yamauchi (Hiroshi Yamauchi)
User

Projects

User does not belong to any projects.

User Details

User Since
Jun 28 2017, 5:32 PM (20 w, 4 d)

Recent Activity

Tue, Nov 14

yamauchi updated the diff for D39980: Add heuristics for irreducible loop metadata under PGO.

Updated comment.

Tue, Nov 14, 4:28 PM
yamauchi updated the summary of D39980: Add heuristics for irreducible loop metadata under PGO.
Tue, Nov 14, 3:18 PM
yamauchi added inline comments to D39980: Add heuristics for irreducible loop metadata under PGO.
Tue, Nov 14, 3:18 PM
yamauchi updated the diff for D39980: Add heuristics for irreducible loop metadata under PGO.

Rebased and addressed comments.

Tue, Nov 14, 3:18 PM
yamauchi committed rL318182: Simplify irreducible loop metadata test code..
Simplify irreducible loop metadata test code.
Tue, Nov 14, 11:49 AM
yamauchi closed D40043: Simplify irreducible loop metadata test code..
Tue, Nov 14, 11:49 AM
yamauchi created D40043: Simplify irreducible loop metadata test code..
Tue, Nov 14, 11:14 AM

Mon, Nov 13

yamauchi updated the summary of D39980: Add heuristics for irreducible loop metadata under PGO.
Mon, Nov 13, 2:12 PM
yamauchi created D39980: Add heuristics for irreducible loop metadata under PGO.
Mon, Nov 13, 2:01 PM

Thu, Nov 2

yamauchi committed rL317278: Irreducible loop metadata for more accurate block frequency under PGO..
Irreducible loop metadata for more accurate block frequency under PGO.
Thu, Nov 2, 3:27 PM
yamauchi closed D39028: Irreducible loop metadata for more accurate block frequency under PGO..
Thu, Nov 2, 3:27 PM
yamauchi updated the diff for D39028: Irreducible loop metadata for more accurate block frequency under PGO..

Rebased.

Thu, Nov 2, 3:26 PM

Fri, Oct 27

yamauchi added inline comments to D39028: Irreducible loop metadata for more accurate block frequency under PGO..
Fri, Oct 27, 3:31 PM
yamauchi updated the diff for D39028: Irreducible loop metadata for more accurate block frequency under PGO..

Addressed comments.

Fri, Oct 27, 3:30 PM

Wed, Oct 25

yamauchi updated the diff for D39028: Irreducible loop metadata for more accurate block frequency under PGO..

Changed the metadata name.
Added the metadata to the doc.

Wed, Oct 25, 3:51 PM

Oct 20 2017

yamauchi added inline comments to D39028: Irreducible loop metadata for more accurate block frequency under PGO..
Oct 20 2017, 6:10 PM
yamauchi updated the diff for D39028: Irreducible loop metadata for more accurate block frequency under PGO..

Addressed comments.

Oct 20 2017, 6:09 PM

Oct 17 2017

yamauchi created D39028: Irreducible loop metadata for more accurate block frequency under PGO..
Oct 17 2017, 5:42 PM

Oct 6 2017

yamauchi added a comment to D38474: Add support for multiple prof metadata on instructions..

To give more context, the motivating issue for us is the block frequency propagation of irreducible loops under PGO.

Oct 6 2017, 11:42 AM
yamauchi added a comment to D38474: Add support for multiple prof metadata on instructions..

This does not look like backward compatible IR change. For the purpose of annotating irreducible loop headers, it might be better to introduce a new type of metadata.

Oct 6 2017, 11:04 AM

Oct 2 2017

yamauchi updated the diff for D38474: Add support for multiple prof metadata on instructions..

A bit of cleanup and added an assert.

Oct 2 2017, 4:43 PM
yamauchi created D38474: Add support for multiple prof metadata on instructions..
Oct 2 2017, 11:49 AM

Sep 13 2017

yamauchi committed rL313220: Add optional profile counts to block frequency dump..
Add optional profile counts to block frequency dump.
Sep 13 2017, 5:22 PM
yamauchi closed D37835: Add optional profile counts to block frequency dump..
Sep 13 2017, 5:21 PM
yamauchi added a comment to D37835: Add optional profile counts to block frequency dump..

The last note is that, as I suspect, we don't include a header in CodeGen/ into a file under Analysis/ which prevents the overload idea from working.

