0
I Use This!
Very High Activity

Commits : Listings

Analyzed about 22 hours ago. based on code collected about 24 hours ago.
Jun 19, 2024 — Jun 19, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
[AMD] Support MI350 ds_read_b64_tr_b8 instruction for int8 (#6018) More... 4 months ago
[BE] `TCGen5MMAScaledOp` accepts scales in shared memory (#6019) More... 4 months ago
[BACKEND] Fix lowering of split op with linear layout (#6031) More... 4 months ago
[AMD] Enable OCP fp8/bf8 tests on MI350 (#6021) More... 4 months ago
[BACKEND] Allow mixed of linear layout and legacy in split/join (#6028) More... 4 months ago
[AMD] [FA] Hoist convert_layout to dotOp for Q out of the loop (#6017) More... 4 months ago
[AMD] NFC: IWYU for RangeAnalysis (#6027) More... 4 months ago
[NFC][LAYOUTS] Kill DotToFMA lowering (#6024) More... 4 months ago
[Blackwell] Support mxfp with 8 warps (#6020) More... 4 months ago
[Frontend][Cache] Include platform information in the key (#6000) More... 4 months ago
[PIPELINER] Refactor pipeliner lowering. (#5989) More... 4 months ago
[Backend] Plumb `ttg.warp_specialize` through LLVM lowering (#5963) More... 4 months ago
cache: add the triton version to the json metadata (#5912) More... 4 months ago
[LAYOUTS] [NFC] Make order accept a RankedTensorType (#6007) More... 4 months ago
typeConverter to llvm support addressSpace attribute (#5951) More... 4 months ago
[BACKEND] Fix condition to do a naive reshape with allow_reorder (#6012) More... 4 months ago
[AMD][NFC] Extract range analysis into its own class (#5977) More... 4 months ago
[AMD] Support dot_scaled(mxfp8, mxfp4) for gfx950 (#5985) More... 4 months ago
[AMD] Add debug prints for pingpong scheduler failures (#5975) More... 4 months ago
[AMD] Fix packed fp16 atomic optimization conditions (#5839) More... 4 months ago
[BACKEND] Fix crash in mmav5 lhs comes from tmem (#6011) More... 4 months ago
[TritonGPU] LICM outer loop before flattening (#6010) More... 4 months ago
[AMD][FA] Improve warp distribution for decode attention dot (#5892) More... 4 months ago
[AMD] Fix block ping-pong reordering for persistent matmul (#5986) More... 4 months ago
[LAYOUTS] Move all get.*ContigPerThread functions to a common API (#6002) More... 4 months ago
[Interface] Add dot interface methods to get A/B tensor (#5984) More... 4 months ago
[LAYOUTS] [NFC] Just accept DistributedEncodings in SliceLayout (#6004) More... 4 months ago
[NFC] Kill isBlockedToDotShortcut (#6003) More... 4 months ago
[LAYOUTS] [NFC] Kill legacy Distributed <> Distributed lowering (#6006) More... 4 months ago
[DIALECT] Take out `supportReduction` (#5997) More... 4 months ago