IBM Support

CUDA_VISIBLE_DEVICES_ORIG and CUDA_VISIBLE_DEVICES

Question & Answer


Question

What's difference between CUDA_VISIBLE_DEVICES_ORIG and CUDA_VISIBLE_DEVICES?

Answer

CUDA_VISIBLE_DEVICES is used to specify which GPUs should be visible to a CUDA application.
 
CUDA_VISIBLE_DEVICES_ORIG is a LSF internal environment variable.
 
Without cgroup enforcement, CUDA_VISIBLE_DEVICES and CUDA_VISIBLE_DEVICES_ORIG should be identical.

With cgroup enforcement, CUDA_VISIBLE_DEVICES_ORIG is the list of GPUs allocated by LSF. CUDA maps the IDs of GPUs in CUDA_VISIBLE_DEVICES_ORIG to the CUDA_VISIBLE_DEVICES list for job to use.

[{"Product":{"code":"SSETD4","label":"Platform LSF"},"Business Unit":{"code":"BU059","label":"IBM Software w\/o TPS"},"Component":"--","Platform":[{"code":"PF016","label":"Linux"}],"Version":"10.1","Edition":"","Line of Business":{"code":"LOB10","label":"Data and AI"}}]

Document Information

Modified date:
17 June 2018

UID

isg3T1024923