I've tested SMPLify-x on two images on CPU and on GPU. I was surprised that the CPU run was much faster. From the time statistics it seems that the CPU run uses multiple cores whereas GPU run not.
CPU
$ time python -m smplifyx \
--config cfg_files/fit_smplx.yaml \
--model_folder models \
--vposer_ckpt vposer \
--data_folder /data/input \
--output_folder /data/output \
--use_cuda=False
...
Processing the data took: 00 hours, 00 minutes, 54 seconds
real 0m56.009s
user 3m24.647s
sys 0m4.912s
GPU
$ time python -m smplifyx \
--config cfg_files/fit_smplx.yaml \
--model_folder models \
--vposer_ckpt vposer \
--data_folder /data/input \
--output_folder /data/output \
--use_cuda=True
...
Processing the data took: 00 hours, 01 minutes, 24 seconds
real 1m26.369s
user 1m22.374s
sys 0m7.757s
Hi,
I've tested SMPLify-x on two images on CPU and on GPU. I was surprised that the CPU run was much faster. From the
time
statistics it seems that the CPU run uses multiple cores whereas GPU run not.CPU
GPU