Commits · e64c76672f5c50d28298222047b3678ee5b2f251 · Roger Ferrer / llvm-epi · GitLab

Nov 10, 2021
- [mlir] recursively convert builtin types to LLVM when possible · e64c7667
  Alex Zinenko authored Oct 27, 2021
  e64c7667
Nov 09, 2021
- [mlir] Remove the non-templated DenseElementsAttr::getSplatValue · 937e40a8
  River Riddle authored Nov 09, 2021
  937e40a8
- [mlir] Refactor ElementsAttr's value access API · ae40d625
  River Riddle authored Nov 09, 2021
  ae40d625
Nov 02, 2021
- [mlir][gpuTonvvm] Remove hardcoded values in MMAType to llvm struct · d88cc079
  thomasraoux authored Nov 02, 2021
  d88cc079
- [mlir][VectorToGPU] Add support for elementwise mma to vector to GPU · 7fbb0678
  thomasraoux authored Nov 02, 2021
  7fbb0678
Nov 01, 2021
- [mlir][gpu] Add basic support to do elementwise ops on mma matrix type · 8a992b20
  thomasraoux authored Nov 01, 2021
  8a992b20
- [mlir][nvvm] Generalize wmma ops to handle more types and shapes · 77eafb84
  thomasraoux authored Oct 27, 2021
  77eafb84
- [mlir][SPIRVToLLVM] Add shufflevector conversion · 3483fc5a
  Weiwei Li authored Nov 01, 2021
  3483fc5a
Oct 30, 2021
- [mlir][spirv] Add conversions from arith.bitcast, std.br, std.cond_br to spirv. · 6e2c0e69
  xndcn authored Oct 29, 2021
  6e2c0e69
Oct 28, 2021
- [mlir][Linalg] NFC - Make more option names consistent. · cd392c0e
  Nicolas Vasilache authored Oct 27, 2021
  cd392c0e
- [mlir][GPUtoNVVM] Relax restriction on wmma op lowering · eacd6e1e
  thomasraoux authored Oct 25, 2021
  eacd6e1e
Oct 27, 2021
- [mlir][Linalg] NFC - Add additional control to lower vector.shape_cast ops · 8e4c806e
  Nicolas Vasilache authored Oct 26, 2021
  8e4c806e
- [mlir:DialectConversion] Restructure how argument/target materializations get invoked · 015192c6
  River Riddle authored Oct 27, 2021
  015192c6
Oct 25, 2021
- [mlir][tosa] Correct tosa.avg_pool2d for specification error · 58901a5a
  Robert Suderman authored Oct 25, 2021
  58901a5a
- [MLIR][Math] Add erf to math dialect · f1b92218
  Boian Petkantchin authored Oct 25, 2021
  f1b92218
- [mlir][Vector] NFC - Add option to hook vector.transpose lowering to strategies. · d054b80b
  Nicolas Vasilache authored Oct 25, 2021
  d054b80b
- [mlir] Switch arith, llvm, std & shape dialects to accessors prefixed both form. · cfb72fd3
  Jacques Pienaar authored Oct 24, 2021
  cfb72fd3
Oct 22, 2021
- [mlir][Vector] NFC - Extract rewrites related to insert/extract strided slice in a separate file. · eda2ebd7
  Nicolas Vasilache authored Oct 22, 2021
  eda2ebd7
Oct 19, 2021
- [mlir][RFC] Refactor layout representation in MemRefType · e41ebbec
  Vladislav Vinogradov authored Oct 11, 2021
  e41ebbec
Oct 16, 2021
- [MLIR][TOSA] Drop "OnTensors" suffix · efc6fe96
  Geoffrey Martin-Noble authored Oct 15, 2021
  efc6fe96
Oct 14, 2021
- [MLIR][arith] fix references to std.constant in comments · cb3aa49e
  Mogball authored Oct 14, 2021
  cb3aa49e
- [mlir][tosa] Fix tosa.cast UiToFp32 for tosa-to-linalg · 59dd418e
  Rob Suderman authored Oct 13, 2021
  59dd418e
Oct 13, 2021
- [MLIR] Replace std ops with arith dialect ops · a54f4eae
  Mogball authored Oct 12, 2021
  a54f4eae
- [mlir][SPIRVToLLVM] Solve ExecutionModeOp redefinition and add OpTypeSampledImage into SPV_Type · c0a6381e
  Weiwei Li authored Oct 13, 2021
  c0a6381e
Oct 12, 2021
- [mlir][tosa] Fix tosa average_pool2d to linalg type issue · 95e4b715
  Rob Suderman authored Oct 12, 2021
  95e4b715
Oct 07, 2021
- [mlir][vector] Split populateVectorContractLoweringPatterns · 3964c1db
  Lei Zhang authored Oct 07, 2021
  3964c1db
- [mlir] Add alignment attribute to memref.global · 8276ac13
  Eugene Zhulenev authored Oct 07, 2021
  8276ac13
Oct 06, 2021
- [mlir][tosa] Create basic dynamic shape support for several ops. · 4c48f7e2
  natashaknk authored Oct 06, 2021
  4c48f7e2
- [mlir][Linalg] Add support for min/max reduction vectorization in linalg.generic · eaf2588a
  Diego Caballero authored Oct 05, 2021
  eaf2588a
Oct 05, 2021
- [mlir][spirv] Add ops and patterns for lowering standard max/min ops · 7a89444c
  Lei Zhang authored Oct 05, 2021
  7a89444c
- [mlir][tosa] tosa.cast support for unsigned integers · d5a4c86d
  Rob Suderman authored Oct 05, 2021
  d5a4c86d
- [mlir] Convert ConstShapeOp to a static tensor type. · d009f6e5
  Adrian Kuegel authored Oct 05, 2021
  d009f6e5
Oct 04, 2021
- [mlir] rename the "packing" flag of linalg.pad_tensor to "nofold" · 01d696e5
  Alex Zinenko authored Oct 04, 2021
  01d696e5
- [mlir][SPIRVToLLVM] Propagate location attribute from spv.GlobalVariable to llvm.mlir.global · 1e4cfe5e
  Weiwei Li authored Oct 05, 2021
  1e4cfe5e
Oct 01, 2021
- [MLIR][GPU] Add GPU launch op support for dynamic shared memory · 08b63db8
  Uday Bondhugula authored Sep 14, 2021
  08b63db8
Sep 27, 2021
- [MLIR][LLVM] Add error if using incorrect attribute type for specifying LLVM linkage · 6dd5b1e3
  William S. Moses authored Sep 27, 2021
  6dd5b1e3
- [mlir] AsyncRuntime: use int64_t for ref counting operations · 92db09cd
  Eugene Zhulenev authored Sep 27, 2021
  92db09cd
Sep 24, 2021
- [mlir] Create a generic reduction detection utility · 2a876a71
  Diego Caballero authored Sep 22, 2021
  2a876a71
- [mlir:OpConversion] Remove the remaing usages of the deprecated matchAndRewrite methods · ef976337
  River Riddle authored Sep 24, 2021
  ef976337
- [mlir:OpConversionPattern] Add overloads for taking an Adaptor instead of ArrayRef · b54c724b
  River Riddle authored Sep 24, 2021
  b54c724b