- User Since
- Feb 7 2020, 5:34 AM (61 w, 5 d)
Addressing review comments:
- rename StaticOperandDims to OperandStaticDims
- spell out auto
- use llvm::append instead of a loop
Use affine apply to perform index computations.
Note that the tile-indexed.mlir test is the counterpart to the tile-indexed-generic.mlir test.
Mon, Apr 12
The new test cases in loops.mlir duplicate the existing tests for the indexed_generic operation (below). The code after lowering is the same the only difference is the input code (linalg.generic + linalg.index instead of linalg.indexed_generic).
Confine the dim attribute to values larger than zero.
Thu, Apr 8
- updated to reflect ongoing changes
Wed, Apr 7
Addressing the following comments:
- added a simple end-to-end integration test
- updated the rational to reflect the index type support
- introduced a method to get the closest module or data layout op
Tue, Apr 6
Mon, Mar 29
Wed, Mar 24
Tue, Mar 23
Mar 8 2021
Mar 1 2021
Feb 22 2021
Feb 14 2021
- addressed Kayjukh's comments
Feb 12 2021
- address comments aartbik's comments
- return error by value
Feb 9 2021
Added missing algorithms header to fix Windows build.
@nicolasvasilache the updated revision should address your comments. The linter may not be happy with the function naming (_mlir_ciface) and I somewhat prefer the older function names.
Avoid std::cout and return number of errors instead of printing success or failure. Restructured the unit tests to limit the scope of the verification code. Also tried to address the lint issues. Howerever the _mlir_ciface_ perfix is fix as far as I understand.
Jan 21 2021
Jan 20 2021
Nov 20 2020
Oct 11 2020
update to latest master
Oct 9 2020
- exit early if the pattern does not apply
- update terminators of then and else bodies in place using the setOperands method
Oct 8 2020
Oct 7 2020
Jul 29 2020
Jun 29 2020
Jun 27 2020
- revert to using static instead of thread_local
- removed more inconsistent spaces
Jun 26 2020
Adressed Alex's comments:
- make default configuration thread_local
- fixed clang tidy issues
- removed unnecessary spaces
Jun 25 2020
- Added a comment that reflects the discussions about clone (can be reverted) vs inline (performance).
- The current implementation is revertible at the cost of cloning the loop body.
- Use std::tie to improve the code readability.