GBNVIDIA AD102 HD Audio3840x2160OpenBenchmarking.orgKernel Details- Transparent Huge Pages: madviseCompiler Details- --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --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-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-12-Pa930Z/gcc-12-12.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-12-Pa930Z/gcc-12-12.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v Processor Details- a: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa601203- b: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa601203- c: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa601203- d: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- e: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- f: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- g: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- h: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- i: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 4080: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 4080 rep: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 4080 xxx: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 4080 zzz: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 3090: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 3090 rep: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 3070: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- RTX 3070 Ti: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 4090: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- 4090 rep: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203- nv 4090: Scaling Governor: acpi-cpufreq performance (Boost: Enabled) - CPU Microcode: 0xa601203Graphics Details- a: BAR1 / Visible vRAM Size: 12272 MB - vBIOS Version: 113-D5121100-101- b: BAR1 / Visible vRAM Size: 12272 MB - vBIOS Version: 113-D5121100-101- c: BAR1 / Visible vRAM Size: 12272 MB - vBIOS Version: 113-D5121100-101- d: BAR1 / Visible vRAM Size: 8192 MiB - vBIOS Version: 95.07.31.00.e3- e: BAR1 / Visible vRAM Size: 8192 MiB - vBIOS Version: 95.07.31.00.e3- f: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 94.06.14.40.46- g: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 94.06.14.40.46- h: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 94.06.14.40.46- i: BAR1 / Visible vRAM Size: 8192 MiB - vBIOS Version: 94.04.25.00.2c- 4080: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 95.03.0e.00.04- 4080 rep: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 95.03.0e.00.04- 4080 xxx: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 95.03.0e.00.04- 4080 zzz: BAR1 / Visible vRAM Size: 16384 MiB - vBIOS Version: 95.03.0e.00.04- 3090: BAR1 / Visible vRAM Size: 32768 MiB - vBIOS Version: 94.02.27.00.02- 3090 rep: BAR1 / Visible vRAM Size: 32768 MiB - vBIOS Version: 94.02.27.00.02- 3070: BAR1 / Visible vRAM Size: 8192 MiB - vBIOS Version: 94.04.25.00.2b- RTX 3070 Ti: BAR1 / Visible vRAM Size: 8192 MiB - vBIOS Version: 94.04.5b.00.02- 4090: BAR1 / Visible vRAM Size: 32768 MiB - vBIOS Version: 95.02.20.00.01- 4090 rep: BAR1 / Visible vRAM Size: 32768 MiB - vBIOS Version: 95.02.20.00.01- nv 4090: BAR1 / Visible vRAM Size: 32768 MiB - vBIOS Version: 95.02.20.00.01Security Details- 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: Not affected + srbds: Not affected + tsx_async_abort: Not affected

vulkan-benchmarks vkpeak: fp16-vec4vkpeak: int32-scalarvkpeak: int16-vec4vkpeak: int32-vec4vkpeak: int16-scalarvkpeak: fp16-scalarvkpeak: fp32-vec4vkpeak: fp32-scalarvkpeak: fp64-scalarvkpeak: fp64-vec4ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - FastestDetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vision_transformerncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - regnety_400mncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - squeezenet_ssdncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - yolov4-tinyncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet50ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - alexnetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet18ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vgg16ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - googlenetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - blazefacencnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - efficientnet-b0ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mnasnetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - shufflenet-v2ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenetvkfft: FFT + iFFT C2C Bluestein benchmark in double precisionvkresample: 2x - Doublencnn: Vulkan GPU-v3-v3 - mobilenet-v3ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - FastestDetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vision_transformerncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - regnety_400mncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - squeezenet_ssdncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - yolov4-tinyncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet50ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - alexnetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet18ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vgg16ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - googlenetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - blazefacencnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - efficientnet-b0ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mnasnetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - shufflenet-v2ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenetvkfft: FFT + iFFT C2C 1D batched in double precisionncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3vkfft: FFT + iFFT C2C Bluestein in single precisionncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - yolov4-tinyncnn: CPU-v3-v3 - mobilenet-v3ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - FastestDetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - vision_transformerncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - regnety_400mncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - squeezenet_ssdncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - resnet50ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - alexnetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - resnet18ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - vgg16ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - googlenetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - blazefacencnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - efficientnet-b0ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - mnasnetncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - shufflenet-v2ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2ncnn: Vulkan GPU-v3-v3-v3-v3-v3-v3 - mobilenetncnn: Vulkan GPU - FastestDetncnn: Vulkan GPU - vision_transformerncnn: Vulkan GPU - regnety_400mncnn: Vulkan GPU - squeezenet_ssdncnn: Vulkan GPU - yolov4-tinyncnn: Vulkan GPU - resnet50ncnn: Vulkan GPU - alexnetncnn: Vulkan GPU - resnet18ncnn: Vulkan GPU - vgg16ncnn: Vulkan GPU - googlenetncnn: Vulkan GPU - blazefacencnn: Vulkan GPU - efficientnet-b0ncnn: Vulkan GPU - mnasnetncnn: Vulkan GPU - shufflenet-v2ncnn: Vulkan GPU-v2-v2 - mobilenet-v2ncnn: Vulkan GPU - mobilenetncnn: CPU - FastestDetncnn: CPU - blazefacencnn: Vulkan GPU-v3-v3-v3-v3-v3 - mobilenet-v3ncnn: CPU-v3-v3-v3-v3-v3 - mobilenet-v3ncnn: CPU - vision_transformerncnn: CPU - regnety_400mncnn: CPU - squeezenet_ssdncnn: CPU - yolov4-tinyncnn: CPU - resnet50ncnn: CPU - alexnetncnn: CPU - resnet18ncnn: CPU - vgg16ncnn: CPU - googlenetncnn: CPU - efficientnet-b0ncnn: CPU - mnasnetncnn: CPU - shufflenet-v2ncnn: CPU-v2-v2 - mobilenet-v2ncnn: CPU - mobilenetncnn: CPU-v3-v3-v3 - FastestDetncnn: CPU-v3-v3-v3 - vision_transformerncnn: CPU-v3-v3-v3 - regnety_400mncnn: CPU-v3-v3-v3 - squeezenet_ssdncnn: CPU-v3-v3-v3 - yolov4-tinyncnn: CPU-v3-v3-v3 - resnet50ncnn: CPU-v3-v3-v3 - alexnetncnn: CPU-v3-v3-v3 - resnet18ncnn: CPU-v3-v3-v3 - vgg16ncnn: CPU-v3-v3-v3 - googlenetncnn: CPU-v3-v3-v3 - blazefacencnn: CPU-v3-v3-v3 - efficientnet-b0ncnn: CPU-v3-v3-v3 - mnasnetncnn: CPU-v3-v3-v3 - shufflenet-v2ncnn: CPU-v3-v3-v3-v2-v2 - mobilenet-v2ncnn: CPU-v3-v3-v3 - mobilenetncnn: Vulkan GPU-v3-v3-v3 - regnety_400mncnn: Vulkan GPU-v3-v3-v3 - FastestDetncnn: Vulkan GPU-v3-v3-v3 - vision_transformerncnn: Vulkan GPU-v3-v3-v3 - squeezenet_ssdncnn: Vulkan GPU-v3-v3-v3 - yolov4-tinyncnn: Vulkan GPU-v3-v3-v3 - resnet50ncnn: Vulkan GPU-v3-v3-v3 - alexnetncnn: Vulkan GPU-v3-v3-v3 - resnet18ncnn: Vulkan GPU-v3-v3-v3 - vgg16ncnn: Vulkan GPU-v3-v3-v3 - googlenetncnn: Vulkan GPU-v3-v3-v3 - blazefacencnn: Vulkan GPU-v3-v3-v3 - efficientnet-b0ncnn: Vulkan GPU-v3-v3-v3 - mnasnetncnn: Vulkan GPU-v3-v3-v3 - shufflenet-v2ncnn: Vulkan GPU-v3-v3-v3-v2-v2 - mobilenet-v2ncnn: Vulkan GPU-v3-v3-v3 - mobilenetvkfft: FFT + iFFT C2C 1D batched in single precisionncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - FastestDetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vision_transformerncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - regnety_400mncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - squeezenet_ssdncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - yolov4-tinyncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet50ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - alexnetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet18ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vgg16ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - googlenetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - blazefacencnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - efficientnet-b0ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mnasnetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - shufflenet-v2ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2ncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenetncnn: CPU-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3vkfft: FFT + iFFT C2C 1D batched in single precision, no reshufflingncnn: CPU-v3-v3-v3-v3-v3-v3 - FastestDetncnn: CPU-v3-v3-v3-v3-v3-v3 - vision_transformerncnn: CPU-v3-v3-v3-v3-v3-v3 - regnety_400mncnn: CPU-v3-v3-v3-v3-v3-v3 - squeezenet_ssdncnn: CPU-v3-v3-v3-v3-v3-v3 - yolov4-tinyncnn: CPU-v3-v3-v3-v3-v3-v3 - resnet50ncnn: CPU-v3-v3-v3-v3-v3-v3 - alexnetncnn: CPU-v3-v3-v3-v3-v3-v3 - resnet18ncnn: CPU-v3-v3-v3-v3-v3-v3 - vgg16ncnn: CPU-v3-v3-v3-v3-v3-v3 - googlenetncnn: CPU-v3-v3-v3-v3-v3-v3 - blazefacencnn: CPU-v3-v3-v3-v3-v3-v3 - efficientnet-b0ncnn: CPU-v3-v3-v3-v3-v3-v3 - mnasnetncnn: CPU-v3-v3-v3-v3-v3-v3 - shufflenet-v2ncnn: CPU-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2ncnn: CPU-v3-v3-v3-v3-v3-v3 - mobilenetvkfft: FFT + iFFT C2C 1D batched in half precisionvkfft: FFT + iFFT C2C multidimensional in single precisionvkfft: FFT + iFFT R2C / C2Rvkresample: 2x - Singleabcdefghi40804080 rep4080 xxx4080 zzz30903090 rep3070RTX 3070 Ti40904090 repnv 409023232.422272.6223123.772658.7313102.7513154.1512730.0813190.09841.40841.8047173.1720816113403.184.131.888.167.0912.8410.014.315.2823.517.901.383.862.983.353.178.053.621.3832.498.187.0712.9010.204.415.2923.757.943.902.973.343.168.05478875050491597330014210511.68623390.442269.2523396.592640.0813070.8113145.1912808.5912807.06839.2836.55469520822112734.0731.858.217.0712.87104.335.2323.567.851.373.822.953.333.148.044.051.383.163.1631.958.187.0612.7410.014.325.223.497.823.852.973.343.167.974.0731.658.057.0312.9810.014.295.2123.57.841.373.822.973.333.158.018.274.0631.717.1412.779.874.425.4223.427.971.373.852.963.333.158479485064391812327514216311.6923387.262269.0623385.442638.6913063.8613136.7912822.0112860.56839.01836.1646703.220847113113.174.0931.7987.0612.81104.335.2123.547.81.373.832.963.323.138.034.111.393.173.1631.778.277.112.8110.114.315.2423.457.933.882.993.353.188.024.0831.788.147.0412.8910.334.285.2623.997.881.383.892.973.343.1487.983.6931.667.0712.8610.034.35.2323.547.831.363.822.963.333.157.95479715059691744328124302111.68816864.478520.027352.858465.825676.028412.3311251.178531.96267.43267.742346500.0143.1712143107193.174.1132.128.177.0812.8510.104.315.2323.567.851.383.872.973.353.168.024.081.3832.438.237.0912.9510.004.305.2323.517.853.852.983.353.178.10426454336585181363283539932.85516865.298505.207336.258465.715675.998397.8011231.728515.58267.41267.252343500.0163.1812168105604.0831.938.107.0512.8710.104.315.2223.607.851.383.842.963.333.148.04426514336585191370903530432.85013440.976827.925959.756800.174480.596812.529006.576837.94214.17214.231814500.013.141056175714.2432.928.347.0813.1710.264.645.4824.558.151.383.862.973.43.138.274.221.373.153.1533.568.087.2313.3211.054.365.6924.197.923.872.973.553.158.453.8533.478.346.9713.0711.054.836.1324.458.071.434.043.123.43.168.568.54.233.367.0914.3410.254.355.324.127.941.373.852.963.333.168.655647657110104146262382659326.73813438.476824.295956.386794.924479.226811.359002.596812.99213.96213.951818500.011105483.16757413.143.143.9233.328.387.1410.334.355.2824.047.961.383.842.973.343.178.54.0732.428.367.113.6410.344.876.2224.28.961.413.862.983.353.188.172.571.383.153.1632.738.37.1317.2310.724.325.5523.787.983.9133.593.1622.743.9733.398.077.2613.0811.254.715.4824.929.151.374.143.053.383.158.987.993.9732.687.3113.3510.434.865.524.718.351.384.632.983.353.178.25645557094104171265412663826.76913490.246800.65978.386772.984495.986838.329036.176810.73213.37210.9610572762256431104298265242417500.0064.87147803.261006113.773.263.8338.018.467.2113.15.15.8827.4310.471.414.193.073.433.2910.025.1436.429.887.4615.1612.966.535.8227.838.751.45.883.23.493.2910.42.661.43.263.2937.89.948.9614.6512.095.35.630.9610.34.052.745.033.528.375.6936.558.218.1615.1114.055.015.8629.1210.171.284.683.393.523.310.087.994.4338.338.3315.4311.154.995.8529.0710.191.254.212.993.363.289.056973871163132270346863372720.935579288.2014.1935.68.677.7113.9311.484.985.9225.678.791.444.043.13.483.263.298.43349741712113.863.244.2834.918.617.6611.44.615.6125.488.41.434.063.063.463.288.844.235.568.397.5813.7910.814.625.67258.421.414.013.073.433.268.444.421.43.273.2835.078.247.7313.8511.164.755.6925.378.423.993.053.413.288.734.234.28.337.6413.8111.114.665.725.18.451.444.023.083.463.318.438.454.234.137.6613.7910.954.695.6525.048.491.424.053.093.443.298.43104556106210211076658696647313.1365583288.1664.1733.938.357.6213.7311.074.645.6725.568.421.424.013.063.443.313.278.38350383.241728713.553.264.1434.18.247.5510.84.725.6125.058.491.413.983.033.393.278.44.2135.078.567.6413.6710.844.655.6125.048.41.414.053.083.443.298.574.341.423.2735.288.677.8614.0311.764.675.6826.118.524.093.063.433.288.414.1834.278.577.5913.5510.794.655.6324.918.521.424.023.093.433.278.488.444.0934.297.6313.6810.844.695.6925.048.581.454.043.073.443.38.461044913.281062054.234.228.727.6713.7110.864.685.6425.018.521.434.073.093.473.38.45211058700686827913.1365587288.0393.7533.98.257.2713.5211.224.715.6525.338.261.313.972.983.343.053.148.31350713.261734313.623.274.1734.238.527.6710.914.675.6725.48.431.424.023.053.433.278.464.234.198.567.6213.6510.944.685.6225.018.421.424.043.073.53.268.374.171.423.333.3134.278.457.6213.610.824.685.5625.038.384.043.063.453.288.374.1934.378.757.6413.6910.914.655.66258.51.424.063.073.473.38.448.584.3135.47.713.9511.55.215.8926.088.991.424.223.133.513.48.881045283.081060993.834.148.387.2713.6311.264.695.7825.448.321.324.0133.43.28.34210713678876906813.1375584288.0283.284.6135.368.378.0615.2612.54.75.7426.098.551.414.053.083.443.263.289.19350583.241718513.613.244.234.18.497.6210.914.685.625.168.421.424.013.083.433.288.464.7934.328.587.6313.811.074.685.5925.828.411.424.043.063.463.298.474.161.43.23.2734.18.377.5513.6311.14.695.6325.48.43.993.043.423.258.384.0434.478.477.3513.8311.214.675.7125.458.551.414.033.063.433.288.48.14.1234.057.5113.6211.094.655.5925.268.371.393.953.013.373.238.381045433.061059263.8234.478.347.2513.4211.14.665.7725.268.291.313.952.963.363.168.25210991700406768913.12641149.120909.0216886.6620820.0913710.8820845.0927797.821269.72653.13653.153.164.0832.168.27.0512.8610.034.35.223.57.831.393.862.973.363.158.014282371.6993.193.8331.948.337.0512.9710.074.325.2123.517.841.383.882.943.323.133.128.03309453.161440613.13.184.0333.227.997.0410.384.315.1923.557.861.363.832.963.333.158.114.1131.948.387.1612.8810.14.355.2723.557.871.393.882.993.393.188.074.211.383.1533.018.257.5214.2610.34.35.2123.437.863.872.993.343.178.64.0431.897.957.0412.8710.054.315.1923.57.831.363.832.953.323.148.068.014.0431.867.0412.889.974.35.2323.57.821.363.852.973.343.1681413571439694.132.18.227.1212.8210.034.335.223.587.91.393.882.983.363.168.07255207510055534710.39941188.0220767.6416881.4720517.6813608.5720953.327807.5820925.3653.634.132.138.197.0712.8910.044.35.223.477.821.383.852.983.363.178.054289371.4223.164.0731.978.077.0812.99.984.35.223.387.861.373.852.973.343.183.178.05311221444912.863.194.0832.098.347.0910.044.315.2223.527.891.393.872.993.373.198.044.0831.918.027.0612.8310.064.315.223.487.821.373.862.973.363.178.034.081.383.153.1531.88.247.1212.779.954.315.2923.437.93.862.973.353.178.014.1131.938.097.0912.8210.074.35.223.547.851.373.852.963.333.178.038.254.132.117.0812.8410.014.35.2423.437.851.383.852.973.363.178.011414374.0731.948.067.0712.9210.274.315.2723.727.861.383.842.963.323.193.158.033.171439564.0731.858.037.0912.8110.064.315.323.47.911.383.852.973.333.168.06265171548145443210.4285.974.4869.4818.2415.4626.3323.5410.6913.3451.2818.663.579.538.156.35.4918.5424.7456.66.7170.2917.6118.8328.7323.4411.8911.349.716.972.999.816.065.595.995.4617.096.5629.87.526.9371.0817.8815.423.5910.0812.1455.4818.61.778.414.5988.3518.398.4170.7616.2215.8228.5923.4810.8812.6848.2919.493.988.995.097.077.8121.118.652.988.065.3875.341813.227.6621.59.6214.0356.6418.259.236.886.829.6717.819.1881.7719.6617.7529.3424.071111.1449.75173.039.016.878.139.1917.8217.238.6373.5116.1529.4923.119.8613.3855.4220.723.187.816.024.897.2416.347.2370.5317.0215.3229.3822.1511.4312.1353.4818.82.696.638.555.897.345.9217.066.437.1265.4118.2514.2728.4122.1910.5912.6450.3219.22.539.196.077.817.2216.5222.0643.524.2538.329.108.3115.5612.525.416.6928.539.581.514.553.253.923.699.9824.8053.644.3238.279.058.6515.0012.425.346.4028.539.901.344.373.103.893.443.669.523.6215.213.614.4137.889.198.2812.735.536.5728.639.841.494.603.343.753.669.624.2638.039.078.4715.5412.605.256.2829.069.871.604.533.403.983.569.353.941.603.653.7637.918.838.1315.2012.735.496.0828.369.694.723.263.773.769.434.3337.869.028.3915.4212.115.556.1828.409.651.794.783.114.093.419.628.894.2638.298.2915.4412.355.676.2328.409.861.714.733.373.953.669.624.1438.509.147.4514.6413.156.255.9427.869.971.404.173.123.483.243.8310.023.704.1838.048.427.5714.5712.816.176.2227.989.682.484.743.244.023.9110.0327.1833.362.8238.8210.057.415.6912.984.947.7827.318.911.454.143.193.475.258.468039172.8833.284.4538.629.879.8115.9512.44.677.5227.449.051.424.154.933.523.623.489.16552143.32037315.553.534.0338.388.17.5714.085.146.5831.5710.871.164.363.175.093.3210.184.3938.768.647.8313.9714.134.645.6928.8210.621.394.343.183.453.310.555.481.273.123.2538.258.137.8613.6814.15.14627.7510.274.233.123.553.310.082.9338.798.139.3215.311.394.945.8128.559.971.174.093.195.183.468.969.63.9439.017.9315.4414.586.115.9728.218.871.334.1833.344.998.811538964.6239.3510.117.4316.05135.146.9627.328.551.354.633.233.563.364.7510.563.331526562.8538.7610.099.5115.8511.724.997.7430.168.381.34.475.193.483.369.0429034281406843519.2843.443.9138.698.459.3416.611.245.165.8429.3510.181.384.443.285.183.68.838119173.0433.34.1139.0310.699.4615.411.515.255.929.1710.391.414.414.993.483.343.369.02553833.332040415.453.413.9638.178.649.1612.175.336.0529.1210.381.464.043.135.273.348.374.5937.598.488.2215.7213.086.796.0127.0410.651.44.093.153.493.3110.235.271.454.93.5337.819.667.3115.3812.735.148.0728.199.534.033.223.514.748.434.1639.1217.159.315.3413.825.277.7527.259.291.344.343.233.593.458.7410.344.1638.737.8116.3913.576.585.8127.598.91.416.283.13.43.319.541539394.5938.658.79.4415.4110.965.345.8729.8510.471.424.13.125.233.353.388.223.31559363.1238.7910.239.3813.8812.475.458.1430.748.971.424.355.113.423.4410.6128765180999813298.9624.814.0138.588.377.7215.6713.466.545.9729.48.852.915.883.13.373.2710.158132172.8873.362.6438.4610.037.0215.5513.134.697.3828.148.351.264.13.073.53.173.428.93549503.262060117.33.472.8139.1810.099.2113.635.148.1627.898.611.184.123.163.515.19.413.9339.049.819.1115.2612.455.27.4429.298.71.44.14.73.513.438.454.511.332.613.3538.910.179.1115.411.415.187.8229.548.934.374.773.453.68.154.0638.999.559.3715.6213.684.677.6127.049.021.164.044.613.463.398.917.735.9238.587.7216.6113.136.115.8427.7710.011.075.262.543.174.4510.541521705.8637.138.348.2616.313.256.325.5827.2510.141.45.823.13.434.963.2910.644.971551483.9338.958.257.4817.6713.296.626.0727.6110.751.425.943.123.323.2912.1229276882875848878.967OpenBenchmarking.org

