-
Notifications
You must be signed in to change notification settings - Fork 29
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update Jax(LLVM, MLIR) and Enzyme #428
Conversation
Hello. You may have forgotten to update the changelog!
|
[sc-53151] |
…yLaneAI/catalyst into update_jax_llvm_mlir_enzyme
mlir/lib/Gradient/Transforms/GradMethods/PS_QuantumGradient.cpp
Outdated
Show resolved
Hide resolved
Blocked by EnzymeAD/Enzyme#1608 |
…yLaneAI/catalyst into update_jax_llvm_mlir_enzyme
…yLaneAI/catalyst into update_jax_llvm_mlir_enzyme
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #428 +/- ##
=======================================
Coverage 99.52% 99.52%
=======================================
Files 45 45
Lines 7941 7942 +1
Branches 537 537
=======================================
+ Hits 7903 7904 +1
Misses 20 20
Partials 18 18 ☔ View full report in Codecov by Sentry. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good
mlir/lib/Gradient/Transforms/GradMethods/PS_QuantumGradient.cpp
Outdated
Show resolved
Hide resolved
Co-authored-by: erick-xanadu <110487834+erick-xanadu@users.noreply.github.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @rmoyard for this long awaited update 💯
@@ -128,7 +128,6 @@ def run_writing_command(command: List[str], compile_options: Optional[CompileOpt | |||
"func.func(hlo-legalize-to-linalg)", | |||
"func.func(mhlo-legalize-to-std)", | |||
"convert-to-signless", | |||
"func.func(scalarize)", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@erick-xanadu I remember you found that either of two passes was providing similar speedups, one at the hlo level and one at the linalg level. Can we replace this pass with the linalg one (since it no longer exists)?
@@ -179,6 +178,7 @@ def run_writing_command(command: List[str], compile_options: Optional[CompileOpt | |||
MLIR_TO_LLVM_PASS = ( | |||
"MLIRToLLVMDialect", | |||
[ | |||
"expand-realloc", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Small comment why this pass is necessary will be nice, since it's hard to see why pipelines are the way they are.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I will get the reference I don't remember where it comes from
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
https://reviews.llvm.org/D159430 This was part of the memref lowering and it was extracted to make lowering simpler
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Right I meant adding it as a comment into the file
Co-authored-by: David Ittah <dime10@users.noreply.github.com>
**Description of the Change:** NullOp -> ZeroOp llvm/llvm-project#67183 CodeGenOpt -> CodeGenOptLevel https://github.com/llvm/llvm-project/pull/66295/files Converters are const llvm/llvm-project@ce25459 Remove typed pointer llvm/llvm-project#71285 thlo gml_st on ice tensorflow/mlir-hlo@7077cec ByteCode is a new interface llvm/llvm-project@837d1ce getArgsOperandMutable was added CallOpInterface llvm/llvm-project@d790a21 Remove redundant memref.tensor_store op https://github.com/llvm/llvm-project/pull/71010/files Lowering of memref.realloc was extracted to its own pass https://reviews.llvm.org/D159430 **Benefits:** **Possible Drawbacks:** **Related GitHub Issues:** --------- Co-authored-by: erick-xanadu <110487834+erick-xanadu@users.noreply.github.com> Co-authored-by: David Ittah <dime10@users.noreply.github.com>
Description of the Change:
NullOp -> ZeroOp
llvm/llvm-project#67183
CodeGenOpt -> CodeGenOptLevel
https://github.com/llvm/llvm-project/pull/66295/files
Converters are const
llvm/llvm-project@ce25459
Remove typed pointer
llvm/llvm-project#71285
thlo gml_st on ice
tensorflow/mlir-hlo@7077cec
ByteCode is a new interface
llvm/llvm-project@837d1ce
getArgsOperandMutable was added CallOpInterface
llvm/llvm-project@d790a21
Remove redundant memref.tensor_store op
https://github.com/llvm/llvm-project/pull/71010/files
Lowering of memref.realloc was extracted to its own pass
https://reviews.llvm.org/D159430
Benefits:
Possible Drawbacks:
Related GitHub Issues: