Krzysztof Drewniak db647f5bd8 [mlir][GPU] Initialize LLVM exactly once during GPU compiles ...

No matter how one constructs their SerializeTo* pass, we want to
ensure that the LLVM initialization code runs once and only once. This
commit adds a static once_flag to ensure that.

I've run into mysterious segfaults when calling MLIR GPU compiles from
multiple threads, and this commit is a potential fix for the issue.

Reviewed By: fmorac

Differential Revision: https://reviews.llvm.org/D155226

2023-07-14 19:10:52 +00:00

..

AllReduceLowering.cpp

[mlir] Move casting calls from methods to function calls

2023-05-12 11:21:25 +02:00

AsyncRegionRewriter.cpp

Cleanup uses of getAttrDictionary() in MLIR to use getDiscardableAttrDictionary() when possible

2023-05-15 11:35:50 -07:00

GlobalIdRewriter.cpp

[mlir][gpu] Add a pattern for transforming gpu.global_id to thread + blockId * blockDim

2023-05-25 20:24:38 +00:00

KernelOutlining.cpp

[mlir] Move casting calls from methods to function calls

2023-05-12 11:21:25 +02:00

MemoryPromotion.cpp

[mlir][NFC] Clean up builder usage around constants/non-foldable ops

2023-06-30 13:56:42 +02:00

ParallelLoopMapper.cpp

…

SerializeToBlob.cpp

Make optimize llvm common to both gpu-to-hsaco/cubin

2023-06-05 10:32:51 +05:30

SerializeToCubin.cpp

[mlir][GPU] Initialize LLVM exactly once during GPU compiles

2023-07-14 19:10:52 +00:00

SerializeToHsaco.cpp

[mlir][GPU] Initialize LLVM exactly once during GPU compiles

2023-07-14 19:10:52 +00:00

ShuffleRewriter.cpp

[mlir][gpu] Add i64 & f64 support to gpu.shuffle

2023-05-25 21:40:25 +00:00