R and the GPU
and to install a module which requires the GPU, e.g. gmatrix:
Code Block (text)
module load R
module load cuda
module load gnu
mkdir RMOD
cd RMOD
git clone https://github.com/njm18/gmatrix.git
rm ./gmatrix/.git -rf
R CMD INSTALL gmatrix --configure-args="--with-arch=sm_35 --with-cuda-home=/local/gensoft2/exe/cuda/8.0.0" --no-test-load --no-clean-on-error
last commands are there to skip the testing, so that we can compile on a machine without CUDA driver. Note that you do not need the internet access except for the "git clone" part
sm, streaming microprocessors, depends on the target architecture:
Tesla Data Center Products#
| GPU | Compute Capability |
|---|---|
| Tesla T4 | 7.5 |
| Tesla V100 | 7.0 |
| Tesla P100 | 6.0 |
| Tesla P40 | 6.1 |
| Tesla P4 | 6.1 |
| Tesla M60 | 5.2 |
| Tesla M40 | 5.2 |
| Tesla K80 | 3.7 |
| Tesla K40 | 3.5 |
| Tesla K20 | 3.5 |
| Tesla K10 | 3.0 |