Page MenuHomePhabricator

Please use GitHub pull requests for new patches. Phabricator shutdown timeline

ebrevnov (Evgeniy)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 23 2018, 4:17 PM (296 w, 3 d)

Recent Activity

Aug 28 2023

ebrevnov added a comment to D155831: [LV][WIP] Lazy creation of BFI when required by cost model.

Any feedback?

Aug 28 2023, 5:12 AM · Restricted Project, Restricted Project

Aug 4 2023

ebrevnov added inline comments to D155960: [NaryReassociate] Use new access type aware getGEPCost.
Aug 4 2023, 1:02 AM · Restricted Project, Restricted Project

Aug 3 2023

ebrevnov added inline comments to D155960: [NaryReassociate] Use new access type aware getGEPCost.
Aug 3 2023, 2:39 AM · Restricted Project, Restricted Project

Aug 2 2023

ebrevnov added a comment to D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .

Just realized the commit title is not in line with the implementation :(

Aug 2 2023, 10:59 PM · Restricted Project, Restricted Project
ebrevnov retitled D151204: [VPlan] Disallow sinking side-effecting first order recurrence users from [VPlan] Allow sinking of instructions with no defs to [VPlan] Disallow sinking side-effecting first order recurrence users .
Aug 2 2023, 10:59 PM · Restricted Project, Restricted Project

Jul 21 2023

ebrevnov requested review of D155948: [LV][WIP] dd-exit loops vectorization.
Jul 21 2023, 5:32 AM · Restricted Project, Restricted Project

Jul 20 2023

ebrevnov added a comment to D147114: [LV] Use BFI to adjust cost of predicated instructions.

@nikic @fhahn please take a look at an alternative solution demonstrating an idea of lazy creation of BFI D155831.

Jul 20 2023, 6:48 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D155831: [LV][WIP] Lazy creation of BFI when required by cost model.

Removed acidental changes

Jul 20 2023, 6:47 AM · Restricted Project, Restricted Project
ebrevnov updated the summary of D155831: [LV][WIP] Lazy creation of BFI when required by cost model.
Jul 20 2023, 6:46 AM · Restricted Project, Restricted Project
ebrevnov requested review of D155831: [LV][WIP] Lazy creation of BFI when required by cost model.
Jul 20 2023, 5:59 AM · Restricted Project, Restricted Project
ebrevnov abandoned D149338: [SLP] Add TreeEntry into PostponedGathers if it depends on another previously postponed TreeEntry.
Jul 20 2023, 3:34 AM · Restricted Project, Restricted Project

Jul 4 2023

ebrevnov added a comment to D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .

Florian, thank you for review.

Jul 4 2023, 2:50 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D154157: [LV] Cost model for out-of-loop reductions.

I support Florian's suggestion. We better take reduction overhead into account the same way as we did for runtime checks.
Today's cost model is very limited in its ability to predict the overhead which comes with the vectorization. For example, the overhead of the scalar epilogue loop is simply ignored. I think, one day we should generalize the cost model so that vectorization overhead is integral part of the model itself rather than a follow up "fix up".

Jul 4 2023, 1:20 AM · Restricted Project, Restricted Project

Jul 2 2023

ebrevnov added a comment to D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .

Florian, I've addressed yours concerns. Please take a look now.

Jul 2 2023, 9:56 PM · Restricted Project, Restricted Project

Jun 29 2023

ebrevnov updated the diff for D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .

Prevent sinking side-effecting instructions

Jun 29 2023, 5:00 AM · Restricted Project, Restricted Project

Jun 21 2023

ebrevnov added a comment to D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .

gentle ping...

Jun 21 2023, 11:02 PM · Restricted Project, Restricted Project

Jun 14 2023

ebrevnov added inline comments to D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .
Jun 14 2023, 1:01 AM · Restricted Project, Restricted Project

Jun 12 2023

ebrevnov added inline comments to D152701: LowerMemIntrinsics: Check address space aliasing for memmove expansion.
Jun 12 2023, 10:06 PM · Restricted Project, Restricted Project

Jun 5 2023

ebrevnov added a comment to D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .

ping

Jun 5 2023, 10:55 PM · Restricted Project, Restricted Project

Jun 1 2023

ebrevnov added inline comments to D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .
Jun 1 2023, 2:12 AM · Restricted Project, Restricted Project

May 23 2023

ebrevnov updated the summary of D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .
May 23 2023, 5:09 AM · Restricted Project, Restricted Project
ebrevnov requested review of D151204: [VPlan] Disallow sinking side-effecting first order recurrence users .
May 23 2023, 5:06 AM · Restricted Project, Restricted Project

Apr 27 2023

ebrevnov added a comment to D149338: [SLP] Add TreeEntry into PostponedGathers if it depends on another previously postponed TreeEntry.
Apr 27 2023, 10:57 PM · Restricted Project, Restricted Project
ebrevnov added a comment to D147562: [SLP] Add TreeEntry into PostponedGathers if it depends on another previously postponed TreeEntry..

This issue affects us as well. To make progress I've duplicated this into D149338 with full context and test case reproducing the issue. If @ikelarev is still interested in this he can add his test anytime.

Apr 27 2023, 5:50 AM · Restricted Project, Restricted Project
ebrevnov updated the summary of D149338: [SLP] Add TreeEntry into PostponedGathers if it depends on another previously postponed TreeEntry.
Apr 27 2023, 5:43 AM · Restricted Project, Restricted Project
ebrevnov requested review of D149338: [SLP] Add TreeEntry into PostponedGathers if it depends on another previously postponed TreeEntry.
Apr 27 2023, 5:40 AM · Restricted Project, Restricted Project

Apr 20 2023

ebrevnov added inline comments to D147114: [LV] Use BFI to adjust cost of predicated instructions.
Apr 20 2023, 2:48 AM · Restricted Project, Restricted Project
ebrevnov added inline comments to D147114: [LV] Use BFI to adjust cost of predicated instructions.
Apr 20 2023, 1:26 AM · Restricted Project, Restricted Project

Apr 14 2023

ebrevnov added a comment to D147114: [LV] Use BFI to adjust cost of predicated instructions.

Are there any additional concerns?

Apr 14 2023, 3:20 AM · Restricted Project, Restricted Project

Apr 5 2023

ebrevnov added inline comments to D147114: [LV] Use BFI to adjust cost of predicated instructions.
Apr 5 2023, 10:16 PM · Restricted Project, Restricted Project

Mar 30 2023

ebrevnov added inline comments to D147114: [LV] Use BFI to adjust cost of predicated instructions.
Mar 30 2023, 10:19 PM · Restricted Project, Restricted Project
ebrevnov updated the diff for D147114: [LV] Use BFI to adjust cost of predicated instructions.

Updated as requested. New tests added.

Mar 30 2023, 10:17 PM · Restricted Project, Restricted Project
ebrevnov added a comment to D146928: [NaryReassociate] Transform expression (X << C1) + C2 to (X + C3) << C1,.

As I see it, you are trying to solve pass ordering issue (InstCombine prevents GVN/CSE). But your solution assumes there is GVN/CSE follows NaryReassociation what creates another potential pass ordering dependence.
Another point is the way the transformation is implemented right now doesn't fit intent of NaryReassociation pass. In the heart of NaryReassociation pass is findClosestMatchingDominator call to find and delete common expressions after reassociation. Since you are not checking for existence of common expressions you can easily do the transformation in some other place (ordinary Reassociate pass, or InstSimplify/InstCombine)
In order to solve both mentioned problems and cover your case you will need to use findClosestMatchingDominator and possibly extend it to look for post dominating expressions as well.

Mar 30 2023, 2:37 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D147114: [LV] Use BFI to adjust cost of predicated instructions.

Do we have a test with profile info? Might be worth having dedicated tests for this feature with a range of branch probabilities if possible

Sure. Will such tests.

Mar 30 2023, 1:37 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D147114: [LV] Use BFI to adjust cost of predicated instructions.

