Adding new env to lit tests

[AMD] NFC: Replace intrinsics for ballot and readlane with ROCDL ops (t…

Adding env var for enabling 2 dots pipelining, disabled by default

Update the comment

Fix for extra barrier

lit tests updated

[TEST] force num_warps to be an int (triton-lang#6319)

Fix for colliding tmem allocation boundaries

Fix for colliding tmem allocation boundaries

Fix typing of do_not_specialize kwarg of triton.jit (triton-lang#…

Merge cleanup

[AMD] Enable packed Bf8/Fp8->Bf16 conversions for gfx950 (triton-lang…

Tweak assign latencies heuristic to include data fed into tmem_store.…

Disable mma pipelining when there is >1 dot in the loop

Fixing buffering issue on hopper

Actually adding the files...

Bringing back some deleted files to make mergeing easier after landin…

Merge branch 'main' into pawel/lower_mmav5

[TUTORIALS] Bypass tma_ws in the tutorial when the GPU doesn't supp…

Reapply "Removing keep_tmem_in_acc"

Merge branch 'main' into pawel/lower_mmav5

Fixing num stages for tma descriptor lowering

Merge branch 'main' into pawel/lower_mmav5

[LAYOUTS] Kill getWarpsPerCTA(Attribute) and prefer LinearLayout-base…

fixing lit

Fixing number of tma descriptor buffers

removing debug change in tests

Cleanup around pipeline utils

Stale env var

Removing keep_tmem_in_acc