Multi-GPU CUDA stress test http://wili.cc/blog/gpu-burn.html
To build GPU Burn:
make
To remove artifacts built by GPU Burn:
make clean
GPU Burn builds with a default Compute Capability of 5.0. To override this with a different value:
make COMPUTE=<compute capability value>
CFLAGS can be added when invoking make to add to the default list of compiler flags:
make CFLAGS=-Wall
LDFLAGS can be added when invoking make to add to the default list of linker flags:
make LDFLAGS=-lmylib
NVCCFLAGS can be added when invoking make to add to the default list of nvcc flags:
make NVCCFLAGS=-ccbin <path to host compiler>
CUDAPATH can be added to point to a non standard install or specific version of the cuda toolkit (default is /usr/local/cuda):
make CUDAPATH=/usr/local/cuda-<version>
CCPATH can be specified to point to a specific gcc (default is /usr/bin):
make CCPATH=/usr/local/bin
GPU Burn
Usage: gpu_burn [OPTIONS] [TIME]
-m 90 Gpu percent
-d Use doubles
-tc Use Tensor cores
-h Show this help message
Example:
gpu-burn -d 3600
gpu_burn -m 96% -d -tc 60
docker build -t gpu-burn .
docker run -it --rm --gpus all gpu-burn