diff options
| author | Mirko BrkuĊĦanin <Mirko.Brkusanin@amd.com> | 2025-08-04 13:06:06 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-08-04 13:06:06 +0200 |
| commit | a24fae3aefd14cfff4bd911101cd0a1fb4dfdafe (patch) | |
| tree | eb38104207caa8a4abe1b6c0706c4cde83ec9964 /llvm/lib/Transforms/Utils/Local.cpp | |
| parent | 8cc4c6d78f08ca38b5051a09a71ec14ae8931cda (diff) | |
| download | llvm-a24fae3aefd14cfff4bd911101cd0a1fb4dfdafe.zip llvm-a24fae3aefd14cfff4bd911101cd0a1fb4dfdafe.tar.gz llvm-a24fae3aefd14cfff4bd911101cd0a1fb4dfdafe.tar.bz2 | |
[AMDGPU][True16][GlobalISel] Fix v2*16 build_vector patterns (#151496)
- Pattern with IMPLICIT_DEF failed to generate an entry in MatchTable and
did not report an error, just silently failed. This is fixed by casting
IMPLICIT_DEF to appropriate type. This also fixes selecting
"build_vector s16, undef" for GlobalISel with True16.
- Add pattern for "build_vector undef, s16" that will work for GlobalISel.
True16 GlobalISel has a G_TRUNC that it needs to deal with.
- Use REG_SEQUENCE for Real16 patterns instead of V_LSHLREV_B32_e64 to
generate more optimal code.
Diffstat (limited to 'llvm/lib/Transforms/Utils/Local.cpp')
0 files changed, 0 insertions, 0 deletions