Commit Graph

6 Commits

Author SHA1 Message Date
Roman Lebedev 0ac56e8eaa
[InstCombine] Fold `(X >>? C1) << C2` patterns to shift+bitmask (PR37872)
This is essentially finalizes a revert of rL155136,
because nowadays the situation has improved, SCEV can model
all these patterns well, and we canonicalize rotate-like patterns
into a funnel shift intrinsics in InstCombine.
So this should not cause any pessimization.

I've verified the canonicalize-{a,l}shr-shl-to-masking.ll transforms
with alive, which confirms that we can freely preserve exact-ness,
and no-wrap flags.

Profs:
* base: https://rise4fun.com/Alive/gPQ
* exact-ness preservation: https://rise4fun.com/Alive/izi
* nuw preservation: https://rise4fun.com/Alive/DmD
* nsw preservation: https://rise4fun.com/Alive/SLN6N
* nuw nsw preservation: https://rise4fun.com/Alive/Qp7

Refs. https://reviews.llvm.org/D46760
2020-10-27 14:42:53 +03:00
Arthur Eubanks 8df17b4dc1 [test][InstCombine][NewPM] Fix InstCombine tests under NPM
Some of these depended on analyses being present that aren't provided
automatically in NPM.

early_dce_clobbers_callgraph.ll was previously inlining a noinline function?

cast-call-combine.ll relied on the legacy always-inline pass being a
CGSCC pass and getting rerun.

Reviewed By: asbirlea

Differential Revision: https://reviews.llvm.org/D88187
2020-10-06 07:39:00 -07:00
Eric Christopher cee313d288 Revert "Temporarily Revert "Add basic loop fusion pass.""
The reversion apparently deleted the test/Transforms directory.

Will be re-reverting again.

llvm-svn: 358552
2019-04-17 04:52:47 +00:00
Eric Christopher a863435128 Temporarily Revert "Add basic loop fusion pass."
As it's causing some bot failures (and per request from kbarton).

This reverts commit r358543/ab70da07286e618016e78247e4a24fcb84077fda.

llvm-svn: 358546
2019-04-17 02:12:23 +00:00
Daniel Neilson 367c2aea4e [InstCombine] Properly change GEP type when reassociating loop invariant GEP chains
Summary:
This is a fix to PR37005.

Essentially, rL328539 ([InstCombine] reassociate loop invariant GEP chains to enable LICM) contains a bug
whereby it will convert:
%src = getelementptr inbounds i8, i8* %base, <2 x i64> %val
%res = getelementptr inbounds i8, <2 x i8*> %src, i64 %val2
into:
%src = getelementptr inbounds i8, i8* %base, i64 %val2
%res = getelementptr inbounds i8, <2 x i8*> %src, <2 x i64> %val

By swapping the index operands if the GEPs are in a loop, and %val is loop variant while %val2
is loop invariant.

This fix recreates new GEP instructions if the index operand swap would result in the type
of %src changing from vector to scalar, or vice versa.

Reviewers: sebpop, spatel

Reviewed By: sebpop

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D45287

llvm-svn: 329331
2018-04-05 18:51:45 +00:00
Sebastian Pop d870aea03e [InstCombine] reassociate loop invariant GEP chains to enable LICM
This change brings performance of zlib up by 10%. The example below is from a
hot loop in longest_match() from zlib.

do.body:
  %cur_match.addr.0 = phi i32 [ %cur_match, %entry ], [ %2, %do.cond ]
  %idx.ext = zext i32 %cur_match.addr.0 to i64
  %add.ptr = getelementptr inbounds i8, i8* %win, i64 %idx.ext
  %add.ptr2 = getelementptr inbounds i8, i8* %add.ptr, i64 %idx.ext1
  %add.ptr3 = getelementptr inbounds i8, i8* %add.ptr2, i64 -1

In this example %idx.ext1 is a loop invariant. It will be moved above the use of
loop induction variable %idx.ext such that it can be hoisted out of the loop by
LICM. The operands that have dependences carried by the loop will be sinked down
in the GEP chain. This patch will produce the following output:

do.body:
  %cur_match.addr.0 = phi i32 [ %cur_match, %entry ], [ %2, %do.cond ]
  %idx.ext = zext i32 %cur_match.addr.0 to i64
  %add.ptr = getelementptr inbounds i8, i8* %win, i64 %idx.ext1
  %add.ptr2 = getelementptr inbounds i8, i8* %add.ptr, i64 -1
  %add.ptr3 = getelementptr inbounds i8, i8* %add.ptr2, i64 %idx.ext

llvm-svn: 328539
2018-03-26 16:19:31 +00:00