diff options
| author | Benjamin Maxwell <benjamin.maxwell@arm.com> | 2025-09-05 12:15:28 +0000 |
|---|---|---|
| committer | Benjamin Maxwell <benjamin.maxwell@arm.com> | 2025-09-12 11:21:41 +0000 |
| commit | b666a28a0581e3a1f7838eca5f241d2550c05e4b (patch) | |
| tree | 41473d9785a63148661490cb95234d4d6f304258 /llvm/lib/Analysis/HashRecognize.cpp | |
| parent | 77596b78e5664fff8d272599c0420fc9b87e2c2d (diff) | |
| download | llvm-users/MacDue/sme-loops.zip llvm-users/MacDue/sme-loops.tar.gz llvm-users/MacDue/sme-loops.tar.bz2 | |
[AArch64][SME] Avoid ZA save state changes in loops in MachineSMEABIPassusers/MacDue/sme-loops
This patch uses the MachineLoopInfo to give blocks within loops a higher
weight when choosing the bundle ZA state. MachineLoopInfo does not find
loop trip counts, so this uses an arbitrary weight (default 10), which
can be configured with the `-aarch64-sme-abi-loop-edge-weight` flag.
This makes the MachineSMEABIPass pass more likely to pick a bundle state
that matches the loop's entry/exit state, which avoids state changes in
the loop (which we assume will happen more than once).
This does require some extra analysis, so this is only enabled at -O1
and above.
Change-Id: If318c809d2f7cc1fca144fbe424ba2a2ca7fb19f
Diffstat (limited to 'llvm/lib/Analysis/HashRecognize.cpp')
0 files changed, 0 insertions, 0 deletions