main

Ccandyhongfix: sync with upstream commit 523a1b2

03c21c85创建于 1月29日历史提交

文件	最后提交记录	最后更新时间
amd	[AMD] Fix buffer op mask operand removal (#7963)	9 个月前
samples	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
accelerate-matmul.mlir	[BACKEND] Don't use mmav5 with num warps < 4 (#7928)	9 个月前
accumulator-init.mlir	[TritonGPU] Enable accum-init optimization for unconditionally zero-ed accumulators (#6395)	1 年前
atomic-cas.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
automatic-warp-specialization.mlir	[TensorDesc] Cleanup ttng IR representation (#7036)	11 个月前
canonicalize.mlir	[Dialect] Layout attr cleanup and tighten invariants (#7714)	10 个月前
coalesce-async-copy.mlir	[BACKEND] Refactor shared memory layout representation (#5786)	1 年前
coalesce.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
combine-select-if.mlir	Add support for Nvidia Blackwell GPUs (#5724)	1 年前
combine.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
consan-negative.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
consan.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
dot-operands.mlir	[LAYOUTS] Implement toLinearLayout for TensorMemoryEncodingAttr (#7748)	9 个月前
fence-inserstion.mlir	[LAYOUTS] Implement toLinearLayout for TensorMemoryEncodingAttr (#7748)	9 个月前
fuse-nested-loops.mlir	[TritonGPU] Propagate num_stages from outer loop when flattening (#6233)	1 年前
global_scratch_alloc.mlir	[DIALECT] Rename triton_gpu to ttg and triton_nvidia_gpu to ttng (#5266)	1 年前
global_scratch_to_llvm.mlir	[PROTON] Intra kernel profiling (#7258)	10 个月前
hoist-tmem-alloc.mlir	[BACKEND] Fix wrong check in tmem_alloc canonicalization pattern (#7719)	10 个月前
inline.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
invalid-attributes.mlir	[AMD] Enable dot_scaled on gfx11 (#7954)	9 个月前
invalid.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
load-mma-specialization.mlir	[WS] reorder partition-loops and lower-aref (#7927)	9 个月前
loop-pipeline-async-latencies.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
loop-pipeline-blackwell.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
loop-pipeline-combine-waits.mlir	[BACKEND] combineRedundantWaitOps should not combine across loops/branches (#7593)	10 个月前
loop-pipeline-cuda.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
loop-pipeline-expand.mlir	[TritonGPU] Split MemDescSubview into MemDescIndex and MemDescSubslice (#7622)	10 个月前
loop-pipeline-hip.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
loop-pipeline-hopper-remove-wait.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
loop-pipeline-hopper.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
loop-pipeline-indirect-load.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
loop-pipeline.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
loop-schedule.mlir	[LAYOUTS] Enable generic swizzling on AMD (#7225)	11 个月前
matmul-loop-pipeline.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
matmul.mlir	[Pipeliner] Merge warp specialization and pipeliner scheduling (#6887)	1 年前
memdesc-subview-split.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
ops.mlir	[Gluon] Actually enable scf+cf+arith canonicalizers (#7775)	9 个月前
optimize-locality.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
optimize-partition-warps.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
optimize_epilogue.mlir	[DIALECT] Rename triton_gpu to ttg and triton_nvidia_gpu to ttng (#5266)	1 年前
partition-loops.mlir	[NVWS] Assign final try_wait to a partition. (#7757)	9 个月前
partition-scheduling.mlir	[NVWS] Assign final try_wait to a partition. (#7757)	9 个月前
pipeline-assign-latencies.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
pipeline-loop-nest.mlir	[TritonGPU] Split MemDescSubview into MemDescIndex and MemDescSubslice (#7622)	10 个月前
pipeline-lower-loop.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
pipeline-schedule-loop.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
prefetch.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
promote-lhs-to-tmem.mlir	[IR] Improve memdesc_index printing to make it obvious what operand is the index (#7876)	9 个月前
prox_fence_insertion.mlir	[BACKEND] Add a new pass to insert fence.proxy.async for write after read hazard (#7262)	11 个月前
reduce-data-duplication.mlir	[Backend] Improve warp-local layout conversion algo using shuffles (#7558)	10 个月前
reorder-instructions.mlir	[Dialect] Actually enable TMEM layout check and fix all the tests (#7723)	10 个月前
rewrite-partition-dependencies.mlir	[NVWS] Assign final try_wait to a partition. (#7757)	9 个月前
tf32x3-matmul.mlir	Remove -split-input-file from tests that don't use file splitting (#6257)	1 年前
verify-blocked-layout.mlir	[DIALECT] Rename triton_gpu to ttg and triton_nvidia_gpu to ttng (#5266)	1 年前