Skip to content

Pull requests: intel/sycl-tla

New pull request New
34 Open 475 Closed
34 Open 475 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Re-implement FlashAttention with new Xe atoms
#547 opened Oct 4, 2025 by petercad Loading…
1
upload 2nd version of sdpa backward
#546 opened Oct 3, 2025 by yuankuns Loading…
[CI] Fix and enable python interface tests
#545 opened Oct 1, 2025 by anupren Loading…
[Reorder] add optimized e8m0 -> f32 upconversion
#544 opened Oct 1, 2025 by petercad Loading…
Add comprehensive Coverity static analysis workflow
#543 opened Oct 1, 2025 by rishi-yadav Loading…
2
Support of FP8 Chunk Prefill kernel
#542 opened Oct 1, 2025 by adityachatter • Draft
Support nullptr value of argument ptr_C for xe_array_epilogue
#541 opened Sep 29, 2025 by sanchitintel Loading…
Use newer version of mma_atom and copy_atom in 00_bmg_gemm
#540 opened Sep 29, 2025 by anamikac-intel Loading…
29
Attention sink support
#533 opened Sep 25, 2025 by kareemshaik80 Loading…
6
Add dimension check to prevent out-of-bounds access in example 05_bmg_gemm_with_epilogue_splitk
#529 opened Sep 23, 2025 by ClarkChin08 Loading…
2
Model-specific tests for flash attention decode
#526 opened Sep 22, 2025 by rishi-yadav Loading…
[WIP] Added support for Rotary Embedding in flash_attention
#523 opened Sep 19, 2025 by pralay-das • Draft
12
Update CI
#522 opened Sep 18, 2025 by anupren • Draft
3
MoEGEMM as an extension of GroupGEMM
#520 opened Sep 18, 2025 by sanchitintel • Draft
15
Add a new tile scheduler for varlen prefill to avoid launching empty work groups
#516 opened Sep 18, 2025 by carsonwang Loading…
Also use column-major B matrix in the example 00_bmg_gemm.cpp
#510 opened Sep 13, 2025 by sanchitintel Loading…
Remove redundant code from GroupGEMM implementation
#508 opened Sep 12, 2025 by sanchitintel Loading…
Example of FP32 -> BF16 conversion in epilogue of GEMM
#506 opened Sep 12, 2025 by sanchitintel • Draft
1 task
Support FP32 -> BF16 conversion in epilogue of GroupedGEMM
#505 opened Sep 12, 2025 by sanchitintel • Draft
1
4
chunk prefill
#498 opened Sep 10, 2025 by sunjiweiswift Loading…
14
Changes to fix compile errors in hopper (SYCL) unit tests
#492 opened Sep 4, 2025 by jojivk73 • Draft
Support FP8 KV cache for prefill
#486 opened Sep 3, 2025 by Valentine233 Loading…
16
Support fp32 accumulation for bf16 gemm and grouped gemm
#482 opened Aug 27, 2025 by wuxun-zhang Loading…
27
[Draft PR, NOT review] improve mixed data type performance
#475 opened Aug 4, 2025 by taozha2 • Draft
14
[WIP] FP8 scaledMM with DeepSeek-style dequantization
#453 opened Jul 2, 2025 by sanchitintel • Draft
4 tasks
ProTip! Adding no:label will show everything without a label.