[SYCL] Fix DMMV dequantization #9279
Merged
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
github-actions
bot
added
the
SYCL
|
I think given the perf implications of this bounds checking, we should dig a little deeper. |
|
@joeatodd Agree |
|
Updated fix and performance is preserved now. |
airMeng
approved these changes
Sep 4, 2024
dsx1986
pushed a commit
to dsx1986/llama.cpp
that referenced
this pull request
Oct 29, 2024
Fixed dmmv dequant for ncols== GGML_SYCL_DMMV_X
arthw
pushed a commit
to arthw/llama.cpp
that referenced
this pull request
Nov 15, 2024
Fixed dmmv dequant for ncols== GGML_SYCL_DMMV_X
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Outdated suggestions cannot be applied.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.
MUL_MAT test-backend-ops currently fail on intel GPUs for Q4_1, Q5_0, Q5_1 and Q8_0 due to a small edge-case issue in the
dequantize_mul_mat_veckernel (when ncols <= GGML_SYCL_DMMV_X specifically).This is a minor fix that halts the access to out-bound/extra quant elements in the kernel reduction step.
All unit-tests are passing following this fix.
Performance on intel GPUs is almost not affected.