Inquiry: Taichi failed on GPU

[Ruoyu66666]

I wrote a small program to compare Taichi and numpy, where Taichi function is supposed to run on Nvidia GPUs on TACC, the supercomputer maintained by the University of Texas System.

Here is the code (please neglect the unindented mistakes due to text paste):

taichi_computation_trial.txt

However, the above code degrades automatically to CPU. When I check GPU working status as program run, no job is running.

If I revise "arch=cuda" to "arch=gpu", the code failed. I attached the errors here:

slurm-733695.txt

Can someone with experience of using GPUs help? I appreciate your help ahead.

[Ruoyu66666]

Update: the TACC staff said he think Taichi developers may help after I showed him my code and the error file. He said the problem should not be the CUDA installation on the TACC side.

Regards,

Ruoyu

[Ruoyu66666]

I submitted an issue report. Thanks.