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 20230730fp16-vec43090 rep3090hgfedcba9K18K27K36K45KSE +/- 0.36, N = 3SE +/- 0.37, N = 3SE +/- 5.96, N = 341188.0241149.1013490.2413438.4713440.9716865.2916864.4723387.2623390.4423232.42

fp16-vec4

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20230730int32-scalar3090 rep3090hgfedcba4K8K12K16K20KSE +/- 0.03, N = 3SE +/- 15.02, N = 3SE +/- 0.34, N = 320767.6420909.026800.606824.296827.928505.208520.022269.062269.252272.62

int32-scalar

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20230730int16-vec43090 rep3090hgfedcba5K10K15K20K25KSE +/- 0.31, N = 3SE +/- 17.33, N = 3SE +/- 21.55, N = 316881.4716886.665978.385956.385959.757336.257352.8523385.4423396.5923123.77

int16-vec4

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20230730int32-vec43090 rep3090hgfedcba4K8K12K16K20KSE +/- 0.05, N = 3SE +/- 0.19, N = 3SE +/- 0.26, N = 320517.6820820.096772.986794.926800.178465.718465.822638.692640.082658.73

int32-vec4

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGIOPS, More Is Bettervkpeak 20230730int16-scalar3090 rep3090hgfedcba3K6K9K12K15KSE +/- 0.02, N = 3SE +/- 0.09, N = 3SE +/- 1.30, N = 313608.5713710.884495.984479.224480.595675.995676.0213063.8613070.8113102.75

int16-scalar

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20230730fp16-scalar3090 rep3090hgfedcba4K8K12K16K20KSE +/- 5.09, N = 3SE +/- 13.46, N = 3SE +/- 4.01, N = 320953.3020845.096838.326811.356812.528397.808412.3313136.7913145.1913154.15

fp16-scalar

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20230730fp32-vec43090 rep3090hgfedcba6K12K18K24K30KSE +/- 2.57, N = 3SE +/- 19.37, N = 3SE +/- 1.81, N = 327807.5827797.809036.179002.599006.5711231.7211251.1712822.0112808.5912730.08

fp32-vec4

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20230730fp32-scalar3090 rep3090hgfedcba5K10K15K20K25KSE +/- 0.30, N = 3SE +/- 16.18, N = 3SE +/- 4.18, N = 320925.3021269.726810.736812.996837.948515.588531.9612860.5612807.0613190.09

fp32-scalar

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20230730fp64-scalar3090 rep3090hgfedcba2004006008001000SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.22, N = 3653.63653.13213.37213.96214.17267.41267.43839.01839.20841.40

fp64-scalar

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

OpenBenchmarking.orgGFLOPS, More Is Bettervkpeak 20230730fp64-vec43090hgfedcba2004006008001000SE +/- 0.00, N = 3SE +/- 0.48, N = 3SE +/- 0.32, N = 3653.15210.96213.95214.23267.25267.74836.16836.55841.80

fp64-vec4

i: The test quit with a non-zero exit status.

4080: The test quit with a non-zero exit status.

4080 rep: The test quit with a non-zero exit status.

4080 xxx: The test quit with a non-zero exit status.

4080 zzz: The test quit with a non-zero exit status.

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

4090: The test quit with a non-zero exit status.

4090 rep: The test quit with a non-zero exit status.

nv 4090: The test quit with a non-zero exit status.

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: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti307030901.34332.68664.02995.37326.7165SE +/- 0.17, N = 154.813.443.363.525.973.16MIN: 3.13 / MAX: 149.75MIN: 3.3 / MAX: 4.34MIN: 3.21 / MAX: 4.83MIN: 2.95 / MAX: 536.1MIN: 2.84 / MAX: 111.8MIN: 3.12 / MAX: 3.671. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30901.0082.0163.0244.0325.04SE +/- 0.29, N = 154.013.912.824.254.484.104.08MIN: 3.87 / MAX: 5.47MIN: 3.77 / MAX: 5.87MIN: 2.69 / MAX: 3.5MIN: 2.46 / MAX: 526.3MIN: 2.2 / MAX: 27.6MIN: 4.06 / MAX: 4.2MIN: 4.04 / MAX: 4.21. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30901530456075SE +/- 0.12, N = 1538.5838.6938.8238.3269.4832.1332.16MIN: 33.06 / MAX: 464.16MIN: 33.32 / MAX: 390.07MIN: 33.83 / MAX: 435.6MIN: 32.26 / MAX: 477.15MIN: 39.08 / MAX: 374.31MIN: 31.95 / MAX: 32.87MIN: 31.94 / MAX: 33.71. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep309048121620SE +/- 0.20, N = 158.378.4510.059.1018.248.198.20MIN: 8.08 / MAX: 10.1MIN: 8.05 / MAX: 12.64MIN: 8.13 / MAX: 173.18MIN: 7.61 / MAX: 454.62MIN: 7.5 / MAX: 201.09MIN: 8.12 / MAX: 8.98MIN: 8.14 / MAX: 8.741. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep309048121620SE +/- 0.22, N = 157.729.347.408.3115.467.077.05MIN: 7.13 / MAX: 8.97MIN: 6.88 / MAX: 268.7MIN: 6.81 / MAX: 8.46MIN: 6.35 / MAX: 364.95MIN: 7.08 / MAX: 147.31MIN: 6.99 / MAX: 7.81MIN: 6.98 / MAX: 7.811. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep3090612182430SE +/- 0.25, N = 1515.6716.6015.6915.5626.3312.8912.86MIN: 12.91 / MAX: 334.44MIN: 12.98 / MAX: 103.04MIN: 13.13 / MAX: 187.93MIN: 12.24 / MAX: 459.8MIN: 12.62 / MAX: 127.32MIN: 12.79 / MAX: 13.77MIN: 12.74 / MAX: 13.681. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep3090612182430SE +/- 0.24, N = 1513.4611.2412.9812.5223.5410.0410.03MIN: 10.6 / MAX: 340.67MIN: 10.22 / MAX: 29.96MIN: 10.26 / MAX: 145.62MIN: 9.95 / MAX: 459.05MIN: 10.3 / MAX: 149.49MIN: 9.94 / MAX: 10.91MIN: 9.88 / MAX: 10.861. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30903691215SE +/- 0.21, N = 156.545.164.945.4110.694.304.30MIN: 4.56 / MAX: 110.58MIN: 4.73 / MAX: 6.38MIN: 4.52 / MAX: 6.23MIN: 4.23 / MAX: 364.66MIN: 4.32 / MAX: 148.92MIN: 4.24 / MAX: 4.85MIN: 4.25 / MAX: 4.631. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30903691215SE +/- 0.24, N = 155.975.847.786.6913.345.205.20MIN: 5.4 / MAX: 8.25MIN: 5.35 / MAX: 8.28MIN: 5.4 / MAX: 168.29MIN: 5.06 / MAX: 462.37MIN: 5.43 / MAX: 279.86MIN: 5.08 / MAX: 6.05MIN: 5.1 / MAX: 6.051. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30901224364860SE +/- 0.26, N = 1529.4029.3527.3128.5351.2823.4723.50MIN: 26.17 / MAX: 411.51MIN: 24.55 / MAX: 485.35MIN: 24.27 / MAX: 230.86MIN: 24.21 / MAX: 515.3MIN: 24.83 / MAX: 242.12MIN: 23.25 / MAX: 24.24MIN: 23.26 / MAX: 24.341. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep3090510152025SE +/- 0.22, N = 158.8510.188.919.5818.667.827.83MIN: 8.16 / MAX: 10.25MIN: 7.81 / MAX: 204.67MIN: 8.3 / MAX: 10.96MIN: 7.62 / MAX: 396.9MIN: 7.42 / MAX: 326.73MIN: 7.72 / MAX: 8.6MIN: 7.73 / MAX: 8.61. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30900.80331.60662.40993.21324.0165SE +/- 0.14, N = 152.911.381.451.513.571.381.39MIN: 1.29 / MAX: 113.97MIN: 1.33 / MAX: 1.98MIN: 1.38 / MAX: 2.98MIN: 1.11 / MAX: 380.46MIN: 1.08 / MAX: 141.04MIN: 1.35 / MAX: 1.88MIN: 1.36 / MAX: 3.121. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30903691215SE +/- 0.16, N = 155.884.444.144.559.533.853.86MIN: 3.96 / MAX: 194.08MIN: 4.24 / MAX: 5.18MIN: 3.93 / MAX: 5.94MIN: 3.84 / MAX: 379.07MIN: 3.77 / MAX: 182.53MIN: 3.81 / MAX: 4.6MIN: 3.82 / MAX: 4.821. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep3090246810SE +/- 0.11, N = 153.103.283.193.258.152.982.97MIN: 2.97 / MAX: 3.92MIN: 3.15 / MAX: 4.32MIN: 3.04 / MAX: 3.98MIN: 2.68 / MAX: 277.21MIN: 2.67 / MAX: 317.68MIN: 2.94 / MAX: 3.36MIN: 2.93 / MAX: 3.451. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep3090246810SE +/- 0.20, N = 153.375.183.473.926.303.363.36MIN: 3.25 / MAX: 5.26MIN: 3.45 / MAX: 200.36MIN: 3.33 / MAX: 5.01MIN: 3.12 / MAX: 496.78MIN: 3.28 / MAX: 147.57MIN: 3.33 / MAX: 3.83MIN: 3.32 / MAX: 3.821. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30901.23532.47063.70594.94126.1765SE +/- 0.20, N = 153.273.605.253.695.493.173.15MIN: 3.11 / MAX: 4.1MIN: 3.44 / MAX: 4.27MIN: 3.11 / MAX: 367.53MIN: 3.07 / MAX: 544.13MIN: 2.97 / MAX: 152.08MIN: 3.12 / MAX: 3.78MIN: 3.11 / MAX: 3.781. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep3090510152025SE +/- 0.24, N = 1510.158.838.469.9818.548.058.01MIN: 8.08 / MAX: 193.04MIN: 8.29 / MAX: 10.15MIN: 8.12 / MAX: 10.14MIN: 7.79 / MAX: 434.9MIN: 8.01 / MAX: 164.45MIN: 7.98 / MAX: 8.94MIN: 7.96 / MAX: 8.471. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

VkFFT

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT C2C Bluestein benchmark in double precisionnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080igfedcba2K4K6K8K10KSE +/- 11.20, N = 3SE +/- 4.37, N = 3SE +/- 0.33, N = 3813281198039428942825584558755835579241718181814234323464670469547171. (CXX) g++ options: -O3

Test: FFT + iFFT C2C Bluestein benchmark in double precision

3070: The test quit with a non-zero exit status. E: VkFFT System: 2909x2909x1 Buffer: 129 MB avg_time_per_step: 72.294 ms std_error: 0.046 num_iter: 31 benchmark: 1828 bandwidth: 27.9

RTX 3070 Ti: The test quit with a non-zero exit status. E: VkFFT System: 2909x2909x1 Buffer: 129 MB avg_time_per_step: 66.798 ms std_error: 0.291 num_iter: 31 benchmark: 1979 bandwidth: 30.2

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: Doublenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfed110220330440550SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3172.89173.04172.8824.8124.75371.42371.70288.03288.04288.17288.20500.01500.01500.01500.02500.011. (CXX) g++ options: -O3

