Page MenuHomePhabricator

cameron.mcinally (Cameron McInally)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 6 2015, 6:21 AM (302 w, 2 d)

Recent Activity

Yesterday

cameron.mcinally added a comment to D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.

Sorry @cameron.mcinally I've not had much time for code reviews this week although will take proper look tomorrow. I have a question though. You've added extra legalisation support but I don't see any explicit tests (or at least ones with matching check lines) for it. Is this something you need for this patch? (I'm guessing sve-fixed-length-fp-reduce.ll's stock NEON run line triggers the cases?) If so then there really should be a neon specific test file that verifies the widening and scalarisation changes as the NEON run line for the "fixed-length" tests is more about ensuring no SVE instructions slip through.

Thu, Oct 22, 10:11 AM · Restricted Project
cameron.mcinally updated the diff for D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.

Try again with 80 column fix...

Thu, Oct 22, 7:51 AM · Restricted Project
cameron.mcinally updated the diff for D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.

Fix 80 column issue. No other changes intended...

Thu, Oct 22, 7:42 AM · Restricted Project

Mon, Oct 19

cameron.mcinally committed rG629d1d117ae0: [SVE] Update vector reduction intrinsics in new tests. (authored by cameron.mcinally).
[SVE] Update vector reduction intrinsics in new tests.
Mon, Oct 19, 11:28 AM
cameron.mcinally added a comment to D88707: [SVE] Lower fixed length VECREDUCE_AND operation.

Can you update the tests to use the new non-experimental intrinsic name?

Mon, Oct 19, 10:22 AM · Restricted Project
cameron.mcinally added inline comments to D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.
Mon, Oct 19, 9:13 AM · Restricted Project
cameron.mcinally added a comment to D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.

Some other notes:

It looks like NEON FADDA support is missing upstream too.

Mon, Oct 19, 9:10 AM · Restricted Project
cameron.mcinally updated the diff for D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.

This is ready for review now...

Mon, Oct 19, 9:08 AM · Restricted Project

Wed, Oct 14

cameron.mcinally added a comment to D89263: [SVE] Lower fixed length VECREDUCE_FADD operation.

Not lowering to SVE for v2f## MVTs makes sense for now but as before when we have proper support for v#i1 our hands will be tied.

Wed, Oct 14, 7:44 AM · Restricted Project
cameron.mcinally committed rG421f1b7294ef: [SVE] Lower fixed length VECREDUCE_FADD operation (authored by cameron.mcinally).
[SVE] Lower fixed length VECREDUCE_FADD operation
Wed, Oct 14, 7:41 AM
cameron.mcinally closed D89263: [SVE] Lower fixed length VECREDUCE_FADD operation.
Wed, Oct 14, 7:41 AM · Restricted Project

Mon, Oct 12

cameron.mcinally updated the summary of D89263: [SVE] Lower fixed length VECREDUCE_FADD operation.
Mon, Oct 12, 1:02 PM · Restricted Project
cameron.mcinally requested review of D89263: [SVE] Lower fixed length VECREDUCE_FADD operation.
Mon, Oct 12, 1:01 PM · Restricted Project
cameron.mcinally committed rGabe14485fed7: [SVE] Fix VBITS_GE_256 typo in fixed-width tests. (authored by cameron.mcinally).
[SVE] Fix VBITS_GE_256 typo in fixed-width tests.
Mon, Oct 12, 12:33 PM
cameron.mcinally added a comment to D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.

Oh, and I can do VECREDUCE_FADD first. Just hit VECREDUCE_SEQ_FADD before I realized I need to add 'fast' to the intrinsic calls.

Mon, Oct 12, 8:40 AM · Restricted Project
cameron.mcinally added a comment to D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.

You're right on all the comments. I stopped midway when I hit the legalisation issue, so that's why this patch is rough. That would need to be built out first before this work could continue. And I think NEON support should be built out before that. It sounds like I'm not stepping on toes, so I'll go in that order. Thanks.

