CUDA: fix crash on large batch size for quant. MoE #13537
Merged
Conversation
This file contains hidden or 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
github-actions
bot
added
Nvidia GPU
slaren
reviewed
May 14, 2025
| @@ -56,13 +56,13 @@ static __global__ void quantize_mmq_q8_1( | |||
| constexpr int vals_per_scale = ds_layout == MMQ_Q8_1_DS_LAYOUT_D2S6 ? 64 : 32; | |||
| constexpr int vals_per_sum = ds_layout == MMQ_Q8_1_DS_LAYOUT_D2S6 ? 16 : 32; | |||
|
|
|||
| const int64_t i0 = ((int64_t)blockDim.x*blockIdx.x + threadIdx.x)*4; | |||
| const int64_t i0 = ((int64_t)blockDim.x*blockIdx.y + threadIdx.x)*4; | |||
There was a problem hiding this comment.
blockDim refers to the maximum extents of threadIdx. The configuration of threads was not changed, therefore blockDim.x is still correct.
slaren
approved these changes
May 14, 2025
Yeah, this fixed it for me - thanks! |
Silver267
pushed a commit
to Silver267/llama.cpp
that referenced
this pull request
May 14, 2025
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Outdated suggestions cannot be applied.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.
Should fix issue described in #13435 (comment) .
This PR swaps the x and y dimensions of the CUDA grid for quantizing the activations since the x dimension has a higher maximum size.