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-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzzifedca246810SE +/- 0.20, N = 14SE +/- 0.02, N = 3SE +/- 0.00, N = 2SE +/- 0.02, N = 33.363.303.283.646.603.163.193.284.873.143.183.173.203.17MIN: 3.21 / MAX: 4.3MIN: 3.15 / MAX: 3.92MIN: 3.15 / MAX: 3.9MIN: 2.87 / MAX: 429.02MIN: 2.98 / MAX: 166.19MIN: 3.11 / MAX: 3.62MIN: 3.14 / MAX: 3.48MIN: 3.13 / MAX: 4.65MIN: 3.14 / MAX: 278.98MIN: 3.09 / MAX: 3.54MIN: 3.11 / MAX: 3.78MIN: 3.1 / MAX: 3.83MIN: 3.16 / MAX: 3.68MIN: 3.11 / MAX: 3.731. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080246810SE +/- 0.27, N = 152.644.114.454.326.714.073.834.613.754.174.19MIN: 2.52 / MAX: 4.14MIN: 3.98 / MAX: 4.73MIN: 4.29 / MAX: 5.05MIN: 2.51 / MAX: 398.91MIN: 2.73 / MAX: 109.52MIN: 4.03 / MAX: 4.18MIN: 3.79 / MAX: 4.09MIN: 4.45 / MAX: 5.92MIN: 3.63 / MAX: 5.24MIN: 4.02 / MAX: 4.75MIN: 4.06 / MAX: 7.411. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40801632486480SE +/- 0.11, N = 1538.4639.0338.6238.2770.2931.9731.9435.3633.9033.9335.60MIN: 32.39 / MAX: 435.46MIN: 33.61 / MAX: 343.67MIN: 33.33 / MAX: 465MIN: 32.29 / MAX: 507.7MIN: 39.39 / MAX: 250.19MIN: 31.71 / MAX: 33.78MIN: 31.72 / MAX: 34.34MIN: 33.87 / MAX: 42.41MIN: 32.72 / MAX: 37.77MIN: 32.77 / MAX: 36.2MIN: 34.13 / MAX: 38.491. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep408048121620SE +/- 0.24, N = 1510.0310.699.879.0517.618.078.338.378.258.358.67MIN: 7.81 / MAX: 171.2MIN: 8.17 / MAX: 339.6MIN: 7.81 / MAX: 243.06MIN: 7.52 / MAX: 417.33MIN: 7.85 / MAX: 165.34MIN: 7.99 / MAX: 8.88MIN: 8.25 / MAX: 9.32MIN: 8.04 / MAX: 10.13MIN: 7.93 / MAX: 9.88MIN: 8.05 / MAX: 9.76MIN: 8.3 / MAX: 14.661. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080510152025SE +/- 0.24, N = 157.029.469.818.6518.837.087.058.067.277.627.71MIN: 6.38 / MAX: 9.36MIN: 7.03 / MAX: 160.39MIN: 7.16 / MAX: 389.1MIN: 6.64 / MAX: 544.17MIN: 6.71 / MAX: 206.11MIN: 7 / MAX: 7.94MIN: 6.97 / MAX: 7.95MIN: 7.42 / MAX: 9.25MIN: 6.74 / MAX: 8.84MIN: 7.01 / MAX: 14.37MIN: 7.15 / MAX: 9.11. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080714212835SE +/- 0.19, N = 1515.5515.4015.9515.0028.7312.9012.9715.2613.5213.7313.93MIN: 12.87 / MAX: 342.3MIN: 13 / MAX: 245.79MIN: 13.38 / MAX: 245.18MIN: 12.75 / MAX: 401.37MIN: 12.83 / MAX: 264.49MIN: 12.77 / MAX: 13.92MIN: 12.83 / MAX: 13.8MIN: 14.19 / MAX: 17.06MIN: 12.72 / MAX: 21.19MIN: 12.78 / MAX: 20.99MIN: 13.08 / MAX: 15.681. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080612182430SE +/- 0.25, N = 1513.1311.5112.4012.4223.449.9810.0712.5011.2211.0711.48MIN: 10.18 / MAX: 247.5MIN: 10.56 / MAX: 13.22MIN: 11.44 / MAX: 14.43MIN: 10.23 / MAX: 444.76MIN: 10.17 / MAX: 219.36MIN: 9.85 / MAX: 11.35MIN: 9.95 / MAX: 10.88MIN: 11.47 / MAX: 14.56MIN: 10.33 / MAX: 12.81MIN: 10.16 / MAX: 13.16MIN: 10.56 / MAX: 12.931. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40803691215SE +/- 0.16, N = 154.695.254.675.3411.894.304.324.704.714.644.98MIN: 4.28 / MAX: 6.33MIN: 4.86 / MAX: 6.33MIN: 4.28 / MAX: 6MIN: 4.25 / MAX: 221.78MIN: 4.34 / MAX: 229.18MIN: 4.24 / MAX: 5.11MIN: 4.25 / MAX: 5.33MIN: 4.28 / MAX: 5.92MIN: 4.26 / MAX: 7.21MIN: 4.24 / MAX: 6MIN: 4.59 / MAX: 7.151. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40803691215SE +/- 0.20, N = 157.385.907.526.4011.305.205.215.745.655.675.92MIN: 5.15 / MAX: 138.85MIN: 5.43 / MAX: 7.49MIN: 5.45 / MAX: 290.49MIN: 5.1 / MAX: 457.07MIN: 5.3 / MAX: 181.7MIN: 5.1 / MAX: 6.09MIN: 5.09 / MAX: 6.13MIN: 5.18 / MAX: 8.08MIN: 5.18 / MAX: 6.76MIN: 5.19 / MAX: 7.38MIN: 5.37 / MAX: 8.241. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40801122334455SE +/- 0.28, N = 1528.1429.1727.4428.5349.7023.3823.5126.0925.3325.5625.67MIN: 24.24 / MAX: 221.5MIN: 24.61 / MAX: 264.85MIN: 24.06 / MAX: 264.59MIN: 23.95 / MAX: 473.83MIN: 25.55 / MAX: 421.44MIN: 23.19 / MAX: 24.27MIN: 23.27 / MAX: 24.38MIN: 24.58 / MAX: 30.18MIN: 24.26 / MAX: 34.98MIN: 24.24 / MAX: 27.92MIN: 24.46 / MAX: 27.341. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep408048121620SE +/- 0.19, N = 158.3510.399.059.9016.977.867.848.558.268.428.79MIN: 7.7 / MAX: 10.46MIN: 7.87 / MAX: 391.66MIN: 8.26 / MAX: 13.34MIN: 7.76 / MAX: 396.66MIN: 7.44 / MAX: 229.93MIN: 7.75 / MAX: 8.71MIN: 7.74 / MAX: 8.72MIN: 7.86 / MAX: 10.08MIN: 7.62 / MAX: 10.47MIN: 7.77 / MAX: 10.52MIN: 8.08 / MAX: 10.271. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40800.67281.34562.01842.69123.364SE +/- 0.03, N = 151.261.411.421.342.991.371.381.411.311.421.44MIN: 1.2 / MAX: 1.76MIN: 1.35 / MAX: 1.91MIN: 1.36 / MAX: 1.92MIN: 1.06 / MAX: 2.66MIN: 1.22 / MAX: 149.55MIN: 1.35 / MAX: 1.48MIN: 1.36 / MAX: 1.53MIN: 1.34 / MAX: 1.88MIN: 1.25 / MAX: 3.14MIN: 1.35 / MAX: 2.89MIN: 1.37 / MAX: 2.071. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40803691215SE +/- 0.13, N = 154.104.414.154.379.813.853.884.053.974.014.04MIN: 3.87 / MAX: 6.14MIN: 4.21 / MAX: 5.82MIN: 3.93 / MAX: 5.94MIN: 3.85 / MAX: 366.28MIN: 3.87 / MAX: 165.38MIN: 3.78 / MAX: 4.83MIN: 3.83 / MAX: 4.72MIN: 3.83 / MAX: 5.42MIN: 3.79 / MAX: 5.93MIN: 3.81 / MAX: 6.04MIN: 3.84 / MAX: 4.831. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080246810SE +/- 0.04, N = 153.074.994.933.106.062.972.943.082.983.063.10MIN: 2.93 / MAX: 4.52MIN: 3.02 / MAX: 235.56MIN: 2.97 / MAX: 124.96MIN: 2.61 / MAX: 4.75MIN: 2.96 / MAX: 42.7MIN: 2.94 / MAX: 3.45MIN: 2.9 / MAX: 3.34MIN: 2.95 / MAX: 3.88MIN: 2.86 / MAX: 4.47MIN: 2.93 / MAX: 5.02MIN: 2.95 / MAX: 4.051. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40801.25782.51563.77345.03126.289SE +/- 0.20, N = 153.503.483.523.895.593.343.323.443.343.443.48MIN: 3.37 / MAX: 4.2MIN: 3.34 / MAX: 4.1MIN: 3.38 / MAX: 4.23MIN: 3.08 / MAX: 345.39MIN: 3.32 / MAX: 42.33MIN: 3.3 / MAX: 3.79MIN: 3.29 / MAX: 3.79MIN: 3.31 / MAX: 4.85MIN: 3.22 / MAX: 3.97MIN: 3.31 / MAX: 4.32MIN: 3.34 / MAX: 4.881. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40801.34782.69564.04345.39126.739SE +/- 0.13, N = 133.173.343.623.445.993.183.133.263.053.313.26MIN: 3.04 / MAX: 4.3MIN: 3.19 / MAX: 3.99MIN: 3.47 / MAX: 4.24MIN: 2.65 / MAX: 361.91MIN: 3.05 / MAX: 26.81MIN: 3.13 / MAX: 3.61MIN: 3.09 / MAX: 3.68MIN: 3.12 / MAX: 4.74MIN: 2.94 / MAX: 3.56MIN: 3.16 / MAX: 3.93MIN: 3.13 / MAX: 4.71. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep40801.22852.4573.68554.9146.1425SE +/- 0.18, N = 153.423.363.483.665.463.173.123.283.143.273.29MIN: 3.15 / MAX: 25.1MIN: 3.17 / MAX: 4.8MIN: 3.32 / MAX: 4.99MIN: 2.73 / MAX: 398.42MIN: 3.27 / MAX: 38.65MIN: 3.12 / MAX: 3.89MIN: 3.07 / MAX: 3.62MIN: 3.11 / MAX: 4.26MIN: 3 / MAX: 3.85MIN: 3.08 / MAX: 4.68MIN: 3.12 / MAX: 3.991. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep408048121620SE +/- 0.25, N = 158.939.029.169.5217.098.058.039.198.318.388.43MIN: 8.33 / MAX: 11.07MIN: 8.42 / MAX: 11.17MIN: 8.5 / MAX: 10.51MIN: 7.97 / MAX: 420.29MIN: 7.89 / MAX: 121.53MIN: 7.96 / MAX: 9.04MIN: 7.96 / MAX: 8.83MIN: 8.51 / MAX: 11.04MIN: 7.85 / MAX: 10.21MIN: 7.94 / MAX: 10.07MIN: 8.03 / MAX: 9.641. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

VkFFT

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT C2C 1D batched in double precisionnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080ihgfedcba12K24K36K48K60KSE +/- 14.62, N = 3SE +/- 12.42, N = 3SE +/- 10.58, N = 3SE +/- 11.67, N = 35495055383552143112230945350583507135038349741478010572105481056112168121432084720822208161. (CXX) g++ options: -O3

Test: FFT + iFFT C2C 1D batched in double precision

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

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-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti307030904080 zzz4080 xxx4080 repig246810SE +/- 0.18, N = 153.263.333.303.626.563.163.243.263.243.263.16MIN: 3.13 / MAX: 3.96MIN: 3.19 / MAX: 4.79MIN: 3.14 / MAX: 4.82MIN: 3 / MAX: 469.9MIN: 3.07 / MAX: 110.87MIN: 3.11 / MAX: 3.77MIN: 3.1 / MAX: 3.88MIN: 3.13 / MAX: 4.08MIN: 3.11 / MAX: 4.37MIN: 3.11 / MAX: 4.7MIN: 3.12 / MAX: 3.581. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

VkFFT

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT C2C Bluestein in single precisionnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080ihgfedcba4K8K12K16K20KSE +/- 83.38, N = 3SE +/- 75.16, N = 15SE +/- 72.34, N = 3SE +/- 62.67, N = 32060120404203731444914406171851734317287171211006176227574757110560107191131111273113401. (CXX) g++ options: -O3

Test: FFT + iFFT C2C Bluestein in single precision

3070: The test quit with a non-zero exit status. E: VkFFT System: 4241x4241x1 Buffer: 137 MB avg_time_per_step: 16.022 ms std_error: 0.041 num_iter: 29 benchmark: 8770 bandwidth: 133.8

RTX 3070 Ti: The test quit with a non-zero exit status. E: VkFFT System: 4241x4241x1 Buffer: 137 MB avg_time_per_step: 10.602 ms std_error: 0.177 num_iter: 29 benchmark: 13253 bandwidth: 202.2

