RPCS3/llvm - llvm - Gitea: Git with a cup of tea

RPCS3/llvm

Fork 0

mirror of https://github.com/RPCS3/llvm.git synced 2025-01-08 04:51:23 +00:00

Commit Graph

Author	SHA1	Message	Date
Lang Hames	3dd951e842	[X86] New and improved VZeroUpperInserter optimization. - Adds support for inserting vzerouppers before tail-calls. This is enabled implicitly by having MachineInstr::copyImplicitOps preserve regmask operands, which allows VZeroUpperInserter to see where tail-calls use vector registers. - Fixes a bug that caused the previous version of this optimization to miss some vzeroupper insertion points in loops. (Loops-with-vector-code that followed loops-without-vector-code were mistakenly overlooked by the previous version). - New algorithm never revisits instructions. Fixes <rdar://problem/16228798> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204021 91177308-0d34-0410-b5e6-96231b3b80d8	2014-03-17 01:22:54 +00:00
Eli Friedman	bd00a934c6	Enhanced vzeroupper insertion pass that avoids inserting vzeroupper where it is unnecessary through local analysis. Patch from Bruno Cardoso Lopes, with some additional changes. I'm going to wait for any review comments and perform some additional testing before turning this on by default. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143750 91177308-0d34-0410-b5e6-96231b3b80d8	2011-11-04 23:46:11 +00:00
Bruno Cardoso Lopes	3bde6fe0df	Introduce a pass to insert vzeroupper instructions to avoid AVX to SSE transition penalty. The pass is enabled through the "x86-use-vzeroupper" llc command line option. This is only the first step (very naive and conservative one) to sketch out the idea, but proper DFA is coming next to allow smarter decisions. Comments and ideas now and in further commits will be very appreciated. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@138317 91177308-0d34-0410-b5e6-96231b3b80d8	2011-08-23 01:14:17 +00:00

Author

SHA1

Message

Date

Lang Hames

3dd951e842

[X86] New and improved VZeroUpperInserter optimization.

- Adds support for inserting vzerouppers before tail-calls.
  This is enabled implicitly by having MachineInstr::copyImplicitOps preserve
  regmask operands, which allows VZeroUpperInserter to see where tail-calls use
  vector registers.

- Fixes a bug that caused the previous version of this optimization to miss some
  vzeroupper insertion points in loops. (Loops-with-vector-code that followed
  loops-without-vector-code were mistakenly overlooked by the previous version).

- New algorithm never revisits instructions.

Fixes <rdar://problem/16228798>



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204021 91177308-0d34-0410-b5e6-96231b3b80d8

2014-03-17 01:22:54 +00:00

Eli Friedman

bd00a934c6

Enhanced vzeroupper insertion pass that avoids inserting vzeroupper where it is unnecessary through local analysis. Patch from Bruno Cardoso Lopes, with some additional changes.

I'm going to wait for any review comments and perform some additional testing before turning this on by default.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@143750 91177308-0d34-0410-b5e6-96231b3b80d8

2011-11-04 23:46:11 +00:00

Bruno Cardoso Lopes

3bde6fe0df

Introduce a pass to insert vzeroupper instructions to avoid AVX to

SSE transition penalty. The pass is enabled through the "x86-use-vzeroupper"
llc command line option. This is only the first step (very naive and
conservative one) to sketch out the idea, but proper DFA is coming next
to allow smarter decisions. Comments and ideas now and in further commits
will be very appreciated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@138317 91177308-0d34-0410-b5e6-96231b3b80d8

2011-08-23 01:14:17 +00:00

3 Commits