Sep 13 2017, 5:14 PM
yamauchi added inline comments to D37835: Add optional profile counts to block frequency dump..
Sep 13 2017, 4:44 PM
yamauchi created D37835: Add optional profile counts to block frequency dump..
Sep 13 2017, 4:30 PM
yamauchi committed rL313159: Add options to dump PGO counts in text..
Add options to dump PGO counts in text.
Sep 13 2017, 10:22 AM
yamauchi closed D37776: Add options to dump PGO counts in text..
Sep 13 2017, 10:21 AM

Sep 12 2017

yamauchi updated the summary of D37776: Add options to dump PGO counts in text..
Sep 12 2017, 5:43 PM
yamauchi updated the diff for D37776: Add options to dump PGO counts in text..

Addressed the comment.

Sep 12 2017, 5:41 PM
yamauchi created D37776: Add options to dump PGO counts in text..
Sep 12 2017, 3:34 PM

Sep 11 2017

yamauchi committed rL312930: Unmerge GEPs to reduce register pressure on IndirectBr edges..
Unmerge GEPs to reduce register pressure on IndirectBr edges.
Sep 11 2017, 10:53 AM
yamauchi closed D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..
Sep 11 2017, 10:53 AM
yamauchi added a comment to D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

Thanks, Hal.

Sep 11 2017, 10:52 AM
yamauchi updated the diff for D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

Rebased.

Sep 11 2017, 10:52 AM

Sep 7 2017

yamauchi updated the diff for D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

Addressed a comment.

Sep 7 2017, 11:59 AM

Sep 6 2017

yamauchi added inline comments to D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..
Sep 6 2017, 3:23 PM
yamauchi updated the diff for D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

One comment addressed and another needs clarification.

Sep 6 2017, 3:22 PM

Aug 28 2017

yamauchi updated the diff for D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

Noted the tradeoff between register pressures and critical path in the comment.

Aug 28 2017, 12:44 PM
yamauchi added a comment to D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

As this patch can affect ARM targets I am doing some benchmarking.
I've got the LNT benchmarks results for AArch64 (Cortex-A57). There is no difference in performance. I'll have got more results soon.
It's interesting to see what benchmarks has been used to measure the improvements.

Aug 28 2017, 11:06 AM

Aug 25 2017

yamauchi committed rL311822: Add options to dump block frequency/branch probability info in text..
Add options to dump block frequency/branch probability info in text.
Aug 25 2017, 5:32 PM
yamauchi closed D37165: Add options to dump block frequency/branch probability info in text..
Aug 25 2017, 5:32 PM
yamauchi added a reviewer for D37165: Add options to dump block frequency/branch probability info in text.: davidxl.
Aug 25 2017, 2:59 PM
yamauchi created D37165: Add options to dump block frequency/branch probability info in text..
Aug 25 2017, 1:58 PM

Aug 24 2017

yamauchi updated the diff for D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

Added getIntImmCost checks. Please take another look.

Aug 24 2017, 5:01 PM

Aug 23 2017

yamauchi added a comment to D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..

Sorry for a delay.

Aug 23 2017, 1:41 PM

Aug 21 2017

yamauchi added a reviewer for D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges.: hfinkel.

Hi Hal, would you take a look at this change?

Aug 21 2017, 5:04 PM
yamauchi abandoned D35445: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.

Replaced by https://reviews.llvm.org/D35499

Aug 21 2017, 4:59 PM

Aug 15 2017

yamauchi created D36772: Unmerge GEPs to reduce register pressure on IndirectBr edges..
Aug 15 2017, 3:47 PM

Aug 9 2017

yamauchi committed rL310545: [LVI] Fix LVI compile time regression around constantFoldUser().
[LVI] Fix LVI compile time regression around constantFoldUser()
Aug 9 2017, 7:24 PM
yamauchi closed D36552: [LVI] Fix LVI compile time regression around constantFoldUser().
Aug 9 2017, 7:24 PM
yamauchi updated the diff for D36552: [LVI] Fix LVI compile time regression around constantFoldUser().

Done.

Aug 9 2017, 7:23 PM
yamauchi updated the diff for D36552: [LVI] Fix LVI compile time regression around constantFoldUser().

Addressed comment.

Aug 9 2017, 5:12 PM
yamauchi updated the diff for D36552: [LVI] Fix LVI compile time regression around constantFoldUser().

Addressed comments.

Aug 9 2017, 3:56 PM
yamauchi updated the diff for D36552: [LVI] Fix LVI compile time regression around constantFoldUser().

Cleaned up.