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-v3-v3-v3-v3-v3-v3 - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig714212835SE +/- 0.28, N = 1517.3015.4515.5515.2129.8012.8613.1013.6113.6213.5513.8613.7713.14MIN: 14.66 / MAX: 441.3MIN: 12.65 / MAX: 445.76MIN: 13.11 / MAX: 307.2MIN: 12.34 / MAX: 380.51MIN: 12.85 / MAX: 216.34MIN: 12.76 / MAX: 13.73MIN: 13.01 / MAX: 14.17MIN: 12.67 / MAX: 19.72MIN: 12.71 / MAX: 15.65MIN: 12.72 / MAX: 15.51MIN: 13.04 / MAX: 15.04MIN: 12.96 / MAX: 14.66MIN: 13 / MAX: 14.021. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igdca246810SE +/- 0.21, N = 14SE +/- 0.00, N = 3SE +/- 0.00, N = 23.473.413.533.617.523.193.183.243.273.263.243.263.143.173.173.18MIN: 3.32 / MAX: 4.91MIN: 3.27 / MAX: 5.24MIN: 3.39 / MAX: 4.31MIN: 2.51 / MAX: 502.85MIN: 2.94 / MAX: 215MIN: 3.15 / MAX: 3.72MIN: 3.14 / MAX: 4.14MIN: 3.11 / MAX: 4.47MIN: 3.13 / MAX: 3.85MIN: 3.09 / MAX: 3.96MIN: 3.09 / MAX: 4.73MIN: 3.14 / MAX: 3.9MIN: 3.1 / MAX: 3.81MIN: 3.12 / MAX: 3.96MIN: 3.15 / MAX: 3.74MIN: 3.14 / MAX: 3.821. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig246810SE +/- 0.20, N = 152.813.964.034.416.934.084.034.204.174.144.283.833.92MIN: 2.68 / MAX: 4.38MIN: 3.79 / MAX: 11.36MIN: 3.89 / MAX: 4.63MIN: 2.06 / MAX: 295.24MIN: 2.57 / MAX: 163.84MIN: 4.04 / MAX: 4.29MIN: 3.99 / MAX: 4.22MIN: 4.01 / MAX: 11.47MIN: 4.03 / MAX: 5.63MIN: 4 / MAX: 5.6MIN: 4.13 / MAX: 4.85MIN: 3.7 / MAX: 4.57MIN: 3.88 / MAX: 4.721. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig1632486480SE +/- 0.20, N = 1539.1838.1738.3837.8871.0832.0933.2234.1034.2334.1034.9138.0133.32MIN: 33.74 / MAX: 520.24MIN: 32.97 / MAX: 462.63MIN: 33.53 / MAX: 477.38MIN: 32.46 / MAX: 518.57MIN: 38.84 / MAX: 374.68MIN: 31.84 / MAX: 32.77MIN: 33.04 / MAX: 36.99MIN: 32.32 / MAX: 38.54MIN: 33.08 / MAX: 37.43MIN: 32.43 / MAX: 38.75MIN: 33.72 / MAX: 36.82MIN: 32.96 / MAX: 388.09MIN: 31.83 / MAX: 104.121. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig48121620SE +/- 0.21, N = 1510.098.648.109.1917.888.347.998.498.528.248.618.468.38MIN: 7.84 / MAX: 366.66MIN: 8.3 / MAX: 10.51MIN: 7.65 / MAX: 10.05MIN: 7.44 / MAX: 524.66MIN: 7.38 / MAX: 190.77MIN: 8.26 / MAX: 9.09MIN: 7.92 / MAX: 8.78MIN: 8.08 / MAX: 9.72MIN: 8.13 / MAX: 9.73MIN: 7.91 / MAX: 9.53MIN: 8.21 / MAX: 10.07MIN: 8.08 / MAX: 10.33MIN: 8.05 / MAX: 27.341. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig48121620SE +/- 0.25, N = 149.219.167.578.2815.407.097.047.627.677.557.667.217.14MIN: 6.83 / MAX: 203.62MIN: 6.73 / MAX: 423.75MIN: 7.02 / MAX: 9MIN: 6.38 / MAX: 381.81MIN: 6.64 / MAX: 132.68MIN: 7.01 / MAX: 7.97MIN: 6.96 / MAX: 7.7MIN: 7 / MAX: 9.93MIN: 7.04 / MAX: 9.1MIN: 6.99 / MAX: 9.08MIN: 7.09 / MAX: 8.97MIN: 6.73 / MAX: 8.82MIN: 7.03 / MAX: 7.991. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig612182430SE +/- 0.23, N = 1513.6312.1714.0812.7323.5910.0410.3810.9110.9110.8011.4013.1010.33MIN: 10.52 / MAX: 488.94MIN: 11.25 / MAX: 13.79MIN: 10.29 / MAX: 247.29MIN: 9.84 / MAX: 518.97MIN: 9.96 / MAX: 177.63MIN: 9.94 / MAX: 10.89MIN: 9.88 / MAX: 18.75MIN: 9.94 / MAX: 14.83MIN: 9.91 / MAX: 13.07MIN: 9.89 / MAX: 12.54MIN: 10.5 / MAX: 13.51MIN: 10.59 / MAX: 267.95MIN: 10.2 / MAX: 11.181. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig3691215SE +/- 0.22, N = 155.145.335.145.5310.084.314.314.684.674.724.615.104.35MIN: 4.65 / MAX: 6.81MIN: 4.83 / MAX: 6.6MIN: 4.76 / MAX: 6.16MIN: 4.22 / MAX: 362.62MIN: 4.36 / MAX: 225.66MIN: 4.26 / MAX: 5.07MIN: 4.25 / MAX: 4.94MIN: 4.26 / MAX: 6.8MIN: 4.27 / MAX: 6.36MIN: 4.25 / MAX: 7.3MIN: 4.24 / MAX: 7.25MIN: 4.75 / MAX: 6.12MIN: 4.28 / MAX: 5.11. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig3691215SE +/- 0.23, N = 158.166.056.586.5712.145.225.195.605.675.615.615.885.28MIN: 5.39 / MAX: 397.44MIN: 5.53 / MAX: 7.66MIN: 6.04 / MAX: 7.81MIN: 4.91 / MAX: 391.33MIN: 5.28 / MAX: 151.53MIN: 5.13 / MAX: 6.1MIN: 5.09 / MAX: 6MIN: 5.09 / MAX: 7.51MIN: 5.1 / MAX: 8.06MIN: 5.07 / MAX: 7.08MIN: 5.09 / MAX: 7.91MIN: 5.36 / MAX: 8.2MIN: 5.16 / MAX: 6.091. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig1224364860SE +/- 0.30, N = 1527.8929.1231.5728.6355.4823.5223.5525.1625.4025.0525.4827.4324.04MIN: 24.5 / MAX: 463.23MIN: 24.62 / MAX: 266.39MIN: 26.09 / MAX: 318.58MIN: 24.13 / MAX: 500.18MIN: 25.94 / MAX: 298.67MIN: 23.33 / MAX: 25.08MIN: 23.31 / MAX: 24.48MIN: 23.97 / MAX: 27.81MIN: 24.05 / MAX: 27.09MIN: 23.78 / MAX: 26.95MIN: 23.88 / MAX: 51.68MIN: 24.65 / MAX: 251.37MIN: 23.48 / MAX: 73.31. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig510152025SE +/- 0.24, N = 158.6110.3810.879.8418.607.897.868.428.438.498.4010.477.96MIN: 7.95 / MAX: 10.07MIN: 7.96 / MAX: 255.68MIN: 8.37 / MAX: 194.11MIN: 7.3 / MAX: 438.04MIN: 8.02 / MAX: 292.16MIN: 7.79 / MAX: 8.84MIN: 7.74 / MAX: 8.62MIN: 7.78 / MAX: 10.7MIN: 7.77 / MAX: 10.4MIN: 7.74 / MAX: 10.76MIN: 7.71 / MAX: 10.64MIN: 8.21 / MAX: 350.07MIN: 7.81 / MAX: 9.051. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig0.39830.79661.19491.59321.9915SE +/- 0.12, N = 141.181.461.161.491.771.391.361.421.421.411.431.411.38MIN: 1.11 / MAX: 1.85MIN: 1.39 / MAX: 2.91MIN: 1.1 / MAX: 2MIN: 1.05 / MAX: 379.08MIN: 1.08 / MAX: 12.53MIN: 1.37 / MAX: 1.52MIN: 1.34 / MAX: 1.46MIN: 1.34 / MAX: 2.84MIN: 1.35 / MAX: 2MIN: 1.34 / MAX: 2.1MIN: 1.36 / MAX: 2.06MIN: 1.35 / MAX: 2.02MIN: 1.35 / MAX: 2.091. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig246810SE +/- 0.19, N = 154.124.044.364.608.413.873.834.014.023.984.064.193.84MIN: 3.86 / MAX: 5.39MIN: 3.85 / MAX: 4.9MIN: 4.14 / MAX: 5.24MIN: 3.79 / MAX: 336.2MIN: 3.76 / MAX: 67.73MIN: 3.81 / MAX: 4.62MIN: 3.78 / MAX: 4.4MIN: 3.79 / MAX: 5.39MIN: 3.8 / MAX: 5.14MIN: 3.77 / MAX: 5.44MIN: 3.85 / MAX: 4.97MIN: 4.01 / MAX: 5.09MIN: 3.78 / MAX: 4.571. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig1.03282.06563.09844.13125.164SE +/- 0.14, N = 153.163.133.173.344.592.992.963.083.053.033.063.072.97MIN: 3.02 / MAX: 4.6MIN: 3.01 / MAX: 3.62MIN: 3.03 / MAX: 3.66MIN: 2.68 / MAX: 393.6MIN: 2.88 / MAX: 20.12MIN: 2.96 / MAX: 3.32MIN: 2.93 / MAX: 3.31MIN: 2.93 / MAX: 4.42MIN: 2.91 / MAX: 3.67MIN: 2.91 / MAX: 4.45MIN: 2.94 / MAX: 3.67MIN: 2.93 / MAX: 3.84MIN: 2.93 / MAX: 3.881. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig246810SE +/- 0.16, N = 153.515.275.093.758.003.373.333.433.433.393.463.433.34MIN: 3.38 / MAX: 4.05MIN: 3.27 / MAX: 191.55MIN: 3.33 / MAX: 161.5MIN: 3.2 / MAX: 361.52MIN: 3.16 / MAX: 190.15MIN: 3.33 / MAX: 3.8MIN: 3.29 / MAX: 3.67MIN: 3.29 / MAX: 3.87MIN: 3.31 / MAX: 3.95MIN: 3.26 / MAX: 3.91MIN: 3.3 / MAX: 5.74MIN: 3.3 / MAX: 4.89MIN: 3.31 / MAX: 4.051. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig246810SE +/- 0.15, N = 155.103.343.323.668.353.193.153.283.273.273.283.293.17MIN: 3.14 / MAX: 138.88MIN: 3.14 / MAX: 4.45MIN: 3.12 / MAX: 4.24MIN: 3.01 / MAX: 311.25MIN: 3.08 / MAX: 103.38MIN: 3.13 / MAX: 4MIN: 3.1 / MAX: 3.68MIN: 3.09 / MAX: 4.98MIN: 3.11 / MAX: 4.73MIN: 3.08 / MAX: 5.18MIN: 3.11 / MAX: 4.16MIN: 3.1 / MAX: 3.96MIN: 3.1 / MAX: 5.031. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3-v3 - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080ig510152025SE +/- 0.27, N = 159.418.3710.189.6218.398.048.118.468.468.408.8410.028.50MIN: 8.98 / MAX: 11.38MIN: 7.98 / MAX: 10.71MIN: 8.18 / MAX: 235.56MIN: 7.71 / MAX: 449.11MIN: 7.92 / MAX: 173.39MIN: 7.96 / MAX: 9.01MIN: 8.02 / MAX: 14.2MIN: 7.97 / MAX: 10.56MIN: 7.95 / MAX: 10.34MIN: 7.93 / MAX: 15.25MIN: 8.31 / MAX: 10.98MIN: 8.07 / MAX: 266.25MIN: 8.42 / MAX: 9.291. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba246810SE +/- 0.29, N = 15SE +/- 0.00, N = 3SE +/- 0.02, N = 3SE +/- 0.00, N = 33.934.594.394.268.414.084.114.794.204.214.205.144.074.244.084.114.094.074.10MIN: 3.8 / MAX: 5.4MIN: 2.62 / MAX: 232.18MIN: 4.25 / MAX: 5.86MIN: 2.5 / MAX: 396.93MIN: 2.89 / MAX: 487.78MIN: 4.04 / MAX: 4.35MIN: 4.07 / MAX: 4.29MIN: 4.64 / MAX: 6.21MIN: 4.03 / MAX: 6.49MIN: 4.04 / MAX: 4.97MIN: 4.02 / MAX: 4.97MIN: 3.7 / MAX: 81.79MIN: 4.02 / MAX: 4.82MIN: 3.88 / MAX: 24.21MIN: 4.03 / MAX: 5.29MIN: 4.01 / MAX: 9.72MIN: 4.05 / MAX: 5.5MIN: 4.04 / MAX: 4.53MIN: 4.06 / MAX: 4.811. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba1632486480SE +/- 0.16, N = 15SE +/- 0.07, N = 3SE +/- 0.21, N = 3SE +/- 0.09, N = 339.0437.5938.7638.0370.7631.9131.9434.3234.1935.0735.5636.4232.4232.9231.9332.1231.7931.8531.88MIN: 33.83 / MAX: 463.88MIN: 34.45 / MAX: 457.98MIN: 33.12 / MAX: 539.58MIN: 32.66 / MAX: 467.28MIN: 38.81 / MAX: 250.01MIN: 31.74 / MAX: 34.28MIN: 31.73 / MAX: 34.21MIN: 32.58 / MAX: 41.88MIN: 32.72 / MAX: 36.79MIN: 33.66 / MAX: 39.36MIN: 33.19 / MAX: 40.43MIN: 33.49 / MAX: 224.86MIN: 31.89 / MAX: 65.47MIN: 32.67 / MAX: 36.93MIN: 31.62 / MAX: 35.85MIN: 31.66 / MAX: 46.9MIN: 31.63 / MAX: 35.57MIN: 31.69 / MAX: 33.06MIN: 31.55 / MAX: 37.471. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba48121620SE +/- 0.21, N = 15SE +/- 0.02, N = 3SE +/- 0.06, N = 3SE +/- 0.11, N = 39.818.488.649.0716.228.028.388.588.568.568.399.888.368.348.108.178.008.218.16MIN: 7.82 / MAX: 241.19MIN: 8.09 / MAX: 9.64MIN: 8.28 / MAX: 10.42MIN: 7.61 / MAX: 402.49MIN: 7.74 / MAX: 314.84MIN: 7.95 / MAX: 8.63MIN: 8.31 / MAX: 8.86MIN: 8.13 / MAX: 9.78MIN: 8.15 / MAX: 9.8MIN: 8.17 / MAX: 10.28MIN: 8 / MAX: 10.29MIN: 8.14 / MAX: 251.77MIN: 8.27 / MAX: 9.08MIN: 7.99 / MAX: 26.72MIN: 7.98 / MAX: 8.84MIN: 7.99 / MAX: 8.97MIN: 7.94 / MAX: 8.88MIN: 8.14 / MAX: 8.84MIN: 7.9 / MAX: 8.991. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba48121620SE +/- 0.26, N = 15SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.02, N = 39.118.227.838.4715.827.067.167.637.627.647.587.467.107.087.057.087.067.077.09MIN: 6.35 / MAX: 130.38MIN: 7.56 / MAX: 9.8MIN: 7.21 / MAX: 9.32MIN: 6.29 / MAX: 533.92MIN: 6.99 / MAX: 82.57MIN: 7 / MAX: 7.82MIN: 7.05 / MAX: 13.55MIN: 7 / MAX: 9.17MIN: 7.01 / MAX: 9.28MIN: 7.05 / MAX: 9.12MIN: 6.98 / MAX: 9.05MIN: 6.9 / MAX: 8.9MIN: 6.99 / MAX: 8.59MIN: 6.98 / MAX: 8.07MIN: 6.95 / MAX: 8MIN: 6.97 / MAX: 7.99MIN: 7 / MAX: 8.03MIN: 7 / MAX: 8.07MIN: 6.98 / MAX: 7.951. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba714212835SE +/- 0.18, N = 15SE +/- 0.04, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 315.2615.7213.9715.5428.5912.8312.8813.8013.6513.6713.7915.1613.6413.1712.8712.8512.8112.8712.84MIN: 12.87 / MAX: 132.82MIN: 13.2 / MAX: 301.81MIN: 13.11 / MAX: 16.15MIN: 12.15 / MAX: 492.01MIN: 12.87 / MAX: 325.37MIN: 12.74 / MAX: 13.59MIN: 12.76 / MAX: 13.67MIN: 12.76 / MAX: 15.76MIN: 12.71 / MAX: 14.99MIN: 12.71 / MAX: 14.88MIN: 12.75 / MAX: 19.63MIN: 12.86 / MAX: 248.64MIN: 13.04 / MAX: 76.32MIN: 13.03 / MAX: 14.1MIN: 12.68 / MAX: 13.84MIN: 12.72 / MAX: 13.93MIN: 12.73 / MAX: 13.08MIN: 12.76 / MAX: 13.73MIN: 12.69 / MAX: 15.331. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba612182430SE +/- 0.26, N = 15SE +/- 0.12, N = 3SE +/- 0.09, N = 3SE +/- 0.02, N = 312.4513.0814.1312.6023.4810.0610.1011.0710.9410.8410.8112.9610.3410.2610.1010.1010.0010.0010.01MIN: 11.55 / MAX: 14.48MIN: 10.11 / MAX: 444.45MIN: 10.63 / MAX: 167.28MIN: 9.82 / MAX: 418.4MIN: 10.06 / MAX: 112.91MIN: 9.95 / MAX: 11.04MIN: 9.97 / MAX: 11.42MIN: 10.1 / MAX: 13.23MIN: 9.95 / MAX: 12.7MIN: 9.93 / MAX: 12.81MIN: 9.95 / MAX: 12.78MIN: 10.23 / MAX: 424.46MIN: 10.14 / MAX: 11.37MIN: 10.09 / MAX: 11.22MIN: 9.84 / MAX: 11.72MIN: 9.86 / MAX: 11.08MIN: 9.91 / MAX: 11.15MIN: 9.92 / MAX: 12.35MIN: 9.88 / MAX: 11.41. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba3691215SE +/- 0.18, N = 15SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 35.206.794.645.2510.884.314.354.684.684.654.626.534.874.644.314.314.334.334.31MIN: 4.82 / MAX: 7.07MIN: 4.23 / MAX: 262.43MIN: 4.26 / MAX: 5.98MIN: 4.23 / MAX: 375.94MIN: 4.38 / MAX: 52.99MIN: 4.26 / MAX: 5.26MIN: 4.28 / MAX: 7.49MIN: 4.26 / MAX: 6.23MIN: 4.26 / MAX: 6.61MIN: 4.26 / MAX: 6.53MIN: 4.26 / MAX: 6.15MIN: 4.57 / MAX: 242.16MIN: 4.8 / MAX: 5.62MIN: 4.57 / MAX: 5.49MIN: 4.23 / MAX: 11.03MIN: 4.25 / MAX: 5.28MIN: 4.26 / MAX: 10.59MIN: 4.28 / MAX: 5.16MIN: 4.24 / MAX: 5.21. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba3691215SE +/- 0.20, N = 15SE +/- 0.01, N = 3SE +/- 0.03, N = 3SE +/- 0.01, N = 37.446.015.696.2812.685.205.275.595.625.615.675.826.225.485.225.235.215.235.28MIN: 5.29 / MAX: 320.54MIN: 5.44 / MAX: 8.18MIN: 5.16 / MAX: 8.22MIN: 4.94 / MAX: 298.06MIN: 5.39 / MAX: 262.62MIN: 5.09 / MAX: 5.98MIN: 5.15 / MAX: 6.19MIN: 5.06 / MAX: 6.95MIN: 5.1 / MAX: 7.65MIN: 5.11 / MAX: 7.44MIN: 5.18 / MAX: 7.22MIN: 5.28 / MAX: 7.02MIN: 6.11 / MAX: 7MIN: 5.33 / MAX: 6.16MIN: 5.09 / MAX: 11.15MIN: 5.08 / MAX: 6.28MIN: 5.11 / MAX: 6.04MIN: 5.13 / MAX: 6.18MIN: 5.17 / MAX: 6.161. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba1122334455SE +/- 0.24, N = 15SE +/- 0.14, N = 3SE +/- 0.10, N = 3SE +/- 0.04, N = 329.2927.0428.8229.0648.2923.4823.5525.8225.0125.0425.0027.8324.2024.5523.6023.5623.5423.5623.51MIN: 24.63 / MAX: 296.95MIN: 24.22 / MAX: 296.13MIN: 24.35 / MAX: 214.1MIN: 24.11 / MAX: 541.55MIN: 24.97 / MAX: 183.12MIN: 23.24 / MAX: 29.21MIN: 23.3 / MAX: 24.45MIN: 24.35 / MAX: 62.94MIN: 23.8 / MAX: 26.41MIN: 24.06 / MAX: 27.35MIN: 23.93 / MAX: 26.69MIN: 24.98 / MAX: 262.23MIN: 23.56 / MAX: 58.31MIN: 23.62 / MAX: 97.69MIN: 23.17 / MAX: 24.71MIN: 23.24 / MAX: 24.78MIN: 23.33 / MAX: 24.61MIN: 23.34 / MAX: 24.72MIN: 23.29 / MAX: 24.681. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba510152025SE +/- 0.22, N = 15SE +/- 0.01, N = 3SE +/- 0.04, N = 3SE +/- 0.02, N = 38.7010.6510.629.8719.497.827.878.418.428.408.428.758.968.157.857.857.807.857.90MIN: 7.96 / MAX: 10.01MIN: 8.29 / MAX: 236.11MIN: 7.83 / MAX: 323.31MIN: 7.33 / MAX: 399.24MIN: 7.4 / MAX: 200.01MIN: 7.69 / MAX: 8.61MIN: 7.76 / MAX: 10.36MIN: 7.72 / MAX: 9.9MIN: 7.73 / MAX: 10.06MIN: 7.77 / MAX: 9.78MIN: 7.79 / MAX: 10.01MIN: 8.08 / MAX: 16.01MIN: 8.82 / MAX: 9.87MIN: 8.02 / MAX: 9.02MIN: 7.71 / MAX: 8.76MIN: 7.71 / MAX: 8.85MIN: 7.72 / MAX: 8.74MIN: 7.76 / MAX: 8.76MIN: 7.74 / MAX: 9.541. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba0.89551.7912.68653.5824.4775SE +/- 0.16, N = 15SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 31.401.401.391.603.981.371.391.421.421.411.411.401.411.381.381.381.371.371.38MIN: 1.34 / MAX: 1.87MIN: 1.33 / MAX: 1.93MIN: 1.33 / MAX: 1.94MIN: 1.11 / MAX: 436.01MIN: 1.31 / MAX: 228.4MIN: 1.36 / MAX: 1.46MIN: 1.37 / MAX: 1.82MIN: 1.36 / MAX: 2.01MIN: 1.36 / MAX: 1.93MIN: 1.35 / MAX: 1.9MIN: 1.35 / MAX: 2.01MIN: 1.33 / MAX: 2MIN: 1.38 / MAX: 2.09MIN: 1.35 / MAX: 2.08MIN: 1.34 / MAX: 1.88MIN: 1.34 / MAX: 2.25MIN: 1.35 / MAX: 1.82MIN: 1.35 / MAX: 1.75MIN: 1.34 / MAX: 1.851. