Commits · 308571074c13ea2a0758aa085aa02f72150f891e · Roger Ferrer / llvm-epi · GitLab

Jan 26, 2020
- Mass update the MLIR license header to mention "Part of the LLVM project" · 30857107
  Mehdi Amini authored Jan 26, 2020
  30857107
Jan 21, 2020
- Create a gpu.module operation for the GPU Dialect. · 9a52ea5c
  Tres Popp authored Jan 17, 2020
  9a52ea5c
Jan 15, 2020
- Revert "[mlir] Create a gpu.module operation for the GPU Dialect." · 0133cc60
  Benjamin Kramer authored Jan 15, 2020
  0133cc60
Jan 14, 2020
- [mlir] Create a gpu.module operation for the GPU Dialect. · 4624a1e8
  Tres Popp authored Jan 14, 2020
  4624a1e8
Jan 13, 2020
- Unbreak the mlir build after 202ab273 · a2cd4fe6
  Benjamin Kramer authored Jan 13, 2020
  
  a2cd4fe6
- [mlir] Added missing GPU lowering ops. · 202ab273
  Julian Gross authored Jan 09, 2020
  202ab273
Jan 11, 2020
- [mlir] NFC: Remove Value::operator* and Value::operator-> now that Value is properly value-typed. · 2bdf33cc
  River Riddle authored Jan 11, 2020
  2bdf33cc
Dec 24, 2019
- NFC: Replace ValuePtr with Value and remove it now that Value is value-typed. · e62a6956
  River Riddle authored Dec 23, 2019
  e62a6956
- Adjust License.txt file to use the LLVM license · 56222a06
  Mehdi Amini authored Dec 23, 2019
  56222a06
Dec 23, 2019
- NFC: Introduce new ValuePtr/ValueRef typedefs to simplify the transition to... · 35807bc4
  River Riddle authored Dec 22, 2019
  35807bc4
Dec 20, 2019
- Add gpu.shuffle op. · 42d46b4e
  Christian Sigg authored Dec 20, 2019
  42d46b4e
Dec 18, 2019
- NFC: Remove unnecessary 'llvm::' prefix from uses of llvm symbols declared in `mlir` namespace. · 4562e389
  River Riddle authored Dec 18, 2019
  4562e389
- Harden the requirements to memory attribution types in gpu.func · 40ef46fb
  Alex Zinenko authored Dec 18, 2019
  40ef46fb
Dec 16, 2019
- Plug gpu.func into the GPU lowering pipelines · 6273fa0c
  Alex Zinenko authored Dec 16, 2019
  6273fa0c
Dec 14, 2019
- Fix maskAndClamp in gpu.all_reduce. · 88465576
  Christian Sigg authored Dec 12, 2019
  88465576
Dec 12, 2019
- Automated rollback of commit f68ac464 · 9b855826
  Christian Sigg authored Dec 12, 2019
  9b855826
- Switch from shfl.bfly to shfl.down. · f68ac464
  Christian Sigg authored Dec 12, 2019
  f68ac464
Dec 11, 2019
- NFC: Fix naming inconsistency: FuncOpLowering -> GPUFuncOpLowering. · c5fb4c13
  Christian Sigg authored Dec 11, 2019
  c5fb4c13
- Add a function to get lowering patterns from GPU to NVVM. · b96f86da
  Stephan Herhut authored Dec 11, 2019
  b96f86da
Dec 07, 2019
- Update the builder API to take ValueRange instead of ArrayRef<Value *> · d6ee6a03
  River Riddle authored Dec 07, 2019
  d6ee6a03
Dec 06, 2019
- LLVM::GlobalOp: take address space as builder argument · be3ed146
  Alex Zinenko authored Dec 06, 2019
  be3ed146
- Add conversions of GPU func with memory attributions to LLVM/NVVM · e216a72a
  Alex Zinenko authored Dec 06, 2019
  e216a72a
Dec 02, 2019
- Introduce Linkage attribute to the LLVM dialect · d5e627f8
  Alex Zinenko authored Dec 02, 2019
  d5e627f8
Nov 23, 2019
- NFC: Remove unnecessarily guarded tablegen includes. · b8ee5634
  River Riddle authored Nov 22, 2019
  b8ee5634
Nov 18, 2019
- Rename CLI flags -lower-gpu-ops-to-*-ops to -convert-gpu-to-* · b8dc3fd8
  Alex Zinenko authored Nov 18, 2019
  b8dc3fd8
Oct 31, 2019
- NFC: Use #ifndef in various .td files instead of #ifdef and #else · 7432234f
  Lei Zhang authored Oct 31, 2019
  7432234f
Oct 24, 2019
- Add custom lowering of ExpOp for NVVM and ROCM. · d2ce435d
  Alexander Belyaev authored Oct 24, 2019
  d2ce435d
Oct 21, 2019
- Unify GPU op definition names with other dialects. · b74af4aa
  Christian Sigg authored Oct 21, 2019
  b74af4aa
Oct 19, 2019
- Get active source lane predicate from shuffle instruction. · c3e56cd1
  Christian Sigg authored Oct 19, 2019
  c3e56cd1
Oct 18, 2019
- Fix OSS target name GPUtoNVVMTransforms -> MLIRGPUtoNVVMTransforms · 3e3ab380
  Nicolas Vasilache authored Oct 18, 2019
  3e3ab380
- Add gpu.barrier op to synchronize invocations of a local workgroup. · fe0ee32d
  Christian Sigg authored Oct 18, 2019
  fe0ee32d
Oct 16, 2019
- Support custom accumulator provided as region to gpu.all_reduce. · d2f0f847
  Christian Sigg authored Oct 16, 2019
  d2f0f847
Oct 10, 2019
- Mark GPU dialect as illegal when lowering to NVVM. · 82dc6c44
  Christian Sigg authored Oct 10, 2019
  82dc6c44
- Use llvm.func to define functions with wrapped LLVM IR function type · 5e7959a3
  Alex Zinenko authored Oct 10, 2019
  5e7959a3
Oct 09, 2019
- Change to doxygen comments. NFC. · 48f819c1
  Christian Sigg authored Oct 09, 2019
  48f819c1
Oct 07, 2019
- Support reduction of partial warps. · 7c765d97
  Christian Sigg authored Oct 07, 2019
  7c765d97
Oct 04, 2019
- Moving the GPUIndexIntrinsicOpLowering template to a common location · d064469f
  Deven Desai authored Oct 04, 2019
  d064469f
- Fix typos, NFC. · 85dcaf19
  Christian Sigg authored Oct 04, 2019
  85dcaf19
Oct 01, 2019
- Change all_reduce lowering to support 2D and 3D blocks. · 1129931a
  Christian Sigg authored Oct 01, 2019
  1129931a
Sep 30, 2019
- Switch comments from GPU dialect terms to CUDA terms (NFC). · 3d9679bd
  Christian Sigg authored Sep 30, 2019
  3d9679bd