Commits

Branch selector

 main

User selector

hanhanW

Datepicker

All time

Commit History

Commits on May 16, 2023

• [LLVMCPU] Enable shuffle_16x16 for pack codegen. (iree-org#13318)

  Show description for c1b424c
  hanhanW

  authored
  c1b424c

  Copy full SHA for c1b424c

Commits on May 12, 2023

• [LLVMCPU] Enable masking vectorization for tensor.pad ops. (iree-org#13496)

  Show description for e99db15
  hanhanW

  authored
  e99db15

  Copy full SHA for e99db15

Commits on May 9, 2023

• [LLVMCPU] Move vector level tiling out of ukernel path. (iree-org#13465)

  Show description for c744551
  hanhanW

  authored
  c744551

  Copy full SHA for c744551

Commits on May 5, 2023

• [LLVMCPU] Add tensor.pad canonicalization patterns to LLVMCPUTensorPad. (iree-org#13420)

  Show description for 8bb2d31
  hanhanW

  authored
  8bb2d31

  Copy full SHA for 8bb2d31

Commits on May 3, 2023

• [LLVMCPU] Add support for peeling tensor.pack ops. (iree-org#13354)

  hanhanW

  authored
  72b3fac

  Copy full SHA for 72b3fac

• Move the tiling logic out of DecomposePackUnPackOps pass. (iree-org#13278)

  Show description for a88bfe9
  hanhanW

  authored
  a88bfe9

  Copy full SHA for a88bfe9

• Remove stack allocation check flags from experimental benchmarks. (iree-org#13373)

  Show description for 9e33727
  hanhanW

  authored
  9e33727

  Copy full SHA for 9e33727

Commits on May 2, 2023

• [LLVMCPU] Properly pass split-transfers option to LLVMCPUVectorLowering. (iree-org#13359)

  Show description for 82b1f28
  hanhanW

  authored
  82b1f28

  Copy full SHA for 82b1f28

Commits on Apr 25, 2023

• Swap tile sizes adjustment about pack/unpack to a proper order. (iree-org#13272)

  Show description for 550b0c4
  hanhanW

  authored
  550b0c4

  Copy full SHA for 550b0c4

Commits on Apr 19, 2023

• Enable passing tests on CPU. (iree-org#13147)

  Show description for 0b92d79
  hanhanW

  authored
  0b92d79

  Copy full SHA for 0b92d79

Commits on Apr 11, 2023

• [NFC] Update comments and var naming after CodegenStrategy deprecation. (iree-org#13017)

  Show description for 8b2f0b4
  hanhanW

  authored
  8b2f0b4

  Copy full SHA for 8b2f0b4

Commits on Apr 10, 2023

• Summarize dispatch op names for SetEncoding/UnsetEncoding ops. (iree-org#12960)

  hanhanW

  authored
  e1721ed

  Copy full SHA for e1721ed

Commits on Apr 7, 2023

• [LLVMCPU] Fix logics about disabling distribution. (iree-org#12980)

  Show description for e95f6f3
  hanhanW

  authored
  e95f6f3

  Copy full SHA for e95f6f3

Commits on Apr 6, 2023

• [NFC] Clean utils up after CodegenStrategy deprecation. (iree-org#12954)

  Show description for d1ae167
  hanhanW

  authored
  d1ae167

  Copy full SHA for d1ae167

• [LLVMCPU] Add a flag to disable distribution. (iree-org#12942)

  Show description for 8d5cf96
  

  hanhanW

  and

  dcaballe

  authored
  8d5cf96

  Copy full SHA for 8d5cf96

Commits on Apr 5, 2023

• Introduce LLVMCPUVectorLowering to get rid of StrategyLowerVectors pass. (iree-org#12845)

  Show description for af4c377
  hanhanW

  authored
  af4c377

  Copy full SHA for af4c377

• [NFC] Unify the usage of debugging macro to LLVM_DEBUG for consistency. (iree-org#12939)

  hanhanW

  authored
  1054c88

  Copy full SHA for 1054c88

Commits on Apr 4, 2023

• Split VectorizePackUnPackOps into decomposition and vectorization passes (iree-org#12865)