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba3691215SE +/- 0.18, N = 15SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.01, N = 34.104.094.344.538.993.863.884.044.044.054.015.883.863.863.843.873.833.823.86MIN: 3.86 / MAX: 5.46MIN: 3.87 / MAX: 5.46MIN: 4.14 / MAX: 5.84MIN: 3.75 / MAX: 396.62MIN: 3.71 / MAX: 129.99MIN: 3.82 / MAX: 4.34MIN: 3.84 / MAX: 4.39MIN: 3.8 / MAX: 5.31MIN: 3.82 / MAX: 5.33MIN: 3.83 / MAX: 6.11MIN: 3.78 / MAX: 5.34MIN: 4.04 / MAX: 364.21MIN: 3.82 / MAX: 4.22MIN: 3.78 / MAX: 10.45MIN: 3.79 / MAX: 4.76MIN: 3.77 / MAX: 9.91MIN: 3.79 / MAX: 4.61MIN: 3.78 / MAX: 4.39MIN: 3.8 / MAX: 4.61. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba1.14532.29063.43594.58125.7265SE +/- 0.16, N = 15SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 34.703.153.183.405.092.972.993.063.073.083.073.202.982.972.962.972.962.952.98MIN: 3 / MAX: 188.08MIN: 3 / MAX: 4.54MIN: 3.05 / MAX: 4.64MIN: 2.72 / MAX: 432.18MIN: 2.86 / MAX: 53.75MIN: 2.94 / MAX: 3.28MIN: 2.96 / MAX: 3.14MIN: 2.92 / MAX: 3.73MIN: 2.95 / MAX: 4.19MIN: 2.94 / MAX: 3.67MIN: 2.93 / MAX: 4.63MIN: 3.07 / MAX: 3.86MIN: 2.94 / MAX: 3.65MIN: 2.93 / MAX: 3.66MIN: 2.91 / MAX: 5.9MIN: 2.92 / MAX: 3.34MIN: 2.93 / MAX: 3.41MIN: 2.92 / MAX: 3.42MIN: 2.92 / MAX: 4.031. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba246810SE +/- 0.20, N = 15SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 33.513.493.453.987.073.363.393.463.503.443.433.493.353.403.333.353.323.333.35MIN: 3.37 / MAX: 4MIN: 3.36 / MAX: 4.33MIN: 3.32 / MAX: 3.99MIN: 3.14 / MAX: 529.82MIN: 3.25 / MAX: 243.32MIN: 3.32 / MAX: 4.06MIN: 3.35 / MAX: 3.69MIN: 3.32 / MAX: 5.24MIN: 3.37 / MAX: 4.85MIN: 3.3 / MAX: 5.36MIN: 3.3 / MAX: 4.22MIN: 3.35 / MAX: 4.24MIN: 3.3 / MAX: 4.02MIN: 3.35 / MAX: 5.89MIN: 3.28 / MAX: 4.14MIN: 3.3 / MAX: 3.82MIN: 3.29 / MAX: 4.19MIN: 3.3 / MAX: 3.59MIN: 3.29 / MAX: 3.851. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba246810SE +/- 0.14, N = 15SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 33.433.313.303.567.813.173.183.293.263.293.263.293.183.133.143.163.133.143.17MIN: 3.25 / MAX: 4.81MIN: 3.14 / MAX: 4.92MIN: 3.12 / MAX: 4.82MIN: 3.09 / MAX: 345.01MIN: 3.07 / MAX: 154.75MIN: 3.12 / MAX: 3.64MIN: 3.14 / MAX: 3.63MIN: 3.11 / MAX: 3.98MIN: 3.1 / MAX: 3.87MIN: 3.12 / MAX: 4.14MIN: 3.1 / MAX: 4.12MIN: 3.12 / MAX: 3.93MIN: 3.13 / MAX: 3.9MIN: 3.07 / MAX: 3.82MIN: 3.08 / MAX: 4.06MIN: 3.09 / MAX: 3.92MIN: 3.08 / MAX: 3.85MIN: 3.1 / MAX: 3.73MIN: 3.09 / MAX: 3.781. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba510152025SE +/- 0.25, N = 15SE +/- 0.02, N = 3SE +/- 0.00, N = 3SE +/- 0.02, N = 38.4510.2310.559.3521.118.038.078.478.378.578.4410.408.178.278.048.028.038.048.05MIN: 8.03 / MAX: 12.61MIN: 8.13 / MAX: 386.42MIN: 8.22 / MAX: 303.1MIN: 7.49 / MAX: 474.12MIN: 7.98 / MAX: 322.43MIN: 7.96 / MAX: 8.77MIN: 7.99 / MAX: 8.8MIN: 8.04 / MAX: 10.17MIN: 7.97 / MAX: 16.09MIN: 7.98 / MAX: 10MIN: 7.98 / MAX: 10.55MIN: 7.97 / MAX: 455.46MIN: 8.08 / MAX: 9.37MIN: 8.17 / MAX: 9.04MIN: 7.95 / MAX: 9.09MIN: 7.95 / MAX: 9.81MIN: 7.98 / MAX: 8.84MIN: 7.95 / MAX: 14.33MIN: 7.95 / MAX: 8.891. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba246810SE +/- 0.23, N = 15SE +/- 0.01, N = 3SE +/- 0.45, N = 34.515.275.483.948.654.084.214.164.174.344.422.662.574.224.084.114.053.62MIN: 4.34 / MAX: 5.96MIN: 4.05 / MAX: 247.02MIN: 2.67 / MAX: 259.34MIN: 2.43 / MAX: 267.02MIN: 3.94 / MAX: 185.21MIN: 4.05 / MAX: 4.84MIN: 4.19 / MAX: 4.41MIN: 4 / MAX: 4.69MIN: 4.05 / MAX: 4.74MIN: 4.19 / MAX: 5.77MIN: 4.25 / MAX: 6.71MIN: 2.54 / MAX: 3.41MIN: 2.53 / MAX: 3.21MIN: 4.18 / MAX: 4.97MIN: 4.02 / MAX: 4.28MIN: 4.08 / MAX: 4.4MIN: 4.02 / MAX: 4.35MIN: 2.7 / MAX: 4.541. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba0.67051.3412.01152.6823.3525SE +/- 0.14, N = 15SE +/- 0.01, N = 3SE +/- 0.00, N = 31.331.451.271.602.981.381.381.401.421.421.401.401.381.371.381.391.381.38MIN: 1.27 / MAX: 1.77MIN: 1.38 / MAX: 2.96MIN: 1.21 / MAX: 1.95MIN: 0.95 / MAX: 433.24MIN: 1.29 / MAX: 144.96MIN: 1.36 / MAX: 1.71MIN: 1.35 / MAX: 2.23MIN: 1.34 / MAX: 2.1MIN: 1.36 / MAX: 2.02MIN: 1.35 / MAX: 1.88MIN: 1.34 / MAX: 2.15MIN: 1.34 / MAX: 2MIN: 1.36 / MAX: 1.62MIN: 1.34 / MAX: 2.11MIN: 1.35 / MAX: 2.05MIN: 1.36 / MAX: 1.53MIN: 1.35 / MAX: 1.67MIN: 1.35 / MAX: 2.061. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti30703090 rep4080 zzz4080 xxx4080 rep4080igfcb246810SE +/- 0.18, N = 152.614.903.123.658.063.153.203.333.273.273.263.153.153.173.16MIN: 2.5 / MAX: 3.12MIN: 3.17 / MAX: 120.84MIN: 2.99 / MAX: 5.09MIN: 2.87 / MAX: 347.75MIN: 2.96 / MAX: 219.87MIN: 3.11 / MAX: 3.83MIN: 3.06 / MAX: 3.84MIN: 3.19 / MAX: 4.2MIN: 3.14 / MAX: 3.99MIN: 3.12 / MAX: 5.24MIN: 3.12 / MAX: 4.19MIN: 3.1 / MAX: 3.87MIN: 3.1 / MAX: 3.8MIN: 3.11 / MAX: 8.89MIN: 3.11 / MAX: 3.751. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080igfcb1.21052.4213.63154.8426.0525SE +/- 0.22, N = 143.353.533.253.765.383.153.153.273.313.283.293.163.153.163.16MIN: 3.21 / MAX: 5.23MIN: 3.2 / MAX: 40.81MIN: 3.11 / MAX: 4.74MIN: 2.89 / MAX: 366.04MIN: 2.74 / MAX: 121.29MIN: 3.11 / MAX: 3.6MIN: 3.11 / MAX: 3.71MIN: 3.14 / MAX: 4.63MIN: 3.16 / MAX: 5.3MIN: 3.14 / MAX: 3.89MIN: 3.15 / MAX: 4.32MIN: 3.11 / MAX: 3.93MIN: 3.11 / MAX: 3.48MIN: 3.12 / MAX: 3.7MIN: 3.12 / MAX: 3.691. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba20406080100SE +/- 0.12, N = 15SE +/- 0.39, N = 3SE +/- 0.29, N = 338.9037.8138.2537.9175.3431.8033.0134.1034.2735.2835.0737.8032.7333.5632.4331.7731.9532.49MIN: 34.2 / MAX: 300.84MIN: 32.66 / MAX: 453.44MIN: 33.04 / MAX: 447.7MIN: 32.08 / MAX: 541.11MIN: 38.72 / MAX: 418.01MIN: 31.66 / MAX: 32.23MIN: 32.88 / MAX: 33.42MIN: 32.65 / MAX: 37.64MIN: 32.82 / MAX: 39.79MIN: 33.9 / MAX: 38.67MIN: 33.14 / MAX: 43.26MIN: 33.74 / MAX: 321.51MIN: 31.44 / MAX: 81.32MIN: 32.98 / MAX: 51.93MIN: 31.56 / MAX: 37.69MIN: 31.61 / MAX: 35.68MIN: 31.79 / MAX: 32.33MIN: 31.67 / MAX: 40.111. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba48121620SE +/- 0.19, N = 15SE +/- 0.06, N = 3SE +/- 0.04, N = 310.179.668.138.8318.008.248.258.378.458.678.249.948.308.088.238.278.188.18MIN: 8.12 / MAX: 209.53MIN: 7.78 / MAX: 95.3MIN: 7.78 / MAX: 9.98MIN: 7.65 / MAX: 351.08MIN: 7.91 / MAX: 176.28MIN: 8.17 / MAX: 8.84MIN: 8.17 / MAX: 8.9MIN: 8.05 / MAX: 10.19MIN: 8.12 / MAX: 9.68MIN: 8.22 / MAX: 15.29MIN: 7.89 / MAX: 9.52MIN: 7.43 / MAX: 166.02MIN: 8.22 / MAX: 9.1MIN: 7.98 / MAX: 10.87MIN: 8.03 / MAX: 8.9MIN: 8.22 / MAX: 9.18MIN: 8.12 / MAX: 8.86MIN: 8.07 / MAX: 9.681. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba3691215SE +/- 0.24, N = 15SE +/- 0.01, N = 3SE +/- 0.01, N = 39.117.317.868.1313.207.127.527.557.627.867.738.967.137.237.097.107.067.07MIN: 6.77 / MAX: 101.58MIN: 6.71 / MAX: 9.3MIN: 7.25 / MAX: 8.98MIN: 6.37 / MAX: 399.11MIN: 6.9 / MAX: 68.61MIN: 7.05 / MAX: 7.63MIN: 7.45 / MAX: 7.74MIN: 7 / MAX: 8.72MIN: 7.01 / MAX: 8.84MIN: 7.22 / MAX: 10.84MIN: 7.13 / MAX: 9.7MIN: 6.92 / MAX: 244.02MIN: 7.04 / MAX: 8.43MIN: 7.15 / MAX: 8.02MIN: 6.99 / MAX: 9.39MIN: 7.05 / MAX: 7.65MIN: 7.01 / MAX: 7.55MIN: 7.01 / MAX: 8.071. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba714212835SE +/- 0.18, N = 15SE +/- 0.05, N = 3SE +/- 0.11, N = 315.4015.3813.6815.2027.6612.7714.2613.6313.6014.0313.8514.6517.2313.3212.9512.8112.7412.90MIN: 12.35 / MAX: 321.43MIN: 12.32 / MAX: 188.07MIN: 12.83 / MAX: 14.63MIN: 12.69 / MAX: 431.37MIN: 12.74 / MAX: 294.9MIN: 12.7 / MAX: 13.02MIN: 14.17 / MAX: 14.53MIN: 12.77 / MAX: 15.36MIN: 12.8 / MAX: 16.23MIN: 13.15 / MAX: 15.97MIN: 12.84 / MAX: 16.75MIN: 12.44 / MAX: 202.68MIN: 12.99 / MAX: 196.66MIN: 12.95 / MAX: 35.49MIN: 12.75 / MAX: 18.88MIN: 12.74 / MAX: 13.2MIN: 12.66 / MAX: 13.28MIN: 12.69 / MAX: 15.881. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba510152025SE +/- 0.26, N = 15SE +/- 0.01, N = 3SE +/- 0.23, N = 311.4112.7314.1012.7321.509.9510.3011.1010.8211.7611.1612.0910.7211.0510.0010.1110.0110.20MIN: 10.57 / MAX: 12.22MIN: 10.22 / MAX: 181.72MIN: 10.27 / MAX: 287MIN: 10.18 / MAX: 541.92MIN: 10.24 / MAX: 116.85MIN: 9.85 / MAX: 10.72MIN: 9.82 / MAX: 17.56MIN: 10.2 / MAX: 13.06MIN: 9.9 / MAX: 12.26MIN: 10.68 / MAX: 44.94MIN: 10.29 / MAX: 15.03MIN: 11.16 / MAX: 13.48MIN: 10.1 / MAX: 108.3MIN: 10.14 / MAX: 162.88MIN: 9.86 / MAX: 11.02MIN: 9.95 / MAX: 16.18MIN: 9.85 / MAX: 11.06MIN: 9.84 / MAX: 12.481. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba3691215SE +/- 0.21, N = 14SE +/- 0.01, N = 3SE +/- 0.11, N = 35.185.145.145.499.624.314.304.694.684.674.755.304.324.364.304.314.324.41MIN: 4.75 / MAX: 7.12MIN: 4.76 / MAX: 6.26MIN: 4.73 / MAX: 6.32MIN: 4.26 / MAX: 363.39MIN: 4.31 / MAX: 147.6MIN: 4.26 / MAX: 5.18MIN: 4.25 / MAX: 4.83MIN: 4.29 / MAX: 5.78MIN: 4.28 / MAX: 6.37MIN: 4.27 / MAX: 5.88MIN: 4.31 / MAX: 13.88MIN: 4.92 / MAX: 7.18MIN: 4.25 / MAX: 5.17MIN: 4.29 / MAX: 5.7MIN: 4.23 / MAX: 5.32MIN: 4.26 / MAX: 4.98MIN: 4.26 / MAX: 5.15MIN: 4.24 / MAX: 5.161. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba48121620SE +/- 0.17, N = 15SE +/- 0.02, N = 3SE +/- 0.07, N = 37.828.076.006.0814.035.295.215.635.565.685.695.605.555.695.235.245.205.29MIN: 5.54 / MAX: 303.05MIN: 5.86 / MAX: 121.03MIN: 5.47 / MAX: 7.29MIN: 4.97 / MAX: 245.95MIN: 5 / MAX: 303.38MIN: 5.18 / MAX: 6.19MIN: 5.09 / MAX: 6.04MIN: 5.08 / MAX: 7.55MIN: 5.09 / MAX: 6.84MIN: 5.17 / MAX: 7.45MIN: 5.16 / MAX: 7.68MIN: 5.13 / MAX: 6.83MIN: 5.19 / MAX: 25.4MIN: 5.22 / MAX: 92.59MIN: 5.1 / MAX: 6.28MIN: 5.15 / MAX: 6.09MIN: 5.1 / MAX: 5.9MIN: 5.09 / MAX: 6.291. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba1326395265SE +/- 0.23, N = 15SE +/- 0.05, N = 3SE +/- 0.30, N = 329.5428.1927.7528.3656.6423.4323.4325.4025.0326.1125.3730.9623.7824.1923.5123.4523.4923.75MIN: 24.77 / MAX: 364.86MIN: 24.69 / MAX: 205.72MIN: 24.58 / MAX: 282.59MIN: 24.13 / MAX: 449.57MIN: 25.75 / MAX: 367.74MIN: 23.23 / MAX: 24.39MIN: 23.2 / MAX: 24.1MIN: 24.09 / MAX: 32.86MIN: 23.85 / MAX: 28.9MIN: 24.54 / MAX: 30.29MIN: 24.26 / MAX: 36.52MIN: 25.92 / MAX: 328.63MIN: 23.52 / MAX: 24.89MIN: 23.99 / MAX: 30.98MIN: 23.19 / MAX: 24.68MIN: 23.26 / MAX: 24.51MIN: 23.36 / MAX: 24.62MIN: 23.31 / MAX: 25.121. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba48121620SE +/- 0.22, N = 15SE +/- 0.01, N = 3SE +/- 0.11, N = 38.939.5310.279.6918.257.907.868.408.388.528.4210.307.987.927.857.937.827.94MIN: 8.27 / MAX: 10.68MIN: 8.86 / MAX: 11.44MIN: 7.95 / MAX: 115.68MIN: 7.29 / MAX: 407.61MIN: 7.5 / MAX: 267.89MIN: 7.79 / MAX: 8.74MIN: 7.76 / MAX: 8.74MIN: 7.72 / MAX: 10.5MIN: 7.72 / MAX: 10.05MIN: 7.84 / MAX: 10.21MIN: 7.75 / MAX: 9.96MIN: 8.19 / MAX: 349.57MIN: 7.86 / MAX: 8.78MIN: 7.8 / MAX: 8.96MIN: 7.71 / MAX: 8.83MIN: 7.82 / MAX: 8.91MIN: 7.73 / MAX: 8.65MIN: 7.71 / MAX: 8.731. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba3691215SE +/- 0.19, N = 15SE +/- 0.00, N = 3SE +/- 0.05, N = 34.374.034.234.729.233.863.873.994.044.093.994.053.913.873.853.883.853.90MIN: 4.15 / MAX: 5.96MIN: 3.86 / MAX: 4.82MIN: 3.98 / MAX: 12.23MIN: 3.37 / MAX: 486.93MIN: 3.43 / MAX: 156.19MIN: 3.81 / MAX: 4.75MIN: 3.83 / MAX: 4.69MIN: 3.8 / MAX: 5.69MIN: 3.83 / MAX: 5.71MIN: 3.86 / MAX: 5.59MIN: 3.79 / MAX: 5.83MIN: 3.78 / MAX: 5.45MIN: 3.85 / MAX: 4.64MIN: 3.81 / MAX: 4.97MIN: 3.81 / MAX: 4.46MIN: 3.84 / MAX: 4.41MIN: 3.81 / MAX: 4.42MIN: 3.82 / MAX: 4.511. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba246810SE +/- 0.14, N = 15SE +/- 0.01, N = 3SE +/- 0.01, N = 34.773.223.123.266.882.972.993.043.063.063.052.743.002.972.982.992.972.97MIN: 3.07 / MAX: 97.57MIN: 3.11 / MAX: 3.71MIN: 2.98 / MAX: 3.79MIN: 2.46 / MAX: 277.54MIN: 3.05 / MAX: 110.25MIN: 2.93 / MAX: 3.28MIN: 2.95 / MAX: 3.88MIN: 2.91 / MAX: 4.47MIN: 2.94 / MAX: 4.45MIN: 2.94 / MAX: 4.51MIN: 2.92 / MAX: 3.82MIN: 2.62 / MAX: 4.22MIN: 2.96 / MAX: 3.68MIN: 2.93 / MAX: 3.95MIN: 2.94 / MAX: 3.83MIN: 2.96 / MAX: 3.44MIN: 2.93 / MAX: 3.45MIN: 2.92 / MAX: 3.481. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba246810SE +/- 0.19, N = 15SE +/- 0.01, N = 3SE +/- 0.01, N = 33.453.513.553.776.823.353.343.423.453.433.415.033.593.553.353.353.343.34MIN: 3.32 / MAX: 4.91MIN: 3.38 / MAX: 5.4MIN: 3.39 / MAX: 5.48MIN: 3.02 / MAX: 511.95MIN: 3.16 / MAX: 64.72MIN: 3.31 / MAX: 3.68MIN: 3.3 / MAX: 4.19MIN: 3.28 / MAX: 4.19MIN: 3.32 / MAX: 3.85MIN: 3.3 / MAX: 4.15MIN: 3.28 / MAX: 4.87MIN: 3.07 / MAX: 228.55MIN: 3.3 / MAX: 25.28MIN: 3.27 / MAX: 22.86MIN: 3.3 / MAX: 3.82MIN: 3.31 / MAX: 3.8MIN: 3.31 / MAX: 3.77MIN: 3.3 / MAX: 3.851. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba3691215SE +/- 0.17, N = 15SE +/- 0.02, N = 3SE +/- 0.00, N = 33.604.743.303.769.673.173.173.253.283.283.283.523.163.153.173.183.163.16MIN: 3.43 / MAX: 4.62MIN: 3.09 / MAX: 140.79MIN: 3.11 / MAX: 4.81MIN: 2.6 / MAX: 364.73MIN: 3.19 / MAX: 225.84MIN: 3.11 / MAX: 4.94MIN: 3.12 / MAX: 4.05MIN: 3.09 / MAX: 4.51MIN: 3.1 / MAX: 4.05MIN: 3.11 / MAX: 4MIN: 3.11 / MAX: 3.88MIN: 3.29 / MAX: 19.18MIN: 3.11 / MAX: 3.83MIN: 3.1 / MAX: 3.65MIN: 3.1 / MAX: 8.86MIN: 3.13 / MAX: 3.84MIN: 3.11 / MAX: 3.61MIN: 3.1 / MAX: 3.81. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfdcba510152025SE +/- 0.22, N = 15SE +/- 0.06, N = 3SE +/- 0.03, N = 38.158.4310.089.4317.818.018.608.388.378.418.738.3722.748.458.108.027.978.05MIN: 7.73 / MAX: 9.34MIN: 8.04 / MAX: 18.04MIN: 8.1 / MAX: 118.32MIN: 7.95 / MAX: 398.1MIN: 8.05 / MAX: 159.41MIN: 7.96 / MAX: 9.85MIN: 8.5 / MAX: 13.72MIN: 7.94 / MAX: 10.16MIN: 7.96 / MAX: 9.72MIN: 8.14 / MAX: 11.03MIN: 8.15 / MAX: 10.96MIN: 8.15 / MAX: 9.75MIN: 8.24 / MAX: 1264.67MIN: 8.37 / MAX: 9.44MIN: 7.94 / MAX: 14.4MIN: 7.98 / MAX: 8.33MIN: 7.94 / MAX: 8.26MIN: 7.97 / MAX: 9.071. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb3691215SE +/- 0.27, N = 144.064.162.934.339.184.114.044.044.194.184.205.693.973.854.084.07MIN: 3.91 / MAX: 5.78MIN: 4 / MAX: 5.58MIN: 2.84 / MAX: 3.38MIN: 2.59 / MAX: 433.58MIN: 3.64 / MAX: 122.65MIN: 4.07 / MAX: 4.21MIN: 4.01 / MAX: 4.15MIN: 3.89 / MAX: 5.01MIN: 4.04 / MAX: 5.47MIN: 4.03 / MAX: 5.07MIN: 4.06 / MAX: 4.86MIN: 3.69 / MAX: 261.71MIN: 3.92 / MAX: 4.75MIN: 3.8 / MAX: 4.65MIN: 4.05 / MAX: 4.36MIN: 4.03 / MAX: 5.831. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb20406080100SE +/- 0.18, N = 1538.9939.1238.7937.8681.7731.9331.8934.4734.3734.2734.2036.5533.3933.4731.7831.65MIN: 34.17 / MAX: 473.06MIN: 33.92 / MAX: 465.83MIN: 33.95 / MAX: 457.41MIN: 32.9 / MAX: 463.9MIN: 44.4 / MAX: 460.28MIN: 31.76 / MAX: 33.09MIN: 31.66 / MAX: 39.97MIN: 33.32 / MAX: 37.42MIN: 33.01 / MAX: 38.7MIN: 33.07 / MAX: 37.01MIN: 32.92 / MAX: 36.19MIN: 33 / MAX: 209.38MIN: 32.73 / MAX: 88.83MIN: 32.89 / MAX: 74.09MIN: 31.64 / MAX: 34.51MIN: 31.53 / MAX: 32.231. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb510152025SE +/- 0.21, N = 159.5517.158.139.0219.668.097.958.478.758.578.338.218.078.348.148.05MIN: 7.5 / MAX: 193.79MIN: 8.02 / MAX: 773.45MIN: 7.75 / MAX: 10.05MIN: 7.69 / MAX: 501.76MIN: 7.5 / MAX: 235.36MIN: 7.99 / MAX: 14.25MIN: 7.88 / MAX: 8.67MIN: 8.13 / MAX: 10.27MIN: 8.35 / MAX: 10.08MIN: 8.21 / MAX: 10.39MIN: 8.02 / MAX: 9.64MIN: 7.9 / MAX: 9.99MIN: 7.97 / MAX: 8.81MIN: 8.26 / MAX: 9.3MIN: 8.08 / MAX: 8.69MIN: 8 / MAX: 8.581. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb48121620SE +/- 0.24, N = 159.379.309.328.3917.757.097.047.357.647.597.648.167.266.977.047.03MIN: 7.07 / MAX: 281.92MIN: 6.92 / MAX: 310.91MIN: 7.1 / MAX: 172.56MIN: 6.53 / MAX: 436.05MIN: 6.47 / MAX: 272.11MIN: 7.02 / MAX: 7.99MIN: 6.96 / MAX: 7.74MIN: 6.79 / MAX: 9.82MIN: 7.03 / MAX: 9.19MIN: 7.02 / MAX: 8.87MIN: 7.05 / MAX: 9.9MIN: 7.51 / MAX: 9.94MIN: 7.14 / MAX: 8.59MIN: 6.83 / MAX: 13.87MIN: 6.96 / MAX: 7.83MIN: 6.97 / MAX: 7.881. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb714212835SE +/- 0.14, N = 1515.6215.3415.3015.4229.3412.8212.8713.8313.6913.5513.8115.1113.0813.0712.8912.98MIN: 12.99 / MAX: 184MIN: 12.94 / MAX: 157.95MIN: 12.87 / MAX: 144.73MIN: 12.21 / MAX: 414.81MIN: 12.17 / MAX: 245.34MIN: 12.72 / MAX: 13.48MIN: 12.75 / MAX: 13.58MIN: 12.89 / MAX: 15.4MIN: 12.73 / MAX: 15.68MIN: 12.75 / MAX: 14.74MIN: 12.84 / MAX: 15.1MIN: 12.93 / MAX: 151.45MIN: 12.96 / MAX: 13.83MIN: 12.95 / MAX: 14.55MIN: 12.84 / MAX: 13.19MIN: 12.73 / MAX: 35.551. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb612182430SE +/- 0.22, N = 1513.6813.8211.3912.1124.0710.0710.0511.2110.9110.7911.1114.0511.2511.0510.3310.01MIN: 10.25 / MAX: 566.67MIN: 10.34 / MAX: 245.6MIN: 10.48 / MAX: 13.29MIN: 10.16 / MAX: 382.56MIN: 10.02 / MAX: 218.35MIN: 9.94 / MAX: 11.06MIN: 9.85 / MAX: 12.64MIN: 10.3 / MAX: 13.25MIN: 9.91 / MAX: 13.1MIN: 9.91 / MAX: 12.75MIN: 10.19 / MAX: 13.03MIN: 11.69 / MAX: 252.21MIN: 10.55 / MAX: 118.12MIN: 10.46 / MAX: 112.6MIN: 10.16 / MAX: 13.97MIN: 9.89 / MAX: 10.861. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb3691215SE +/- 0.23, N = 154.675.274.945.5511.004.304.314.674.654.654.665.014.714.834.284.29MIN: 4.28 / MAX: 5.7MIN: 4.78 / MAX: 7.7MIN: 4.51 / MAX: 6.64MIN: 4.2 / MAX: 281.58MIN: 4.33 / MAX: 199.92MIN: 4.24 / MAX: 4.99MIN: 4.25 / MAX: 5.13MIN: 4.28 / MAX: 6.29MIN: 4.28 / MAX: 6.42MIN: 4.26 / MAX: 6.13MIN: 4.29 / MAX: 6.1MIN: 4.6 / MAX: 6.68MIN: 4.65 / MAX: 5.57MIN: 4.76 / MAX: 5.74MIN: 4.24 / MAX: 5.12MIN: 4.24 / MAX: 5.641. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb3691215SE +/- 0.19, N = 157.617.755.816.1811.145.205.195.715.665.635.705.865.486.135.265.21MIN: 5.23 / MAX: 90.18MIN: 5.57 / MAX: 125.43MIN: 5.27 / MAX: 7.16MIN: 5.17 / MAX: 262.79MIN: 4.79 / MAX: 65.12MIN: 5.1 / MAX: 5.97MIN: 5.09 / MAX: 6.13MIN: 5.12 / MAX: 8.19MIN: 5.14 / MAX: 7.49MIN: 5.09 / MAX: 7.75MIN: 5.15 / MAX: 7.9MIN: 5.35 / MAX: 7.79MIN: 5.37 / MAX: 6.51MIN: 5.41 / MAX: 151.51MIN: 5.18 / MAX: 6.27MIN: 5.12 / MAX: 6.221. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb1122334455SE +/- 0.24, N = 1527.0427.2528.5528.4049.7523.5423.5025.4525.0024.9125.1029.1224.9224.4523.9923.50MIN: 24.33 / MAX: 215.56MIN: 24.14 / MAX: 379.93MIN: 24.05 / MAX: 201.8MIN: 24.12 / MAX: 509.06MIN: 25.45 / MAX: 273.86MIN: 23.33 / MAX: 24.41MIN: 23.17 / MAX: 24.44MIN: 24.22 / MAX: 27.73MIN: 23.91 / MAX: 27.99MIN: 23.8 / MAX: 26.87MIN: 24.12 / MAX: 27.57MIN: 26.33 / MAX: 310.23MIN: 24.58 / MAX: 31.89MIN: 24.26 / MAX: 25.26MIN: 23.72 / MAX: 24.98MIN: 23.3 / MAX: 24.411. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb48121620SE +/- 0.24, N = 159.029.299.979.6517.007.857.838.558.508.528.4510.179.158.077.887.84MIN: 8.41 / MAX: 11.08MIN: 7.98 / MAX: 83.03MIN: 7.67 / MAX: 258.52MIN: 7.59 / MAX: 472.81MIN: 7.35 / MAX: 277.79MIN: 7.75 / MAX: 8.69MIN: 7.71 / MAX: 8.8MIN: 7.85 / MAX: 10.35MIN: 7.79 / MAX: 9.94MIN: 7.81 / MAX: 10.78MIN: 7.79 / MAX: 10.32MIN: 7.94 / MAX: 150.01MIN: 7.84 / MAX: 198.46MIN: 7.92 / MAX: 8.86MIN: 7.79 / MAX: 8.78MIN: 7.74 / MAX: 8.71. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb0.68181.36362.04542.72723.409SE +/- 0.19, N = 151.161.341.171.793.031.371.361.411.421.421.441.281.371.431.381.37MIN: 1.11 / MAX: 1.67MIN: 1.27 / MAX: 1.95MIN: 1.11 / MAX: 1.9MIN: 1.13 / MAX: 312.12MIN: 1.28 / MAX: 96.94MIN: 1.35 / MAX: 1.46MIN: 1.34 / MAX: 1.46MIN: 1.34 / MAX: 1.91MIN: 1.36 / MAX: 1.92MIN: 1.36 / MAX: 2.2MIN: 1.37 / MAX: 3.45MIN: 1.23 / MAX: 1.73MIN: 1.34 / MAX: 2.07MIN: 1.4 / MAX: 1.77MIN: 1.36 / MAX: 1.58MIN: 1.35 / MAX: 1.521. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb3691215SE +/- 0.22, N = 154.044.344.094.789.013.853.834.034.064.024.024.684.144.043.893.82MIN: 3.78 / MAX: 4.9MIN: 4.16 / MAX: 5.28MIN: 3.86 / MAX: 4.83MIN: 3.82 / MAX: 411.19MIN: 3.98 / MAX: 188.57MIN: 3.81 / MAX: 4.53MIN: 3.78 / MAX: 4.41MIN: 3.82 / MAX: 5.43MIN: 3.83 / MAX: 5.55MIN: 3.82 / MAX: 5.39MIN: 3.82 / MAX: 5.66MIN: 4.48 / MAX: 6.02MIN: 4.09 / MAX: 5.13MIN: 3.99 / MAX: 4.82MIN: 3.83 / MAX: 9.72MIN: 3.79 / MAX: 4.341. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb246810SE +/- 0.04, N = 154.613.233.193.116.872.962.953.063.073.093.083.393.053.122.972.97MIN: 2.78 / MAX: 222.99MIN: 3.1 / MAX: 3.75MIN: 3.06 / MAX: 3.75MIN: 2.8 / MAX: 4.98MIN: 2.93 / MAX: 216.41MIN: 2.94 / MAX: 3.38MIN: 2.92 / MAX: 3.29MIN: 2.93 / MAX: 3.64MIN: 2.94 / MAX: 3.6MIN: 2.95 / MAX: 4.52MIN: 2.94 / MAX: 4.52MIN: 3.26 / MAX: 4.86MIN: 3.01 / MAX: 3.88MIN: 3.08 / MAX: 3.86MIN: 2.94 / MAX: 3.43MIN: 2.94 / MAX: 3.431. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb246810SE +/- 0.21, N = 153.463.595.184.098.133.333.323.433.473.433.463.523.383.403.343.33MIN: 3.32 / MAX: 5.2MIN: 3.46 / MAX: 4.09MIN: 3.34 / MAX: 283.54MIN: 3.12 / MAX: 435.28MIN: 3.09 / MAX: 147.21MIN: 3.3 / MAX: 3.67MIN: 3.28 / MAX: 3.66MIN: 3.31 / MAX: 3.94MIN: 3.33 / MAX: 5.01MIN: 3.3 / MAX: 4.03MIN: 3.34 / MAX: 3.93MIN: 3.39 / MAX: 4.05MIN: 3.34 / MAX: 4.15MIN: 3.35 / MAX: 4.17MIN: 3.32 / MAX: 3.79MIN: 3.3 / MAX: 3.791. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb3691215SE +/- 0.10, N = 153.393.453.463.419.193.173.143.283.303.273.313.303.153.163.143.15MIN: 3.21 / MAX: 4.24MIN: 3.23 / MAX: 4.55MIN: 3.29 / MAX: 4.38MIN: 2.99 / MAX: 184.91MIN: 3.04 / MAX: 232.12MIN: 3.11 / MAX: 4.5MIN: 3.08 / MAX: 3.7MIN: 3.1 / MAX: 4MIN: 3.12 / MAX: 4.03MIN: 3.1 / MAX: 4.34MIN: 3.12 / MAX: 4.76MIN: 3.14 / MAX: 4.82MIN: 3.1 / MAX: 3.63MIN: 3.09 / MAX: 3.89MIN: 3.1 / MAX: 3.67MIN: 3.1 / MAX: 3.681. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3 - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb48121620SE +/- 0.23, N = 158.918.748.969.6217.828.038.068.408.448.488.4310.088.988.568.008.01MIN: 8.33 / MAX: 10.07MIN: 8.25 / MAX: 10.5MIN: 8.39 / MAX: 10.77MIN: 7.76 / MAX: 454.91MIN: 7.57 / MAX: 211.62MIN: 7.98 / MAX: 8.77MIN: 7.94 / MAX: 13.92MIN: 8.12 / MAX: 10.11MIN: 7.97 / MAX: 10.71MIN: 7.96 / MAX: 10.32MIN: 7.99 / MAX: 10.44MIN: 8.08 / MAX: 286.28MIN: 8.1 / MAX: 124.43MIN: 8.04 / MAX: 75.44MIN: 7.96 / MAX: 8.63MIN: 7.95 / MAX: 8.951. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb48121620SE +/- 0.25, N = 147.7310.349.608.8917.238.258.018.108.588.448.457.997.998.507.988.27MIN: 7.43 / MAX: 9.41MIN: 8.21 / MAX: 214.16MIN: 7.66 / MAX: 210.23MIN: 7.74 / MAX: 476.28MIN: 7.8 / MAX: 193.14MIN: 8.12 / MAX: 14MIN: 7.93 / MAX: 8.35MIN: 7.77 / MAX: 15.42MIN: 8.23 / MAX: 10.39MIN: 8.04 / MAX: 10.17MIN: 8.05 / MAX: 10.3MIN: 7.62 / MAX: 9.27MIN: 7.91 / MAX: 8.8MIN: 8.04 / MAX: 30.12MIN: 7.93 / MAX: 8.65MIN: 8.22 / MAX: 9.011. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb246810SE +/- 0.15, N = 155.924.163.944.268.634.104.044.124.314.094.204.433.974.203.694.06MIN: 4.25 / MAX: 103.26MIN: 4.03 / MAX: 4.73MIN: 3.8 / MAX: 5.41MIN: 2.71 / MAX: 347.03MIN: 4.27 / MAX: 144.3MIN: 4.06 / MAX: 4.21MIN: 4 / MAX: 4.15MIN: 3.97 / MAX: 6.99MIN: 4.14 / MAX: 6.11MIN: 3.92 / MAX: 5.5MIN: 4.04 / MAX: 5.82MIN: 4.28 / MAX: 5.01MIN: 3.93 / MAX: 4.73MIN: 4.15 / MAX: 4.92MIN: 3.66 / MAX: 3.92MIN: 4.03 / MAX: 4.31. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb1632486480SE +/- 0.13, N = 1538.5838.7339.0138.2973.5132.1131.8634.0535.4034.2934.1338.3332.6833.3631.6631.71MIN: 33.77 / MAX: 476.18MIN: 33.81 / MAX: 362.17MIN: 33.91 / MAX: 411.66MIN: 32.31 / MAX: 557.38MIN: 39.27 / MAX: 288.2MIN: 31.94 / MAX: 33.01MIN: 31.58 / MAX: 35.84MIN: 32.83 / MAX: 38.57MIN: 33.93 / MAX: 39.3MIN: 33.11 / MAX: 40.12MIN: 32.98 / MAX: 36.11MIN: 34.14 / MAX: 246.43MIN: 32.02 / MAX: 87.72MIN: 32.83 / MAX: 76.21MIN: 31.52 / MAX: 32.14MIN: 31.56 / MAX: 33.031. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb48121620SE +/- 0.19, N = 157.727.817.938.2916.157.087.047.517.707.637.668.337.317.097.077.14MIN: 7.12 / MAX: 23.25MIN: 7.24 / MAX: 9.04MIN: 7.31 / MAX: 9.45MIN: 6.37 / MAX: 448.22MIN: 7.25 / MAX: 210.69MIN: 7.01 / MAX: 7.93MIN: 6.97 / MAX: 7.76MIN: 6.94 / MAX: 9.51MIN: 7.11 / MAX: 9.19MIN: 7.02 / MAX: 9.71MIN: 7.02 / MAX: 9.08MIN: 6.32 / MAX: 222.03MIN: 6.96 / MAX: 30.1MIN: 6.98 / MAX: 8.01MIN: 7.01 / MAX: 7.75MIN: 7.06 / MAX: 7.951. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb714212835SE +/- 0.23, N = 1516.6116.3915.4415.4429.4912.8412.8813.6213.9513.6813.7915.4313.3514.3412.8612.77MIN: 12.32 / MAX: 375.99MIN: 12.97 / MAX: 369.64MIN: 12.92 / MAX: 211.43MIN: 12.61 / MAX: 387.62MIN: 13.03 / MAX: 182.99MIN: 12.76 / MAX: 13.7MIN: 12.75 / MAX: 13.79MIN: 12.75 / MAX: 15.79MIN: 13.03 / MAX: 15.9MIN: 12.77 / MAX: 15.57MIN: 12.79 / MAX: 15.92MIN: 13.1 / MAX: 210.2MIN: 12.87 / MAX: 58.52MIN: 14.23 / MAX: 15.12MIN: 12.76 / MAX: 13.98MIN: 12.69 / MAX: 13.711. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb612182430SE +/- 0.27, N = 1513.1313.5714.5812.3523.1110.019.9711.0911.5010.8410.9511.1510.4310.2510.039.87MIN: 10.56 / MAX: 323.44MIN: 10.45 / MAX: 199.55MIN: 10.67 / MAX: 324.82MIN: 9.83 / MAX: 424.28MIN: 10.22 / MAX: 140.41MIN: 9.91 / MAX: 10.74MIN: 9.86 / MAX: 10.84MIN: 10.18 / MAX: 13.12MIN: 10.5 / MAX: 13.47MIN: 9.93 / MAX: 12.83MIN: 9.91 / MAX: 17.11MIN: 10.31 / MAX: 12.97MIN: 10.19 / MAX: 11.32MIN: 10.05 / MAX: 11.08MIN: 9.93 / MAX: 10.96MIN: 9.79 / MAX: 10.731. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb3691215SE +/- 0.23, N = 156.116.586.115.679.864.304.304.655.214.694.694.994.864.354.304.42MIN: 4.83 / MAX: 124.76MIN: 4.61 / MAX: 91.07MIN: 4.73 / MAX: 81.72MIN: 4.21 / MAX: 365.75MIN: 4.25 / MAX: 157.02MIN: 4.25 / MAX: 4.7MIN: 4.25 / MAX: 5.08MIN: 4.26 / MAX: 5.97MIN: 4.79 / MAX: 6.66MIN: 4.26 / MAX: 7.17MIN: 4.26 / MAX: 6.15MIN: 4.59 / MAX: 6.56MIN: 4.8 / MAX: 6.37MIN: 4.27 / MAX: 5.16MIN: 4.26 / MAX: 5.16MIN: 4.32 / MAX: 5.11. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb3691215SE +/- 0.16, N = 155.845.815.976.2313.385.245.235.595.895.695.655.855.505.305.235.42MIN: 5.35 / MAX: 7.72MIN: 5.3 / MAX: 6.82MIN: 5.46 / MAX: 7.02MIN: 4.99 / MAX: 309.18MIN: 5.43 / MAX: 208.42MIN: 5.14 / MAX: 5.99MIN: 5.1 / MAX: 6.07MIN: 5.09 / MAX: 7.7MIN: 5.36 / MAX: 7.53MIN: 5.11 / MAX: 6.94MIN: 5.14 / MAX: 6.93MIN: 5.3 / MAX: 8.27MIN: 5.4 / MAX: 6.38MIN: 5.17 / MAX: 5.93MIN: 5.11 / MAX: 6.03MIN: 5.36 / MAX: 6.271. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb1224364860SE +/- 0.27, N = 1527.7727.5928.2128.4055.4223.4323.5025.2626.0825.0425.0429.0724.7124.1223.5423.42MIN: 24.82 / MAX: 264.66MIN: 24.34 / MAX: 396.09MIN: 24.57 / MAX: 270.76MIN: 23.98 / MAX: 456MIN: 25.32 / MAX: 281.46MIN: 23.26 / MAX: 24.3MIN: 23.23 / MAX: 24.26MIN: 24.14 / MAX: 27.73MIN: 24.52 / MAX: 27.73MIN: 23.81 / MAX: 27.15MIN: 23.87 / MAX: 28.04MIN: 24.45 / MAX: 263.33MIN: 23.88 / MAX: 119.23MIN: 23.57 / MAX: 46.44MIN: 23.32 / MAX: 24.54MIN: 23.27 / MAX: 24.321. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb510152025SE +/- 0.21, N = 1510.018.908.879.8620.727.857.828.378.998.588.4910.198.357.947.837.97MIN: 7.29 / MAX: 259.11MIN: 8.22 / MAX: 11.07MIN: 8.18 / MAX: 11.09MIN: 7.54 / MAX: 396.21MIN: 7.49 / MAX: 355.33MIN: 7.75 / MAX: 8.64MIN: 7.69 / MAX: 8.6MIN: 7.76 / MAX: 10.31MIN: 8.25 / MAX: 10.27MIN: 7.79 / MAX: 10.48MIN: 7.82 / MAX: 11.98MIN: 7.73 / MAX: 212.36MIN: 8.2 / MAX: 9.39MIN: 7.8 / MAX: 8.78MIN: 7.74 / MAX: 8.61MIN: 7.89 / MAX: 8.71. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb0.71551.4312.14652.8623.5775SE +/- 0.18, N = 151.071.411.331.713.181.381.361.391.421.451.421.251.381.371.361.37MIN: 1.02 / MAX: 1.52MIN: 1.35 / MAX: 1.89MIN: 1.27 / MAX: 1.98MIN: 1.09 / MAX: 448.17MIN: 1.31 / MAX: 185.03MIN: 1.36 / MAX: 1.9MIN: 1.34 / MAX: 1.61MIN: 1.34 / MAX: 1.89MIN: 1.36 / MAX: 1.92MIN: 1.36 / MAX: 8.73MIN: 1.35 / MAX: 2.15MIN: 1.19 / MAX: 2.61MIN: 1.36 / MAX: 1.76MIN: 1.35 / MAX: 1.62MIN: 1.34 / MAX: 1.44MIN: 1.35 / MAX: 1.391. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb246810SE +/- 0.18, N = 155.266.284.184.737.813.853.853.954.224.044.054.214.633.853.823.85MIN: 3.48 / MAX: 250.88MIN: 3.91 / MAX: 337.73MIN: 4 / MAX: 5.25MIN: 3.79 / MAX: 418.72MIN: 3.73 / MAX: 159.47MIN: 3.81 / MAX: 4.62MIN: 3.8 / MAX: 4.43MIN: 3.76 / MAX: 4.84MIN: 4 / MAX: 5.58MIN: 3.81 / MAX: 5.08MIN: 3.83 / MAX: 5MIN: 3.96 / MAX: 4.94MIN: 3.8 / MAX: 159.43MIN: 3.8 / MAX: 4.6MIN: 3.78 / MAX: 4.53MIN: 3.82 / MAX: 4.481. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb246810SE +/- 0.16, N = 142.543.103.003.376.022.972.973.013.133.073.092.992.982.962.962.96MIN: 2.44 / MAX: 3.58MIN: 2.97 / MAX: 3.72MIN: 2.89 / MAX: 3.46MIN: 2.86 / MAX: 278.87MIN: 2.79 / MAX: 50.49MIN: 2.94 / MAX: 3.39MIN: 2.92 / MAX: 3.28MIN: 2.91 / MAX: 3.6MIN: 3 / MAX: 5.1MIN: 2.94 / MAX: 3.72MIN: 2.94 / MAX: 3.79MIN: 2.86 / MAX: 4.38MIN: 2.95 / MAX: 3.63MIN: 2.92 / MAX: 3.81MIN: 2.93 / MAX: 3.41MIN: 2.93 / MAX: 3.41. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb1.10032.20063.30094.40125.5015SE +/- 0.22, N = 153.173.403.343.954.893.363.343.373.513.443.443.363.353.333.333.33MIN: 3.04 / MAX: 3.78MIN: 3.26 / MAX: 4.84MIN: 3.23 / MAX: 4.78MIN: 3.19 / MAX: 410.41MIN: 3.04 / MAX: 18.32MIN: 3.32 / MAX: 3.7MIN: 3.31 / MAX: 3.6MIN: 3.25 / MAX: 3.95MIN: 3.37 / MAX: 4.26MIN: 3.32 / MAX: 4.16MIN: 3.31 / MAX: 4.88MIN: 3.25 / MAX: 4.02MIN: 3.31 / MAX: 4.01MIN: 3.29 / MAX: 3.99MIN: 3.31 / MAX: 3.81MIN: 3.3 / MAX: 3.771. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb246810SE +/- 0.18, N = 154.453.314.993.667.243.173.163.233.403.303.293.283.173.163.153.15MIN: 2.65 / MAX: 216.76MIN: 3.12 / MAX: 4.6MIN: 3.1 / MAX: 201.8MIN: 3.01 / MAX: 437.59MIN: 3.04 / MAX: 261.68MIN: 3.12 / MAX: 4.03MIN: 3.11 / MAX: 3.51MIN: 3.06 / MAX: 4.66MIN: 3.23 / MAX: 4.8MIN: 3.12 / MAX: 4.7MIN: 3.12 / MAX: 4.64MIN: 3.09 / MAX: 5.28MIN: 3.13 / MAX: 3.58MIN: 3.1 / MAX: 3.71MIN: 3.11 / MAX: 3.85MIN: 3.11 / MAX: 3.881. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: Vulkan GPU-v3-v3-v3 - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfcb48121620SE +/- 0.26, N = 1510.549.548.819.6216.348.018.008.388.888.468.439.058.208.657.958.00MIN: 8.41 / MAX: 134.08MIN: 8.94 / MAX: 10.54MIN: 8.32 / MAX: 10.7MIN: 7.76 / MAX: 502.83MIN: 8.13 / MAX: 80.69MIN: 7.95 / MAX: 8.35MIN: 7.94 / MAX: 8.78MIN: 7.95 / MAX: 10.41MIN: 8.31 / MAX: 10.01MIN: 7.99 / MAX: 10.62MIN: 7.99 / MAX: 10.66MIN: 8.48 / MAX: 11.28MIN: 8.12 / MAX: 9.4MIN: 8.55 / MAX: 9.53MIN: 7.89 / MAX: 8.79MIN: 7.95 / MAX: 8.991. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

