You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have multiple GPUs on a machine and am trying to understand how to use them. In the C++ binding function execute I can see there is a line PjRtDevice *device = client->devices()[0]; which seems to always use the first available GPU. I tried to test changing 0 to 1, 2, etc., but am having the following error:
Execution of replica 0 failed: INVALID_ARGUMENT: Buffer passed to Execute() as argument 0 to replica 0 is on device gpu:3, but replica is assigned to device gpu:0.
Is there any suggestions how I should modified the code to use multiple GPUs? Thanks.
The text was updated successfully, but these errors were encountered:
I have multiple GPUs on a machine and am trying to understand how to use them. In the C++ binding function
execute
I can see there is a linePjRtDevice *device = client->devices()[0];
which seems to always use the first available GPU. I tried to test changing 0 to 1, 2, etc., but am having the following error:Is there any suggestions how I should modified the code to use multiple GPUs? Thanks.
The text was updated successfully, but these errors were encountered: