Repository navigation
[SYCL] support sparse FA - #28796
Merged
Merged
[SYCL] support sparse FA#28796
Conversation
github-actions
Bot
added
documentation
Titaniumtown
requested changes
Sep 18, 2026
| GGML_LOG_INFO(" GGML_SYCL_SUPPORT_VMM: no\n"); | ||
| #endif | ||
|
|
||
| //Print the running environment variables for SYCL backend |
CISC
reviewed
Sep 18, 2026
| static const bool sparse_fa = getenv("GGML_SYCL_SPARSE_FA") && atoi(getenv("GGML_SYCL_SPARSE_FA")) != 0; | ||
|
|
||
| ggml_tensor * cur = build_attn_mha(q, k, v, nullptr, kq_mask_top_k, nullptr, nullptr, | ||
| sparse_fa ? top_k->ne[0] : 0, kq_scale, il); |
arthw
force-pushed
the
support_sparse_fa
branch
from
September 22, 2026 06:11
1662b8c to
992747a
Compare
Contributor
Author
|
Could you help verify this PR? Thank you! |
|
Just tested the current version of this PR on top of current master and seems to work well. Until now I have been using the original version of the patch daily with no issues at all. |
Contributor
Author
|
OK, it's great! Thank you for your feedback and support! |
arthw
added
the
merge ready
feal87
added a commit
to feal87/myllama.cpp
that referenced
this pull request
Sep 25, 2026
Merge upstream commits: - llama: add llama_prec_policy + model-driven W4A4 path (ggml-org#24364) - llama: fix tensor split for fused qkv with uneven K/V head sizes (ggml-org#29294) - metal: split fa kernels into per-dtype libraries (ggml-org#29329) - metal: FWHT kernels for block widths above 512 (ggml-org#29095) - CUDA: fuse RMS_NORM + SCALE into one kernel (ggml-org#29393) - common: extract shared unicode path/string helpers (ggml-org#29415) - common,rpc: simplify fs_create_directory_with_parents() (ggml-org#29432) - rpc: include nb in the get_alloc_size cache key (ggml-org#29283) - [SYCL] support sparse FA (ggml-org#28796) - musa: fix PH1 operator failures and build issues (ggml-org#29193) - HIP: bump HIP_VERSION required for fp8 (ggml-org#29231) - opencl: add q5_k bin kernel (ggml-org#29401) - hexagon: add q5_k quant type support (ggml-org#29123) - hexagon: use DMA for contiguous dim1 CONCAT (ggml-org#29404) - mtmd: fix mel preprocessor in LFM2 audio (ggml-org#29403) - vulkan: fix legacy GLSLC without cooperativeMatrix (ggml-org#29409) - gguf-py: ByteLevel processing defaults bos/eos to False (ggml-org#29422) - gguf-py: TemplateProcessing has final word on add_special_token (ggml-org#29417) Assisted-by: Pi
sky-mighty
pushed a commit
to sky-mighty/llama.cpp
that referenced
this pull request
Sep 26, 2026
* fix conflict * fix format issue * rm unused code
1 task
frostyautumnleaf
pushed a commit
to frostyautumnleaf/llama.cpp
that referenced
this pull request
Oct 5, 2026
* fix conflict * fix format issue * rm unused code
edwardyoon
pushed a commit
to edwardyoon/focus-llama
that referenced
this pull request
Oct 7, 2026
* fix conflict * fix format issue * rm unused code (cherry picked from commit cd74ef6)
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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.
Overview
This PR is used to fix the requirement in #28695.
Additional information
It's provided by logari81.
Support long-context decode for qwen4exp on Intel Arc B70.
It's controled by env vars:
GGML_SYCL_SPARSE_FA
GGML_SYCL_SPARSE_FA_DEBUG
GGML_SYCL_SPARSE_FA_MARGIN=256
It's disabled as default.
You need to
export GGML_SYCL_SPARSE_FA=1to enable .Thank logari81 to contribute the patch!
Requirements