VkFFT

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT C2C 1D batched in single precisionnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080ihgfedcba30K60K90K120K150KSE +/- 25.50, N = 3SE +/- 1.67, N = 3SE +/- 2.73, N = 3SE +/- 9.54, N = 31521701539391538961414371413571045431045281044911045566973856431564555647642651426454797147948478871. (CXX) g++ options: -O3

Test: FFT + iFFT C2C 1D batched in single precision

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

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: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep246810SE +/- 0.15, N = 35.864.594.624.147.234.07MIN: 3.9 / MAX: 190.17MIN: 4.44 / MAX: 5.2MIN: 4.48 / MAX: 5.16MIN: 3.73 / MAX: 5.07MIN: 3.75 / MAX: 121.71MIN: 4.04 / MAX: 4.251. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep1632486480SE +/- 0.10, N = 337.1338.6539.3538.5070.5331.94MIN: 33.97 / MAX: 443.1MIN: 33.07 / MAX: 476.08MIN: 34.22 / MAX: 466.65MIN: 33.7 / MAX: 418.06MIN: 39.2 / MAX: 276.33MIN: 31.73 / MAX: 32.751. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep48121620SE +/- 0.54, N = 38.348.7010.119.1417.028.06MIN: 8.01 / MAX: 12.36MIN: 8.29 / MAX: 12.6MIN: 8.03 / MAX: 259.38MIN: 8.14 / MAX: 400.02MIN: 7.65 / MAX: 216.63MIN: 7.98 / MAX: 8.61. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep48121620SE +/- 0.14, N = 38.269.447.437.4515.327.07MIN: 7.64 / MAX: 11.08MIN: 7.17 / MAX: 94.63MIN: 6.84 / MAX: 8.82MIN: 6.59 / MAX: 9.11MIN: 6.66 / MAX: 139.17MIN: 6.98 / MAX: 9.711. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep714212835SE +/- 0.94, N = 316.3015.4116.0514.6429.3812.92MIN: 14.11 / MAX: 184.46MIN: 12.75 / MAX: 226.87MIN: 12.93 / MAX: 474.03MIN: 12.77 / MAX: 383.28MIN: 12.95 / MAX: 201.31MIN: 12.79 / MAX: 18.51. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep510152025SE +/- 0.30, N = 313.2510.9613.0013.1522.1510.27MIN: 10.61 / MAX: 154.12MIN: 10.09 / MAX: 12.99MIN: 10.34 / MAX: 397.57MIN: 10.26 / MAX: 349.93MIN: 10.11 / MAX: 123.04MIN: 10.12 / MAX: 11.191. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep3691215SE +/- 0.57, N = 36.325.345.146.2511.434.31MIN: 4.26 / MAX: 195.95MIN: 4.87 / MAX: 6.57MIN: 4.75 / MAX: 7.34MIN: 4.27 / MAX: 334.55MIN: 4.24 / MAX: 178.83MIN: 4.26 / MAX: 4.831. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep3691215SE +/- 0.05, N = 35.585.876.965.9412.135.27MIN: 5.09 / MAX: 6.98MIN: 5.41 / MAX: 7.58MIN: 5.3 / MAX: 242.18MIN: 5.32 / MAX: 8.32MIN: 5.32 / MAX: 123.4MIN: 5.15 / MAX: 6.111. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep1224364860SE +/- 0.28, N = 327.2529.8527.3227.8653.4823.72MIN: 24.12 / MAX: 252.53MIN: 24.25 / MAX: 400.86MIN: 24.36 / MAX: 262.38MIN: 24.17 / MAX: 416.36MIN: 25.52 / MAX: 296.52MIN: 23.56 / MAX: 24.591. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep510152025SE +/- 0.55, N = 310.1410.478.559.9718.807.86MIN: 7.85 / MAX: 257.61MIN: 7.86 / MAX: 191.94MIN: 7.85 / MAX: 11.39MIN: 8.16 / MAX: 381.49MIN: 7.78 / MAX: 141.46MIN: 7.75 / MAX: 8.571. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep0.60531.21061.81592.42123.0265SE +/- 0.04, N = 31.401.421.351.402.691.38MIN: 1.34 / MAX: 1.86MIN: 1.36 / MAX: 2.03MIN: 1.28 / MAX: 1.84MIN: 1.28 / MAX: 1.91MIN: 1.35 / MAX: 48.81MIN: 1.36 / MAX: 1.731. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep246810SE +/- 0.08, N = 35.824.104.634.176.633.84MIN: 3.98 / MAX: 197.79MIN: 3.88 / MAX: 5.04MIN: 4.38 / MAX: 6.01MIN: 3.86 / MAX: 5.52MIN: 3.75 / MAX: 22.34MIN: 3.8 / MAX: 4.671. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep246810SE +/- 0.02, N = 33.103.123.233.128.552.96MIN: 2.97 / MAX: 3.73MIN: 3 / MAX: 4.1MIN: 3.08 / MAX: 4.73MIN: 2.97 / MAX: 4.65MIN: 2.99 / MAX: 185.5MIN: 2.92 / MAX: 3.271. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep1.32532.65063.97595.30126.6265SE +/- 0.02, N = 33.435.233.563.485.893.32MIN: 3.29 / MAX: 5.31MIN: 3.34 / MAX: 185.57MIN: 3.43 / MAX: 4.24MIN: 3.33 / MAX: 5.22MIN: 3.19 / MAX: 97.88MIN: 3.29 / MAX: 3.621. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti30703090 rep246810SE +/- 0.04, N = 34.963.353.363.247.343.19MIN: 3.14 / MAX: 189.43MIN: 3.22 / MAX: 3.99MIN: 3.22 / MAX: 4.62MIN: 3.05 / MAX: 5.14MIN: 3.09 / MAX: 155.33MIN: 3.13 / MAX: 3.611. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep1.3322.6643.9965.3286.66SE +/- 0.53, N = 33.293.384.753.835.923.15MIN: 3.12 / MAX: 4.27MIN: 3.2 / MAX: 4MIN: 2.93 / MAX: 147.66MIN: 3.11 / MAX: 343.21MIN: 3.16 / MAX: 103.24MIN: 3.1 / MAX: 3.751. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep48121620SE +/- 0.13, N = 310.648.2210.5610.0217.068.03MIN: 8.4 / MAX: 127.99MIN: 7.75 / MAX: 9.41MIN: 8.32 / MAX: 239.95MIN: 7.8 / MAX: 372.36MIN: 8 / MAX: 101.45MIN: 7.97 / MAX: 8.911. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v3-v3 - Model: mobilenet-v3nv 40904090 rep4090RTX 3070 Ti30703090 rep4080 zzz4080 xxx4080 rep246810SE +/- 0.53, N = 34.973.303.333.706.433.173.063.083.28MIN: 3.15 / MAX: 291.01MIN: 3.15 / MAX: 3.91MIN: 3.2 / MAX: 4.4MIN: 2.98 / MAX: 261.6MIN: 2.85 / MAX: 164.91MIN: 3.12 / MAX: 3.75MIN: 2.94 / MAX: 3.94MIN: 2.97 / MAX: 3.67MIN: 3.13 / MAX: 4.781. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

VkFFT

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT C2C 1D batched in single precision, no reshufflingnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080igfedcba30K60K90K120K150KSE +/- 2.08, N = 3SE +/- 2.33, N = 3SE +/- 8.89, N = 315514815593615265614395614396910592610609910620510621071163570945711043365433655059650643505041. (CXX) g++ options: -O3

