Repository navigation

All pull requests

[Ascend] Infer SIMT UB ownership from physical access geometry
#3633happyhappyhappy-arch opened Oct 23, 2026
4
[Ascend] Fix strided stores in compound per-core tasks
#3632erhsh opened Oct 23, 2026Contributor
2
[Transform] Simplify reassociated layout div/mod expressions
#3631sepcnt opened Oct 23, 2026Contributor
2
[Ascend] Add write-counter multi-buffering and reorder DeepGEMM loads
#3629AutumnKite opened Oct 23, 2026Collaborator
2
[BugFix] Check access_ptr reads in VerifyBufferInit
#3626liuhao-labs opened Oct 23, 2026
7
[Ascend] Preserve configuration register phases in manual pipelines
#3624Denverjin opened Oct 23, 2026Contributor
2
[Ascend] Preserve buffer view layouts and pointer units
#3623Denverjin opened Oct 23, 2026Contributor
2
[Fix][Ascend] Correct Philox uniform range docs and tests
#3622erhsh opened Oct 23, 2026Contributor
2
[BugFix][ROCm] Don't lower T.exp, T.sin, and T.cos to fastmath by default on HIP
#3621jzluo opened Oct 22, 2026
6
[Transform] Locate VerifyBufferInit reads outside tile ops
#3620RyanL2 opened Oct 22, 2026Contributor
2
[Codegen] Add unsigned mulhi intrinsic
#3617sepcnt opened Oct 22, 2026Contributor
2
[Ascend] cross core sync with mode=0 operator requires setting the batch mode
#3614kevin3113 opened Oct 22, 2026
15
[Do not review][Language][CUDA][ROCm] Add native argmin and argmax reductions
#3613LeiWang1999 opened Oct 22, 2026Member
2
[Transform] Keep assumption checks with local dependencies in scope
#3611sepcnt opened Oct 22, 2026Contributor
2
[Transform] Preserve tl.assume predicates that simplify to true
#3610sepcnt opened Oct 22, 2026Contributor
2
[CPU] Support clz in C and C host code generation
#3609sepcnt opened Oct 22, 2026Contributor
2
[BugFix][Ascend] Reject odd FP4 view geometry in GM->L1 copies
#3608173787247 opened Oct 22, 2026Contributor
7
[Ascend] Support exact placement of versioned on-chip buffers
#3607Denverjin opened Oct 22, 2026Contributor
3
[CUDA] Add NaN-propagating symmetric clamp
#3605KellyFrog opened Oct 22, 2026Collaborator
4
[BugFix][CUDA] Round FP32 literals before hexadecimal formatting
#3604JayceSu98 opened Oct 22, 2026Contributor
2
[BugFix][CUDA] Check shared regions before selecting 1D TMA
#3603JayceSu98 opened Oct 22, 2026Contributor
4
[BugFix][JIT] Resolve scalar parameters in output shapes
#3602erhsh opened Oct 22, 2026Contributor
2
[BugFix][Ascend] Exclude the rounded uniform RNG endpoint
#3601erhsh opened Oct 22, 2026Contributor
2
[BugFix][Transform] Pipeline shared buffers accessed through T.reshape/T.view
#3600lance-lii opened Oct 22, 2026
2
[BugFix] Preserve signedness for mixed-signedness shifts
#3597Ethereal49 opened Oct 21, 2026Contributor
2

Pull requests