CUDA: fix scratch buffers being allocated on non-main device #3220
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This was referenced Sep 17, 2023
slaren
reviewed
Sep 17, 2023
ggml-cuda.cu
Outdated
Comment on lines
6972
to
6974
|
|
||
| ggml_cuda_set_device(g_main_device); | ||
|
|
There was a problem hiding this comment.
Could this be moved inside the if block below? It seems to be the only CUDA op relevant here.
JohannesGaessler
force-pushed
the
cuda-fix-malloc-wrong-device
branch
from
70ffed6 to
391dab7
Compare
slaren
approved these changes
Sep 17, 2023
pkrmf
pushed a commit
to morlockstudios-com/llama.cpp
that referenced
this pull request
Sep 26, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #3163 .
The issue from what I can tell is that
ggml_cuda_assign_scratch_offsetdoes not set the device to the main device before allocating memory. As a consequence it is possible that the scratch buffers end up on other devices which then later causes an error. This PR simply adds a call toggml_cuda_set_devicebefore any of the other CUDA calls.