Mon, Oct 12, 8:29 AM · Restricted Project
cameron.mcinally committed rG974ddb54c9ad: [SVE] Lower fixed length VECREDUCE_XOR operation (authored by cameron.mcinally).
[SVE] Lower fixed length VECREDUCE_XOR operation
Mon, Oct 12, 8:12 AM
cameron.mcinally closed D88974: [SVE] Lower fixed length VECREDUCE_XOR operation.
Mon, Oct 12, 8:12 AM · Restricted Project

Fri, Oct 9

cameron.mcinally requested review of D89162: [SVE] Lower fixed length VECREDUCE_SEQ_FADD operation.
Fri, Oct 9, 2:00 PM · Restricted Project

Wed, Oct 7

cameron.mcinally added a comment to D88707: [SVE] Lower fixed length VECREDUCE_AND operation.

Legalisation tests added for sve-fixed-length-fp-reduce.ll in:

Wed, Oct 7, 1:23 PM · Restricted Project
cameron.mcinally committed rG365ef499d600: [SVE] Add legalisation tests to sve-fixed-length-fp-reduce.ll (authored by cameron.mcinally).
[SVE] Add legalisation tests to sve-fixed-length-fp-reduce.ll
Wed, Oct 7, 1:22 PM
cameron.mcinally updated the diff for D88974: [SVE] Lower fixed length VECREDUCE_XOR operation.

Correct OR->EOR variable name.

Wed, Oct 7, 9:09 AM · Restricted Project
cameron.mcinally requested review of D88974: [SVE] Lower fixed length VECREDUCE_XOR operation.
Wed, Oct 7, 9:05 AM · Restricted Project
cameron.mcinally committed rG333b2ab60b61: [SVE] Lower fixed length VECREDUCE_OR operation (authored by cameron.mcinally).
[SVE] Lower fixed length VECREDUCE_OR operation
Wed, Oct 7, 7:56 AM
cameron.mcinally closed D88847: [SVE] Lower fixed length VECREDUCE_OR operation.
Wed, Oct 7, 7:56 AM · Restricted Project

Mon, Oct 5

cameron.mcinally added a comment to D88707: [SVE] Lower fixed length VECREDUCE_AND operation.

Legalisation tests added for sve-fixed-length-int-reduce.ll in:

Mon, Oct 5, 2:06 PM · Restricted Project
cameron.mcinally committed rG6bec45e25585: [SVE] Add legalisation tests to sve-fixed-length-int-reduce.ll (authored by cameron.mcinally).
[SVE] Add legalisation tests to sve-fixed-length-int-reduce.ll
Mon, Oct 5, 2:06 PM
cameron.mcinally requested review of D88847: [SVE] Lower fixed length VECREDUCE_OR operation.
Mon, Oct 5, 12:50 PM · Restricted Project
cameron.mcinally accepted D88671: [SVE] Lower fixed length vector floating point rounding operations..

LGTM

Mon, Oct 5, 12:46 PM · Restricted Project
cameron.mcinally added a comment to D88707: [SVE] Lower fixed length VECREDUCE_AND operation.

Committed. It looks like the legalisations seem reasonable. Something like:

Mon, Oct 5, 9:38 AM · Restricted Project
cameron.mcinally committed rG9642ded8ba64: [SVE] Lower fixed length VECREDUCE_AND operation (authored by cameron.mcinally).
[SVE] Lower fixed length VECREDUCE_AND operation
Mon, Oct 5, 9:34 AM
cameron.mcinally closed D88707: [SVE] Lower fixed length VECREDUCE_AND operation.
Mon, Oct 5, 9:34 AM · Restricted Project

Fri, Oct 2

cameron.mcinally added inline comments to D88683: [SVE] Lower fixed length vector fneg and fsqrt operations..
Fri, Oct 2, 6:25 AM · Restricted Project
cameron.mcinally added a comment to D88707: [SVE] Lower fixed length VECREDUCE_AND operation.

we start adding proper v#i1 support

Fri, Oct 2, 6:18 AM · Restricted Project

Thu, Oct 1

cameron.mcinally requested review of D88707: [SVE] Lower fixed length VECREDUCE_AND operation.
Thu, Oct 1, 3:33 PM · Restricted Project
cameron.mcinally accepted D88683: [SVE] Lower fixed length vector fneg and fsqrt operations..