Test: FFT + iFFT C2C 1D batched in single precision, no reshuffling

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

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: CPU-v3-v3-v3-v3-v3-v3 - Model: FastestDetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep246810SE +/- 0.87, N = 33.933.122.854.187.124.074.103.823.804.20MIN: 3.76 / MAX: 11.77MIN: 2.97 / MAX: 4.42MIN: 2.74 / MAX: 4.36MIN: 2.53 / MAX: 295.11MIN: 3.72 / MAX: 188.7MIN: 4.03 / MAX: 4.2MIN: 4.07 / MAX: 4.34MIN: 3.65 / MAX: 9.77MIN: 3.65 / MAX: 6.08MIN: 4.04 / MAX: 5.631. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: vision_transformernv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep1530456075SE +/- 0.11, N = 338.9538.7938.7638.0465.4131.8532.1034.4734.1434.22MIN: 34.04 / MAX: 486.96MIN: 34.02 / MAX: 460.15MIN: 33.38 / MAX: 423.24MIN: 33.11 / MAX: 346.94MIN: 39.08 / MAX: 230.59MIN: 31.67 / MAX: 35.74MIN: 31.9 / MAX: 33.03MIN: 33.05 / MAX: 39.69MIN: 32.5 / MAX: 37.13MIN: 33.01 / MAX: 37.091. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: regnety_400mnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep48121620SE +/- 0.29, N = 38.2510.2310.098.4218.258.038.228.348.388.72MIN: 7.87 / MAX: 10.07MIN: 8.22 / MAX: 197.1MIN: 8.01 / MAX: 418.58MIN: 7.66 / MAX: 10.74MIN: 7.8 / MAX: 238.29MIN: 7.97 / MAX: 8.65MIN: 8.14 / MAX: 8.67MIN: 8.03 / MAX: 10.23MIN: 8.04 / MAX: 9.63MIN: 8.32 / MAX: 10.481. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: squeezenet_ssdnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep48121620SE +/- 0.23, N = 37.489.389.517.5714.277.097.127.257.277.67MIN: 6.85 / MAX: 9.67MIN: 6.77 / MAX: 224.11MIN: 7.11 / MAX: 307.17MIN: 6.69 / MAX: 10MIN: 7.01 / MAX: 51.13MIN: 7.02 / MAX: 7.86MIN: 7.04 / MAX: 7.97MIN: 6.72 / MAX: 8.05MIN: 6.73 / MAX: 8.77MIN: 7.06 / MAX: 9.961. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: yolov4-tinynv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep714212835SE +/- 0.81, N = 317.6713.8815.8514.5728.4112.8112.8213.4213.6313.71MIN: 14.92 / MAX: 343.93MIN: 13.09 / MAX: 14.77MIN: 13.26 / MAX: 253.23MIN: 12.33 / MAX: 312.42MIN: 12.49 / MAX: 151.04MIN: 12.7 / MAX: 13.69MIN: 12.72 / MAX: 13.66MIN: 12.65 / MAX: 16.19MIN: 12.77 / MAX: 16.93MIN: 12.78 / MAX: 15.621. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: resnet50nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep510152025SE +/- 0.04, N = 313.2912.4711.7212.8122.1910.0610.0311.1011.2610.86MIN: 10.54 / MAX: 456.82MIN: 11.5 / MAX: 14.68MIN: 10.8 / MAX: 12.8MIN: 10.06 / MAX: 349.03MIN: 10.16 / MAX: 181.74MIN: 9.86 / MAX: 11.9MIN: 9.93 / MAX: 10.87MIN: 10.19 / MAX: 18.3MIN: 10.32 / MAX: 13.29MIN: 9.98 / MAX: 12.461. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: alexnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep3691215SE +/- 0.43, N = 36.625.454.996.1710.594.314.334.664.694.68MIN: 4.28 / MAX: 339.62MIN: 4.93 / MAX: 7.98MIN: 4.56 / MAX: 6.91MIN: 4.5 / MAX: 261.75MIN: 4.3 / MAX: 177.68MIN: 4.26 / MAX: 5.07MIN: 4.26 / MAX: 5.19MIN: 4.24 / MAX: 5.97MIN: 4.26 / MAX: 6.07MIN: 4.27 / MAX: 6.081. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: resnet18nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep3691215SE +/- 0.30, N = 36.078.147.746.2212.645.305.205.775.785.64MIN: 5.49 / MAX: 15.12MIN: 5.39 / MAX: 122.47MIN: 5.25 / MAX: 312.09MIN: 5.3 / MAX: 8.22MIN: 5.3 / MAX: 53.81MIN: 5.21 / MAX: 6.24MIN: 5.1 / MAX: 6.16MIN: 5.22 / MAX: 7.06MIN: 5.21 / MAX: 6.97MIN: 5.11 / MAX: 7.511. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: vgg16nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep1122334455SE +/- 0.54, N = 327.6130.7430.1627.9850.3223.4023.5825.2625.4425.01MIN: 24.67 / MAX: 401.29MIN: 25.36 / MAX: 428.68MIN: 24.66 / MAX: 332.49MIN: 24.35 / MAX: 423.63MIN: 25.92 / MAX: 281.06MIN: 23.2 / MAX: 24.07MIN: 23.35 / MAX: 24.43MIN: 24.29 / MAX: 27.75MIN: 24.27 / MAX: 27.68MIN: 23.88 / MAX: 26.661. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: googlenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep510152025SE +/- 0.80, N = 310.758.978.389.6819.207.917.908.298.328.52MIN: 7.92 / MAX: 447.83MIN: 8.22 / MAX: 10.51MIN: 7.78 / MAX: 10.43MIN: 8.16 / MAX: 382.41MIN: 7.84 / MAX: 193.36MIN: 7.81 / MAX: 8.62MIN: 7.8 / MAX: 8.73MIN: 7.63 / MAX: 9.87MIN: 7.71 / MAX: 10.39MIN: 7.85 / MAX: 10.561. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: blazefacenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep0.56931.13861.70792.27722.8465SE +/- 0.48, N = 31.421.421.302.482.531.381.391.311.321.43MIN: 1.34 / MAX: 1.99MIN: 1.34 / MAX: 2.37MIN: 1.24 / MAX: 1.92MIN: 1.17 / MAX: 344.52MIN: 1.08 / MAX: 118.73MIN: 1.35 / MAX: 1.64MIN: 1.37 / MAX: 1.48MIN: 1.25 / MAX: 1.76MIN: 1.26 / MAX: 2.03MIN: 1.36 / MAX: 2.021. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: efficientnet-b0nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep3691215SE +/- 0.46, N = 35.944.354.474.749.193.853.883.954.014.07MIN: 3.97 / MAX: 208.59MIN: 4.08 / MAX: 5.62MIN: 4.23 / MAX: 5.82MIN: 3.68 / MAX: 295.7MIN: 3.85 / MAX: 131.42MIN: 3.81 / MAX: 4.75MIN: 3.83 / MAX: 4.61MIN: 3.79 / MAX: 4.59MIN: 3.83 / MAX: 5.28MIN: 3.85 / MAX: 4.791. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: mnasnetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep246810SE +/- 0.13, N = 33.125.115.193.246.072.972.982.963.003.09MIN: 2.98 / MAX: 3.71MIN: 2.96 / MAX: 247.47MIN: 3.04 / MAX: 436.91MIN: 2.9 / MAX: 5.34MIN: 2.94 / MAX: 129.1MIN: 2.93 / MAX: 3.3MIN: 2.95 / MAX: 3.9MIN: 2.85 / MAX: 3.82MIN: 2.88 / MAX: 4.37MIN: 2.96 / MAX: 4.981. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: shufflenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep246810SE +/- 0.60, N = 33.323.423.484.027.813.333.363.363.403.47MIN: 3.19 / MAX: 4.76MIN: 3.29 / MAX: 3.94MIN: 3.35 / MAX: 4.05MIN: 3.27 / MAX: 328.59MIN: 3.3 / MAX: 131.26MIN: 3.3 / MAX: 3.78MIN: 3.32 / MAX: 3.66MIN: 3.23 / MAX: 3.99MIN: 3.28 / MAX: 3.87MIN: 3.33 / MAX: 5.391. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3-v2-v2 - Model: mobilenet-v2nv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep246810SE +/- 0.53, N = 33.293.443.363.917.223.163.163.163.203.30MIN: 3.13 / MAX: 4.29MIN: 3.27 / MAX: 4.93MIN: 3.21 / MAX: 4.78MIN: 3.04 / MAX: 394.66MIN: 3.17 / MAX: 69.66MIN: 3.09 / MAX: 4.06MIN: 3.11 / MAX: 3.95MIN: 3.01 / MAX: 5.17MIN: 3.05 / MAX: 4.67MIN: 3.11 / MAX: 4.011. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20230517Target: CPU-v3-v3-v3-v3-v3-v3 - Model: mobilenetnv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep48121620SE +/- 0.14, N = 312.1210.619.0410.0316.528.068.078.258.348.45MIN: 9.16 / MAX: 505.01MIN: 8.34 / MAX: 225.97MIN: 8.49 / MAX: 10.96MIN: 7.86 / MAX: 346.64MIN: 7.9 / MAX: 82.53MIN: 8 / MAX: 8.96MIN: 8.01 / MAX: 8.62MIN: 7.78 / MAX: 9.61MIN: 7.89 / MAX: 9.42MIN: 8.01 / MAX: 10.861. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

VkFFT

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT C2C 1D batched in half precisionnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080ihgfedcba60K120K180K240K300KSE +/- 133.47, N = 3SE +/- 26.03, N = 3SE +/- 18.50, N = 3SE +/- 83.55, N = 329276828765129034226517125520721099121071321105821107613227010429810417110414685191851819174491812915971. (CXX) g++ options: -O3

Test: FFT + iFFT C2C 1D batched in half precision

3070: The test quit with a non-zero exit status.

RTX 3070 Ti: The test quit with a non-zero exit status.

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT C2C multidimensional in single precisionnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080igfedcba20K40K60K80K100KSE +/- 555.86, N = 3SE +/- 437.33, N = 3SE +/- 116.12, N = 3SE +/- 57.83, N = 382875809998140654814510057004067887700686586934686265412623837090363283281232751330011. (CXX) g++ options: -O3

Test: FFT + iFFT C2C multidimensional in single precision

3070: The test quit with a non-zero exit status. E: VkFFT System: 3840x2160x1 Buffer: 63 MB avg_time_per_step: 2.236 ms std_error: 0.035 num_iter: 64 benchmark: 28982 bandwidth: 331.7

RTX 3070 Ti: The test quit with a non-zero exit status. E: VkFFT System: 3840x2160x1 Buffer: 63 MB avg_time_per_step: 1.462 ms std_error: 0.004 num_iter: 64 benchmark: 44332 bandwidth: 507.3

OpenBenchmarking.orgBenchmark Score, More Is BetterVkFFT 1.2.31Test: FFT + iFFT R2C / C2Rnv 40904090 rep40903090 rep30904080 zzz4080 xxx4080 rep4080ihgfedcba20K40K60K80K100KSE +/- 796.66, N = 3SE +/- 3.71, N = 3SE +/- 118.74, N = 3SE +/- 200.55, N = 38488781329843515443255347676896906868279664733372726524266382659335304353994302142163421051. (CXX) g++ options: -O3

Test: FFT + iFFT R2C / C2R

3070: The test quit with a non-zero exit status. E: VkFFT System: 512x512x128 Buffer: 128 MB avg_time_per_step: 4.833 ms std_error: 0.038 num_iter: 31 benchmark: 27226 bandwidth: 311.6

RTX 3070 Ti: The test quit with a non-zero exit status. E: VkFFT System: 512x512x128 Buffer: 128 MB avg_time_per_step: 3.494 ms std_error: 0.002 num_iter: 31 benchmark: 37664 bandwidth: 431.0

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: Singlenv 40904090 rep4090RTX 3070 Ti30703090 rep30904080 zzz4080 xxx4080 rep4080igfedcba816243240SE +/- 0.029, N = 3SE +/- 0.000, N = 3SE +/- 0.004, N = 3SE +/- 0.001, N = 38.9678.9629.28427.18322.06410.42810.39913.12613.13713.13613.13620.93026.76926.73832.85032.85511.68811.69011.6861. (CXX) g++ options: -O3

173 Results Shown

vkpeak:
  fp16-vec4
  int32-scalar
  int16-vec4
  int32-vec4
  int16-scalar
  fp16-scalar
  fp32-vec4
  fp32-scalar
  fp64-scalar
  fp64-vec4
NCNN:
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - FastestDet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vision_transformer
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - regnety_400m
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - squeezenet_ssd
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - yolov4-tiny
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet50
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - alexnet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet18
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vgg16
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - googlenet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - blazeface
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - efficientnet-b0
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mnasnet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - shufflenet-v2
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet
VkFFT
VkResample
NCNN:
  Vulkan GPU-v3-v3 - mobilenet-v3
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - FastestDet
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vision_transformer
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - regnety_400m
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - squeezenet_ssd
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - yolov4-tiny
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet50
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - alexnet
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet18
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vgg16
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - googlenet
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - blazeface
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - efficientnet-b0
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mnasnet
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - shufflenet-v2
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet
VkFFT
NCNN
VkFFT
NCNN:
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - yolov4-tiny
  CPU-v3-v3 - mobilenet-v3
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - FastestDet
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - vision_transformer
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - regnety_400m
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - squeezenet_ssd
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - resnet50
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - alexnet
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - resnet18
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - vgg16
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - googlenet
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - blazeface
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - efficientnet-b0
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - mnasnet
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - shufflenet-v2
  Vulkan GPU-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2
  Vulkan GPU-v3-v3-v3-v3-v3-v3 - mobilenet
  Vulkan GPU - FastestDet
  Vulkan GPU - vision_transformer
  Vulkan GPU - regnety_400m
  Vulkan GPU - squeezenet_ssd
  Vulkan GPU - yolov4-tiny
  Vulkan GPU - resnet50
  Vulkan GPU - alexnet
  Vulkan GPU - resnet18
  Vulkan GPU - vgg16
  Vulkan GPU - googlenet
  Vulkan GPU - blazeface
  Vulkan GPU - efficientnet-b0
  Vulkan GPU - mnasnet
  Vulkan GPU - shufflenet-v2
  Vulkan GPU-v2-v2 - mobilenet-v2
  Vulkan GPU - mobilenet
  CPU - FastestDet
  CPU - blazeface
  Vulkan GPU-v3-v3-v3-v3-v3 - mobilenet-v3
  CPU-v3-v3-v3-v3-v3 - mobilenet-v3
  CPU - vision_transformer
  CPU - regnety_400m
  CPU - squeezenet_ssd
  CPU - yolov4-tiny
  CPU - resnet50
  CPU - alexnet
  CPU - resnet18
  CPU - vgg16
  CPU - googlenet
  CPU - efficientnet-b0
  CPU - mnasnet
  CPU - shufflenet-v2
  CPU-v2-v2 - mobilenet-v2
  CPU - mobilenet
  CPU-v3-v3-v3 - FastestDet
  CPU-v3-v3-v3 - vision_transformer
  CPU-v3-v3-v3 - regnety_400m
  CPU-v3-v3-v3 - squeezenet_ssd
  CPU-v3-v3-v3 - yolov4-tiny
  CPU-v3-v3-v3 - resnet50
  CPU-v3-v3-v3 - alexnet
  CPU-v3-v3-v3 - resnet18
  CPU-v3-v3-v3 - vgg16
  CPU-v3-v3-v3 - googlenet
  CPU-v3-v3-v3 - blazeface
  CPU-v3-v3-v3 - efficientnet-b0
  CPU-v3-v3-v3 - mnasnet
  CPU-v3-v3-v3 - shufflenet-v2
  CPU-v3-v3-v3-v2-v2 - mobilenet-v2
  CPU-v3-v3-v3 - mobilenet
  Vulkan GPU-v3-v3-v3 - regnety_400m
  Vulkan GPU-v3-v3-v3 - FastestDet
  Vulkan GPU-v3-v3-v3 - vision_transformer
  Vulkan GPU-v3-v3-v3 - squeezenet_ssd
  Vulkan GPU-v3-v3-v3 - yolov4-tiny
  Vulkan GPU-v3-v3-v3 - resnet50
  Vulkan GPU-v3-v3-v3 - alexnet
  Vulkan GPU-v3-v3-v3 - resnet18
  Vulkan GPU-v3-v3-v3 - vgg16
  Vulkan GPU-v3-v3-v3 - googlenet
  Vulkan GPU-v3-v3-v3 - blazeface
  Vulkan GPU-v3-v3-v3 - efficientnet-b0
  Vulkan GPU-v3-v3-v3 - mnasnet
  Vulkan GPU-v3-v3-v3 - shufflenet-v2
  Vulkan GPU-v3-v3-v3-v2-v2 - mobilenet-v2
  Vulkan GPU-v3-v3-v3 - mobilenet
VkFFT
NCNN:
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - FastestDet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vision_transformer
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - regnety_400m
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - squeezenet_ssd
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - yolov4-tiny
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet50
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - alexnet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - resnet18
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - vgg16
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - googlenet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - blazeface
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - efficientnet-b0
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mnasnet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - shufflenet-v2
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2
  CPU-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet
  CPU-v3-v3-v3-v3-v3-v3-v3-v3 - mobilenet-v3
VkFFT
NCNN:
  CPU-v3-v3-v3-v3-v3-v3 - FastestDet
  CPU-v3-v3-v3-v3-v3-v3 - vision_transformer
  CPU-v3-v3-v3-v3-v3-v3 - regnety_400m
  CPU-v3-v3-v3-v3-v3-v3 - squeezenet_ssd
  CPU-v3-v3-v3-v3-v3-v3 - yolov4-tiny
  CPU-v3-v3-v3-v3-v3-v3 - resnet50
  CPU-v3-v3-v3-v3-v3-v3 - alexnet
  CPU-v3-v3-v3-v3-v3-v3 - resnet18
  CPU-v3-v3-v3-v3-v3-v3 - vgg16
  CPU-v3-v3-v3-v3-v3-v3 - googlenet
  CPU-v3-v3-v3-v3-v3-v3 - blazeface
  CPU-v3-v3-v3-v3-v3-v3 - efficientnet-b0
  CPU-v3-v3-v3-v3-v3-v3 - mnasnet
  CPU-v3-v3-v3-v3-v3-v3 - shufflenet-v2
  CPU-v3-v3-v3-v3-v3-v3-v2-v2 - mobilenet-v2
  CPU-v3-v3-v3-v3-v3-v3 - mobilenet
VkFFT:
  FFT + iFFT C2C 1D batched in half precision
  FFT + iFFT C2C multidimensional in single precision
  FFT + iFFT R2C / C2R
VkResample