Skip to content

Pull requests: openxla/xla

New pull request New
1,823 Open 15,726 Closed
1,823 Open 15,726 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

Handle gather/scatter batching dims in Get(Gather/Scatter)SizeInChunkRatio for cost analysis.
#17955 opened Oct 5, 2024 by copybara-service bot Loading…
@tomnatan30
[XLA] Refactor LayoutMode / MemorySpaceColor handling to avoid code duplication.
#17954 opened Oct 5, 2024 by copybara-service bot Loading…
@mooskagh
[XLA:GPU] Support dot_bf16_bf16_f32 algorithm with cuBLAS by adding convert before the dot call.
#17953 opened Oct 5, 2024 by copybara-service bot Loading…
@loislo
Rollback of PR #76831 because it was related to a build error due to openmp symbol missing.
#17952 opened Oct 5, 2024 by copybara-service bot Loading…
[XLA:GPU] Support 0D Tensors in the Generic Triton Emitter.
#17951 opened Oct 5, 2024 by copybara-service bot Loading…
@dimitar-asenov
Remove Post Layout Assignment Collective Pipeliner
#17950 opened Oct 5, 2024 by philipphack Loading…
Remove stream_executor/platform/platform.h in favor of simply using tsl/platform/platform.h.
#17949 opened Oct 4, 2024 by copybara-service bot Loading…
@klucke
Skip nextafter test with IFRT for unsupported dtype.
#17948 opened Oct 4, 2024 by copybara-service bot Loading…
@reedwm
Reverts a2f9ed71b6a5f4bc1109682c12a9286c83d1b5f9
#17947 opened Oct 4, 2024 by copybara-service bot Loading…
@klucke
[XLA:GPU] Split triton_fusion_emitter into three files.
#17946 opened Oct 4, 2024 by copybara-service bot Loading…
@dimitar-asenov
PR #17295: [ROCm] clang support
#17945 opened Oct 4, 2024 by copybara-service bot Loading…
Revert [XLA:GPU] Use MaterializeOp for side outputs in transpose fusion emitter
#17943 opened Oct 4, 2024 by copybara-service bot Loading…
@thomasjoerg
[IFRT] Extend assembly/disassembly operations to distinguish between addressable-shard and all-shard processing
#17942 opened Oct 4, 2024 by copybara-service bot Loading…
[XLA] Delete unused code intended to handle auto layout in HLO.
#17940 opened Oct 4, 2024 by copybara-service bot Loading…
@mooskagh
Dynamic slice fusion causes failure with maxtext
#17938 opened Oct 4, 2024 by shraiysh Loading…
[XLA:GPU] Allow kSend and kRecv to have no users if they are the root of a computation.
#17936 opened Oct 4, 2024 by copybara-service bot Loading…
FunctionArgTypeConverterFn under BufferizationOptions takes FunctionOpInterface instead of func::FuncOp
#17935 opened Oct 4, 2024 by tzunghanjuang Loading…
1
Integrate LLVM at llvm/llvm-project@82f5acfbec65
#17934 opened Oct 4, 2024 by copybara-service bot Loading…
Internal change only
#17933 opened Oct 4, 2024 by copybara-service bot Loading…
@beckerhe
Integrate LLVM at llvm/llvm-project@82f5acfbec65
#17932 opened Oct 4, 2024 by copybara-service bot Loading…
[XLA:GPU] Minor cleanup in select folder
#17931 opened Oct 4, 2024 by copybara-service bot Loading…
@frgossen
Integrate LLVM at llvm/llvm-project@82f5acfbec65
#17930 opened Oct 4, 2024 by copybara-service bot Loading…
PR #76831: aarch64: set compute library openmp runtime to off by default
#17929 opened Oct 4, 2024 by copybara-service bot Loading…
1
[XLA:GPU] Disable xla_gpu_enable_all_gather_combine_by_dim by default.
#17928 opened Oct 4, 2024 by copybara-service bot Loading…
@golechwierowicz
Allow vector size 8 in VectorizeLoadsAndStoresPass.
#17927 opened Oct 4, 2024 by copybara-service bot Loading…
@akuegel
ProTip! no:milestone will show everything without a milestone.