Merge OpenAI Triton commit `6294db5`#5583

Closed

anmyachev wants to merge 15 commits intomainfrom

amyachev/merge111

Commits on Nov 28, 2025

Commits on Nov 30, 2025

more fixes && new 'rank' function for 'IntelDPASLayout'
anmyachev
committed

Commits on Dec 1, 2025

[BACKEND] run remove backward prop until a fix point (#8776)

ThomasRaoux
authored and
anmyachev
committed
[AMD] Preserve Denorms for precise sqrt (#8697)

mengfei-jiang
authored and
anmyachev
committed
[AMD] Fix `lowerLoops`: only erase load ops which are converted (#8737)

leonling-ll
authored and
anmyachev
committed
[consan] Handle all tmem allocations (#8787)

ThomasRaoux
authored and
anmyachev
committed
[AMD] Enabling Buffer Atomic for RDNA4 (#8778)

saeid-rostami
authored and
anmyachev
committed
[PROTON][AMD] Fix failing proton tests for AMD GPUs (#8763)

authored and
anmyachev
committed
[AMD] Make kWidth to mandatory for WMMA v3 (#8783)

borontion
authored and
anmyachev
committed
[AMD] Allow async load global to load block dimension duplication (#8788)

AlexAUT
authored and
anmyachev
committed
[Reland] Fix handling of unvisited operands in AxisInfoAnalysis (#8758)

neildhar
authored and
anmyachev
committed
[KERNELS] fix persistent matmul heuristics (#8791)

aeng-openai
authored and
anmyachev
committed
Revert "[Reland] Fix handling of unvisited operands in AxisInfoAnalysis (#8758)"
anmyachev
committed