A synthetic benchmarking tool to measure peak capabilities of vulkan devices. It only measures the peak metrics that can be achieved using vector operations and does not represent a real-world use case.
Download Windows/Linux/MacOS Executable for Intel/AMD/Nvidia GPU
https://github.com/nihui/vkpeak/releases
vkpeak.exe 0The only parameter 0 is the device id.
If you encounter a crash or error, try upgrading your GPU driver:
- Intel: https://downloadcenter.intel.com/product/80939/Graphics-Drivers
- AMD: https://www.amd.com/en/support
- NVIDIA: https://www.nvidia.com/Download/index.aspx
- Download and setup the Vulkan SDK from https://vulkan.lunarg.com/
- For Linux distributions, you can either get the essential build requirements from package manager
dnf install vulkan-headers vulkan-loader-develapt-get install libvulkan-devpacman -S vulkan-headers vulkan-icd-loader- Clone this project with all submodules
git clone https://github.com/nihui/vkpeak.git
cd vkpeak
git submodule update --init --recursive- Build with CMake
- You can pass -DUSE_STATIC_MOLTENVK=ON option to avoid linking the vulkan loader library on MacOS
mkdir build
cd build
cmake ..
cmake --build . -j 4[nihui@nihui-pc build]$ ./vkpeak 0
device = GeForce RTX 2070
fp32-scalar = 8536.18 GFLOPS
fp32-vec4 = 8473.82 GFLOPS
fp16-scalar = 8405.30 GFLOPS
fp16-vec4 = 16261.30 GFLOPS
fp64-scalar = 262.86 GFLOPS
fp64-vec4 = 262.86 GFLOPS
int32-scalar = 8363.63 GIOPS
int32-vec4 = 8313.07 GIOPS
int16-scalar = 5518.05 GIOPS
int16-vec4 = 7138.91 GIOPS
nihui@nihui-macbook-air vkpeak-20210424-macos % ./vkpeak 0
device = Apple M1
fp32-scalar = 2093.55 GFLOPS
fp32-vec4 = 2369.02 GFLOPS
fp16-scalar = 2195.79 GFLOPS
fp16-vec4 = 2513.04 GFLOPS
fp64-scalar = 0.00 GFLOPS
fp64-vec4 = 0.00 GFLOPS
int32-scalar = 653.38 GIOPS
int32-vec4 = 649.56 GIOPS
int16-scalar = 653.42 GIOPS
int16-vec4 = 652.94 GIOPS
- https://github.com/Tencent/ncnn for fast neural network inference on ALL PLATFORMS