diff options
| author | Thomas Symalla <github@thomassymalla.de> | 2023-05-08 20:50:20 +0200 |
|---|---|---|
| committer | Thomas Symalla <thomas.symalla@amd.com> | 2023-05-19 12:02:21 +0200 |
| commit | 91a7aa4c9b29fa217f45badeaa87c4667a28f807 (patch) | |
| tree | 26a430888df862993ebc77d0c29cbfa652baf30a /clang/lib/CodeGen/CodeGenFunction.cpp | |
| parent | c10138a54f7f50c7a4d92b79a8887c51c52e2c20 (diff) | |
| download | llvm-91a7aa4c9b29fa217f45badeaa87c4667a28f807.zip llvm-91a7aa4c9b29fa217f45badeaa87c4667a28f807.tar.gz llvm-91a7aa4c9b29fa217f45badeaa87c4667a28f807.tar.bz2 | |
[AMDGPU] Improve abs modifier usage
If a call to the llvm.fabs intrinsic has users in another reachable
BB, SelectionDAG will not apply the abs modifier to these users and
instead generate a v_and ..., 0x7fffffff instruction.
For fneg instructions, the issue is similar.
This patch implements `AMDGPUIselLowering::shouldSinkOperands`,
which allows CodegenPrepare to call `tryToSinkFreeOperands`.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D150347
Diffstat (limited to 'clang/lib/CodeGen/CodeGenFunction.cpp')
0 files changed, 0 insertions, 0 deletions