  Show description for 820222c
  hanhanW

  authored
  820222c

  Copy full SHA for 820222c

• Migrate LLVMCPU vectorization to its own pass. (iree-org#12837)

  Show description for eeb7189
  hanhanW

  authored
  eeb7189

  Copy full SHA for eeb7189

Commits on Apr 1, 2023

• Update the tensor.pack tests to not fuse inputs and pack ops. (iree-org#12879)

  hanhanW

  authored
  4a0b9d2

  Copy full SHA for 4a0b9d2

Commits on Mar 30, 2023

• Delete upstreamed AffineBuilder. (iree-org#12863)

  Show description for 4067e74
  hanhanW

  authored
  4067e74

  Copy full SHA for 4067e74

• Delete SCFTilingPattern declaration which was missed in previous commit. (iree-org#12864)

  Show description for 3fe488b
  hanhanW

  authored
  3fe488b

  Copy full SHA for 3fe488b

• Delete LinalgExt FoldIntoPackAndUnpackOps pass. (iree-org#12860)

  Show description for 72247dd
  hanhanW

  authored
  72247dd

  Copy full SHA for 72247dd

• [LLVMCPU] Move SplitReduction from CodegenDriver to a LLVMCPU pass. (iree-org#12786)

  Show description for 0587a95
  hanhanW

  authored
  0587a95

  Copy full SHA for 0587a95

Commits on Mar 29, 2023

• [LLVMCPU] Clamp inner tiling sizes for padding approach to avoid masking (iree-org#12814)

  Show description for 14078cf
  hanhanW

  authored
  14078cf

  Copy full SHA for 14078cf

• Switch to use scf::tileConsumerAndFuseProducerGreedilyUsingSCFForOp (iree-org#12811)

  Show description for fbdf570
  hanhanW

  authored
  fbdf570

  Copy full SHA for fbdf570

• Introduce LLVMCPUTile to get rid of CodegenStrategy. (iree-org#12798)

  Show description for 3892765
  hanhanW

  authored
  3892765

  Copy full SHA for 3892765

Commits on Mar 28, 2023

• [NFC] Deprecate filter based decomposition patterns. (iree-org#12812)

  Show description for 5c0c4ea
  hanhanW

  authored
  5c0c4ea

  Copy full SHA for 5c0c4ea

• Wrap convolution decomposition patterns to a common pass. (iree-org#12799)

  Show description for 09bc7fd
  hanhanW

  authored
  09bc7fd

  Copy full SHA for 09bc7fd

Commits on Mar 27, 2023

• Only exclude e2e_matmul_mmt4d_i8_small on RISC-V. (iree-org#12764)

  Show description for d1e7ef8
  hanhanW

  authored
  d1e7ef8

  Copy full SHA for d1e7ef8

Commits on Mar 24, 2023

• [NFC] Delete dead code. (iree-org#12752)

  hanhanW

  authored
  3064b0f

  Copy full SHA for 3064b0f

• Drop support for LinalgFuse, StrategyTileAndFuse and StrategyPad (iree-org#12668)

  hanhanW

  authored
  2f6dcc4

  Copy full SHA for 2f6dcc4

Commits on Mar 23, 2023

• Move peeling from CodegenStrategy to a LLVMCPU pass. (iree-org#12707)

  hanhanW

  authored
  66cb2bd

  Copy full SHA for 66cb2bd

Commits on Mar 22, 2023

• Introduce LLVMCPUTensorPad pass (iree-org#12667)

  Show description for c6b6aba
  hanhanW

  authored
  c6b6aba

  Copy full SHA for c6b6aba

• [LLVMCPU] Move tiling on reduction to use SingleTilingExpert. (iree-org#12711)

  Show description for ac24f60
  hanhanW

  authored
  ac24f60

  Copy full SHA for ac24f60

Pagination

Previous

Next