Aug 9 2017, 2:46 PM
yamauchi created D36552: [LVI] Fix LVI compile time regression around constantFoldUser().
Aug 9 2017, 2:30 PM

Aug 3 2017

yamauchi committed rL309986: [LVI] Constant-propagate a zero extension of the switch condition value through….
[LVI] Constant-propagate a zero extension of the switch condition value through…
Aug 3 2017, 2:12 PM
yamauchi closed D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.
Aug 3 2017, 2:12 PM
yamauchi added inline comments to D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.
Aug 3 2017, 1:49 PM
yamauchi updated the diff for D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

Addressed the comment.

Aug 3 2017, 1:49 PM
yamauchi added a comment to D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

Addressed the inline comments.

Aug 3 2017, 12:03 PM
yamauchi updated the diff for D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

Agreed on the phrasing and that there are cases where we could do the propagation/transformation.

Aug 3 2017, 12:03 PM

Aug 2 2017

yamauchi added a comment to D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

This is ready for review.

Aug 2 2017, 4:09 PM
yamauchi added a comment to D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

The issue with D34822 was that it is not correct to propagate the constant range of the value that's derived from the switch condition to the *default* edge.

Aug 2 2017, 4:03 PM
yamauchi updated the diff for D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

This second diff is the new one that fixes the bug (see test20).

Aug 2 2017, 4:00 PM
yamauchi added a comment to D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

The first diff is the same as the reverted https://reviews.llvm.org/D34822.

Aug 2 2017, 3:53 PM
yamauchi created D36247: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.
Aug 2 2017, 3:52 PM

Jul 28 2017

yamauchi committed rL309415: [LVI] Constant-propagate a zero extension of the switch condition value through….
[LVI] Constant-propagate a zero extension of the switch condition value through…
Jul 28 2017, 11:36 AM
yamauchi closed D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.
Jul 28 2017, 11:36 AM

Jul 27 2017

yamauchi committed rL309304: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs….
[InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs…
Jul 27 2017, 11:27 AM
yamauchi closed D35499: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.
Jul 27 2017, 11:27 AM

Jul 26 2017

yamauchi updated the diff for D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

Addressed the comments.

Jul 26 2017, 6:33 PM
yamauchi added inline comments to D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.
Jul 26 2017, 4:39 PM
yamauchi updated the diff for D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

Addressed the comments.

Jul 26 2017, 4:39 PM
yamauchi committed rL309192: Fix a comment (test commit)..
Fix a comment (test commit).
Jul 26 2017, 2:55 PM

Jul 24 2017

yamauchi updated the diff for D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.
Jul 24 2017, 4:09 PM

Jul 20 2017

yamauchi added a comment to D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

I think this situation occurs due to the combination of the following two optimizations that shrink/narrow the width of operands.

Jul 20 2017, 10:40 AM
yamauchi added a reviewer for D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges: sanjoy.
Jul 20 2017, 10:15 AM

Jul 19 2017

yamauchi added a comment to D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

For more context:

Jul 19 2017, 11:42 AM

Jul 18 2017

yamauchi added inline comments to D35499: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.
Jul 18 2017, 5:04 PM
yamauchi updated the diff for D35499: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.

Now using llvm::count_if + make_range.

Jul 18 2017, 5:01 PM
yamauchi added inline comments to D35499: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.
Jul 18 2017, 2:30 PM
yamauchi updated the diff for D35499: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.

Addressed one more comment.

Jul 18 2017, 2:29 PM
yamauchi updated the diff for D35499: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.

Addressed the comments.

Jul 18 2017, 2:17 PM

Jul 17 2017

yamauchi created D35499: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.
Jul 17 2017, 12:24 PM

Jul 14 2017

yamauchi updated subscribers of D35445: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.
Jul 14 2017, 6:56 PM
yamauchi updated the summary of D35445: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.
Jul 14 2017, 6:42 PM
yamauchi created D35445: [InstCombine] Simplify pointer difference subtractions (GEP-GEP) where GEPs have other uses and one non-constant index.
Jul 14 2017, 6:39 PM

Jul 10 2017

yamauchi added a comment to D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

Wei, can you review this?

Jul 10 2017, 10:47 AM

Jun 29 2017

yamauchi updated the diff for D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.

Added a test.

Jun 29 2017, 5:21 PM
yamauchi added reviewers for D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges: wmi, dberlin.
Jun 29 2017, 11:25 AM
yamauchi created D34822: [LVI] Constant-propagate a zero extension of the switch condition value through case edges.
Jun 29 2017, 11:12 AM