LGTM

Thu, Oct 1, 3:09 PM · Restricted Project

Tue, Sep 29

cameron.mcinally committed rG80381c4dc925: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable (authored by cameron.mcinally).
[SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable
Tue, Sep 29, 2:31 PM
cameron.mcinally closed D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.
Tue, Sep 29, 2:31 PM · Restricted Project
cameron.mcinally added a comment to D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.

Ok, I think that's all of them. Looks like it started with D87796 and was buried in other changes. To confirm:

Tue, Sep 29, 9:24 AM · Restricted Project
cameron.mcinally added a comment to D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.

Sorry again, Paul. Still looking at how far this propagated. Looks like it was introduced with the VECREDUC_ADD patch. Need some more time...

Tue, Sep 29, 9:14 AM · Restricted Project
cameron.mcinally updated the diff for D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.

Update more typos...

Tue, Sep 29, 9:10 AM · Restricted Project
cameron.mcinally added inline comments to D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.
Tue, Sep 29, 8:33 AM · Restricted Project
cameron.mcinally committed rG01c95f79424d: [SVE] Fix typo in CHECK lines for sve-fixed-length-int-reduce.ll (authored by cameron.mcinally).
[SVE] Fix typo in CHECK lines for sve-fixed-length-int-reduce.ll
Tue, Sep 29, 8:25 AM
cameron.mcinally added inline comments to D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.
Tue, Sep 29, 8:14 AM · Restricted Project
cameron.mcinally added inline comments to D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.
Tue, Sep 29, 7:44 AM · Restricted Project
cameron.mcinally updated the diff for D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.

Address reviews...

Tue, Sep 29, 7:43 AM · Restricted Project

Mon, Sep 28

cameron.mcinally added a comment to D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.

Also, while I have everyone's attention, there are a number of unhandled vector reduction intrinsics with SVE support. Do we want to add lowerings for those? E.g. ANDV.

Mon, Sep 28, 1:17 PM · Restricted Project
cameron.mcinally requested review of D88444: [SVE] Lower fixed length VECREDUCE_[FMAX|FMIN] to Scalable.
Mon, Sep 28, 1:10 PM · Restricted Project
cameron.mcinally committed rG9b0b09671cdd: [SVE] Lower fixed length VECREDUCE_[UMAX|UMIN] to Scalable (authored by cameron.mcinally).
[SVE] Lower fixed length VECREDUCE_[UMAX|UMIN] to Scalable
Mon, Sep 28, 7:29 AM
cameron.mcinally closed D88317: [SVE] Lower fixed length VECREDUCE_[UMAX|UMIN] to Scalable.
Mon, Sep 28, 7:29 AM · Restricted Project
cameron.mcinally added a comment to D88317: [SVE] Lower fixed length VECREDUCE_[UMAX|UMIN] to Scalable.

Just a heads up that @kmclaughlin is starting to look at legalisation/lowering for scalable vector types. From an operation legalisation point of view you've done most of the plumbing so hopefully any toe treading will be minimal.

Mon, Sep 28, 7:06 AM · Restricted Project

Fri, Sep 25

cameron.mcinally updated the diff for D88317: [SVE] Lower fixed length VECREDUCE_[UMAX|UMIN] to Scalable.

Correct bad string replacement.

Fri, Sep 25, 9:22 AM · Restricted Project
cameron.mcinally added inline comments to D88317: [SVE] Lower fixed length VECREDUCE_[UMAX|UMIN] to Scalable.
Fri, Sep 25, 9:20 AM · Restricted Project
cameron.mcinally requested review of D88317: [SVE] Lower fixed length VECREDUCE_[UMAX|UMIN] to Scalable.
Fri, Sep 25, 9:17 AM · Restricted Project
cameron.mcinally added inline comments to D88259: [SVE] Lower fixed length VECREDUCE_[SMAX|SMIN] to Scalable.
Fri, Sep 25, 8:18 AM · Restricted Project
cameron.mcinally committed rG9a4767411e89: [SVE] Revert accidental change from 405e22fbe8719cff6c40eec15c2044f42527f116 (authored by cameron.mcinally).
[SVE] Revert accidental change from 405e22fbe8719cff6c40eec15c2044f42527f116
Fri, Sep 25, 8:14 AM
cameron.mcinally added a comment to D78778: Add SupportsDebugUnwindInformation to MCAsmInfo.

Cray/HPE is motivated to see this patch land. My confidence is low in this area, but I don't see any review comments that are fatal. Any reason we can't move forward with this change?

Fri, Sep 25, 8:03 AM · debug-info, Restricted Project
cameron.mcinally committed rGe2ccf7f178dd: [SVE] Lower fixed length VECREDUCE_[SMAX|SMIN] to Scalable (authored by cameron.mcinally).
[SVE] Lower fixed length VECREDUCE_[SMAX|SMIN] to Scalable
Fri, Sep 25, 7:58 AM
cameron.mcinally closed D88259: [SVE] Lower fixed length VECREDUCE_[SMAX|SMIN] to Scalable.
Fri, Sep 25, 7:58 AM · Restricted Project
cameron.mcinally added inline comments to D88259: [SVE] Lower fixed length VECREDUCE_[SMAX|SMIN] to Scalable.
Fri, Sep 25, 7:33 AM · Restricted Project

Thu, Sep 24

cameron.mcinally requested review of D88259: [SVE] Lower fixed length VECREDUCE_[SMAX|SMIN] to Scalable.
Thu, Sep 24, 1:13 PM · Restricted Project

Wed, Sep 23

cameron.mcinally committed rGe8413ac97f6c: [AArch64] Expand some vector of i64 reductions on NEON (authored by cameron.mcinally).
[AArch64] Expand some vector of i64 reductions on NEON
Wed, Sep 23, 2:02 PM
cameron.mcinally closed D88161: [AArch64] No NEON instructions to support vector of i64 reductions.
Wed, Sep 23, 2:01 PM · Restricted Project
cameron.mcinally updated the diff for D88161: [AArch64] No NEON instructions to support vector of i64 reductions.

How about something like this? Not really exhaustive coverage, but it matches the current legalization tests for the reduce intrinsics.

Wed, Sep 23, 1:03 PM · Restricted Project
cameron.mcinally updated the diff for D88161: [AArch64] No NEON instructions to support vector of i64 reductions.

You're right about v2i64 UADDV. The existing test didn't catch my regression, so I added a new CHECK line to that test. Also added a line to Custom lower that as before.

Wed, Sep 23, 12:22 PM · Restricted Project
cameron.mcinally requested review of D88161: [AArch64] No NEON instructions to support vector of i64 reductions.
Wed, Sep 23, 9:30 AM · Restricted Project
cameron.mcinally accepted D76879: [AMDGPU] Begin emitting CFI for AMDGCN.

LGTM

Wed, Sep 23, 7:28 AM · debug-info, Restricted Project
cameron.mcinally committed rGdb40a7434429: [SVE] Lower fixed length ISD::VECREDUCE_ADD to Scalable (authored by cameron.mcinally).
[SVE] Lower fixed length ISD::VECREDUCE_ADD to Scalable
Wed, Sep 23, 7:08 AM
cameron.mcinally closed D87796: [SVE][WIP] Lower fixed length VECREDUCE_ADD to Scalable.
Wed, Sep 23, 7:08 AM · Restricted Project

Sep 22 2020

cameron.mcinally accepted D76880: [AMDGPU] Emit entry function CFI.

LGTM with moderate confidence. Does anyone see a reason that this shouldn't land?

Sep 22 2020, 2:05 PM · debug-info, Restricted Project
cameron.mcinally accepted D76884: [AMDGPU] Implement -amdgpu-spill-cfi-saved-regs.

LGTM with moderate confidence. Does anyone see a reason that this shouldn't land?

Sep 22 2020, 1:13 PM · debug-info, Restricted Project
cameron.mcinally updated the diff for D87796: [SVE][WIP] Lower fixed length VECREDUCE_ADD to Scalable.

A few other cleanliness opportunities popped up during the refactor. How about something like this?

Sep 22 2020, 8:34 AM · Restricted Project

Sep 21 2020

cameron.mcinally updated the diff for D87796: [SVE][WIP] Lower fixed length VECREDUCE_ADD to Scalable.

Thanks for the patches, @paulwalker-arm. Those helped a lot.

Sep 21 2020, 3:05 PM · Restricted Project

Sep 17 2020

cameron.mcinally committed rGa35c7f30769b: [SVE][WIP] Implement lowering for fixed length VSELECT to Scalable (authored by cameron.mcinally).
[SVE][WIP] Implement lowering for fixed length VSELECT to Scalable
Sep 17 2020, 12:03 PM
cameron.mcinally closed D85364: [SVE][WIP] Implement lowering for fixed width select.
Sep 17 2020, 12:03 PM · Restricted Project
cameron.mcinally planned changes to D87796: [SVE][WIP] Lower fixed length VECREDUCE_ADD to Scalable.

This is certainly not the expected behaviour so I'll get it fixed.

Sep 17 2020, 9:04 AM · Restricted Project
cameron.mcinally updated the diff for D85364: [SVE][WIP] Implement lowering for fixed width select.

Here's the most obvious choice -- glueing the new check to the AArch64ISD::EXT generation. That's good because the intention is obvious, and future code changes won't get in the way. But it could also be argued that the NEON-sized vector check should precede this entire if-else statement. I.e. the EXTRACT_SUBVECTORs have the same problem as AArch64ISD::EXT. None of this code is ready for SVE-sized fixed vectors.

Sep 17 2020, 9:02 AM · Restricted Project

Sep 16 2020

cameron.mcinally requested review of D87796: [SVE][WIP] Lower fixed length VECREDUCE_ADD to Scalable.
Sep 16 2020, 2:15 PM · Restricted Project

Sep 14 2020

cameron.mcinally added inline comments to D85364: [SVE][WIP] Implement lowering for fixed width select.
Sep 14 2020, 2:13 PM · Restricted Project
cameron.mcinally added inline comments to D85364: [SVE][WIP] Implement lowering for fixed width select.
Sep 14 2020, 7:45 AM · Restricted Project
cameron.mcinally updated the diff for D85364: [SVE][WIP] Implement lowering for fixed width select.

Remove the BUILD_VECTOR->EXTRACT_SUBVECTOR transform so that this is easier to review. Will post a separate Diff for that.

Sep 14 2020, 7:42 AM · Restricted Project

Sep 8 2020

cameron.mcinally updated the diff for D85364: [SVE][WIP] Implement lowering for fixed width select.

Fix comment.

Sep 8 2020, 8:49 AM · Restricted Project
cameron.mcinally updated the diff for D85364: [SVE][WIP] Implement lowering for fixed width select.

Turns out that ReconstructShuffle(...) needs a lot of work to be updated. The EXTRACT_SUBVECTOR index assumes we're extracting a 1/2 width vector. I missed that.

Sep 8 2020, 8:44 AM · Restricted Project

Sep 3 2020

cameron.mcinally added inline comments to D85364: [SVE][WIP] Implement lowering for fixed width select.
Sep 3 2020, 8:40 AM · Restricted Project
cameron.mcinally added inline comments to D85364: [SVE][WIP] Implement lowering for fixed width select.
Sep 3 2020, 6:35 AM · Restricted Project

Sep 2 2020

cameron.mcinally updated the diff for D85364: [SVE][WIP] Implement lowering for fixed width select.

Update tests -- don't use pointer args for NEON sized vectors.

Sep 2 2020, 1:02 PM · Restricted Project
cameron.mcinally updated the diff for D85364: [SVE][WIP] Implement lowering for fixed width select.

Fix bad copy-and-paste in CHECK lines. No significant changes made.

Sep 2 2020, 12:18 PM · Restricted Project
cameron.mcinally added a reviewer for D85364: [SVE][WIP] Implement lowering for fixed width select: t.p.northover.
Sep 2 2020, 8:45 AM · Restricted Project
cameron.mcinally updated subscribers of D85364: [SVE][WIP] Implement lowering for fixed width select.
Sep 2 2020, 8:45 AM · Restricted Project
cameron.mcinally updated the diff for D85364: [SVE][WIP] Implement lowering for fixed width select.

Updating Diff for recent upstream changes.

Sep 2 2020, 8:38 AM · Restricted Project

Sep 1 2020

cameron.mcinally committed rGcfe2b81710c4: [SVE] Update INSERT_SUBVECTOR DAGCombine to use getVectorElementCount(). (authored by cameron.mcinally).
[SVE] Update INSERT_SUBVECTOR DAGCombine to use getVectorElementCount().
Sep 1 2020, 2:52 PM
cameron.mcinally closed D86894: [SVE] Disable INSERT_SUBVECTOR DAGCombine for scalable vectors.
Sep 1 2020, 2:52 PM · Restricted Project
cameron.mcinally added a comment to D86894: [SVE] Disable INSERT_SUBVECTOR DAGCombine for scalable vectors.

Is there any way to write a test for the transform? Maybe not, given the limited places we use INSERT_SUBVECTOR.

Sep 1 2020, 1:04 PM · Restricted Project
cameron.mcinally updated the diff for D86894: [SVE] Disable INSERT_SUBVECTOR DAGCombine for scalable vectors.

Updated with @david-arm's comments for Thursday's Sync-up call.

Sep 1 2020, 9:01 AM · Restricted Project
cameron.mcinally added a comment to D86894: [SVE] Disable INSERT_SUBVECTOR DAGCombine for scalable vectors.

Thanks, David. Ok, I'll put a pin in this until Thursday's call. But yeah, isKnownMultipleOf(...) sounds fair. I'll prepare a patch to see if any surprises come up.

Sep 1 2020, 7:19 AM · Restricted Project

Aug 31 2020

cameron.mcinally added a comment to D86894: [SVE] Disable INSERT_SUBVECTOR DAGCombine for scalable vectors.

I don't really have an opinion on any of this. Just thinking aloud...

Aug 31 2020, 4:23 PM · Restricted Project
cameron.mcinally updated the diff for D86894: [SVE] Disable INSERT_SUBVECTOR DAGCombine for scalable vectors.

Ok, I guess that makes sense. So if we're considering Min elements, should we just overload % then?

Aug 31 2020, 2:47 PM · Restricted Project
cameron.mcinally requested review of D86894: [SVE] Disable INSERT_SUBVECTOR DAGCombine for scalable vectors.
Aug 31 2020, 1:29 PM · Restricted Project

Aug 26 2020

cameron.mcinally added a comment to D85364: [SVE][WIP] Implement lowering for fixed width select.

˜Bah, egg on my face. You're right that D86394 fixes the immediate issue. Sorry for the noise.

Aug 26 2020, 7:55 AM · Restricted Project

Aug 24 2020

cameron.mcinally added a comment to D85546: [SVE] Add ISD nodes for predicated integer extend inreg operations.

D86394 addresses a different issue, I think. Posted some new code in D85364 to expose the AArch64ISD::ZERO_EXTEND_INREG_MERGE_PASSTHRU issue.

Aug 24 2020, 1:41 PM · Restricted Project