Skip to content

Pull requests: iree-org/iree

New pull request New
155 Open 15,485 Closed
155 Open 15,485 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
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fixing execution region result placement. compiler/dialects Relating to the IREE compiler dialects (flow, hal, vm)
#19872 opened Jan 31, 2025 by benvanik Loading…
Don't hoist bit-extend op as a leaf
#19871 opened Jan 31, 2025 by IanWood1 Loading…
Add links to new iree-turbine documentation website. documentation ✏️ Improvements or additions to documentation integrations/pytorch PyTorch integration work
#19870 opened Jan 31, 2025 by ScottTodd Loading…
fix(iree-opt): add missing passes and dialects for CUDA
#19867 opened Jan 31, 2025 by chrsmcgrr Loading…
[Codegen] Add initial pad encoding layout attrs
#19865 opened Jan 31, 2025 by kuhar Loading…
Bump llvm to llvm/llvm-project@5a8fe9e9a777
#19863 opened Jan 31, 2025 by pashu123 Loading…
[Dispatch] Extend gather fusion pattern
#19862 opened Jan 30, 2025 by IanWood1 Loading…
[Dispatch] Fold attention with consumer transpose
#19861 opened Jan 30, 2025 by IanWood1 Loading…
[Codegen][LLVMGPU] Swap MFMA operands for better store performance
#19860 opened Jan 30, 2025 by qedawkins • Draft
1
[LLVMGPU] Prefer larger virtual intrinsics for TileAndFuse
#19858 opened Jan 30, 2025 by qedawkins • Draft
[GPU] Match Tile And Fuse skinny matmul bail-out to Vector Distribute
#19857 opened Jan 30, 2025 by nirvedhmeshram Loading…
6
[DispatchCreation] Allow horizontal fusion when all contractions feed to the same trunc operator.
#19856 opened Jan 30, 2025 by MaheshRavishankar • Draft
[LLVMGPU] Enable TileAndFuse for matmul by default
#19854 opened Jan 30, 2025 by qedawkins • Draft
[LLVMGPU][DT] Add MaterializeDeviceEncodingPass to LLVMGPU passes behind flag
#19849 opened Jan 29, 2025 by Max191 Loading…
2
[DispatchCreation] Make truncate operations fuse with producers.
#19847 opened Jan 29, 2025 by MaheshRavishankar • Draft
9
[CPU] Remove casting math.powf from fp16 to fp32
#19844 opened Jan 29, 2025 by ita9naiwa Loading…
1
[NFC][Im2Col] Move im2tocall to affine.delinearize_index
#19840 opened Jan 28, 2025 by krzysz00 Loading…
[Codegen] Use affine.delinearize_index in workgroup distribution
#19839 opened Jan 28, 2025 by krzysz00 Loading…
Only consider linalg ops for attention mask
#19838 opened Jan 28, 2025 by IanWood1 • Draft
[LLVMGPUVectorDistribute] Add support for distributing masked reads/writes
#19830 opened Jan 28, 2025 by manupak Loading…
Only fuse gather-like producer for query operand of attention ops.
#19829 opened Jan 28, 2025 by MaheshRavishankar • Draft
1
Add bf16 from from numpy.
#19825 opened Jan 27, 2025 by AWoloszyn Loading…
1
[LLVMCPU] Update tilerootfuseconsumerproduer pass to tile softmax
#19821 opened Jan 27, 2025 by pashu123 Loading…
2
[linalgExt::Attention] add AttentionOP boolean useExp2 to choose either exp or exp2
#19820 opened Jan 27, 2025 by ita9naiwa Loading…
1
Disable Attention V operand transposition.
#19810 opened Jan 24, 2025 by MaheshRavishankar Loading…
ProTip! Updated in the last three days: updated:>2025-01-29.