Do you have any benchmark data to back it up?
I have motivating example which shows almost 2x degradation due to vectorization. I will add it as lit test.

Mar 30 2023, 1:35 AM · Restricted Project, Restricted Project

Mar 28 2023

ebrevnov added reviewers for D147114: [LV] Use BFI to adjust cost of predicated instructions: fhahn, dmgreen, mssimpso.
Mar 28 2023, 11:25 PM · Restricted Project, Restricted Project
ebrevnov requested review of D147114: [LV] Use BFI to adjust cost of predicated instructions.
Mar 28 2023, 11:17 PM · Restricted Project, Restricted Project

Feb 16 2023

ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Fixed fuchsia-x86_64-linux, ppc64le-lld-multistage-test, clang-with-lto-ubuntu buildbot failures

Feb 16 2023, 12:58 AM · Restricted Project, Restricted Project
ebrevnov reopened D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Feb 16 2023, 12:58 AM · Restricted Project, Restricted Project

Jan 26 2023

ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Added extra RUN; line into select.ll test as requested

Jan 26 2023, 11:55 PM · Restricted Project, Restricted Project
ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Rebase

Jan 26 2023, 10:26 PM · Restricted Project, Restricted Project

Jan 11 2023

ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Check that (TrueWeight + FalseWeight) are not zero.

Jan 11 2023, 12:57 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Removed code related to PostDomTree + rebase

Jan 11 2023, 12:45 AM · Restricted Project, Restricted Project

Dec 20 2022

ebrevnov added a comment to D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Ping

Dec 20 2022, 2:28 AM · Restricted Project, Restricted Project

Nov 25 2022

ebrevnov added inline comments to D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Nov 25 2022, 2:42 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Invalidate all existing not preserved analysis before running "extrnal" analysis

Nov 25 2022, 2:39 AM · Restricted Project, Restricted Project
ebrevnov reopened D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Nov 25 2022, 2:39 AM · Restricted Project, Restricted Project

Nov 24 2022

ebrevnov updated the diff for D137149: Use PassGate from LLVMContext if any otherwise global one.

Fixed build error in flang/lib/Frontend/FrontendActions.cpp

Nov 24 2022, 1:55 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Nov 23 2022

ebrevnov updated the diff for D137149: Use PassGate from LLVMContext if any otherwise global one.

Updated

Nov 23 2022, 3:49 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Nov 22 2022

ebrevnov added inline comments to D137149: Use PassGate from LLVMContext if any otherwise global one.
Nov 22 2022, 10:14 PM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov added inline comments to D137149: Use PassGate from LLVMContext if any otherwise global one.
Nov 22 2022, 1:11 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Nov 16 2022

ebrevnov added reviewers for D138132: [JumpThreading] Preserve profile metadata during select unfolding: anemet, manmanren.
Nov 16 2022, 10:08 PM · Restricted Project, Restricted Project
ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Addressed review comments

Nov 16 2022, 10:04 PM · Restricted Project, Restricted Project
ebrevnov added inline comments to D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Nov 16 2022, 9:57 PM · Restricted Project, Restricted Project

Nov 15 2022

ebrevnov updated the diff for D137149: Use PassGate from LLVMContext if any otherwise global one.

Pass LLVMContext to the StandardInstrumentations constructor

Nov 15 2022, 11:50 PM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov added a comment to D137149: Use PassGate from LLVMContext if any otherwise global one.

@aeubanks Any comments?

Nov 15 2022, 12:38 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Nov 11 2022

ebrevnov added a comment to D137149: Use PassGate from LLVMContext if any otherwise global one.

Gentle ping.

Nov 11 2022, 1:54 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov added inline comments to D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Nov 11 2022, 1:52 AM · Restricted Project, Restricted Project

Nov 9 2022

ebrevnov added a comment to D137165: [JumpThreading] Copy profile metadata on select unfolding.

Looks OK, except mentioned nit.

Nov 9 2022, 4:34 AM · Restricted Project, Restricted Project
ebrevnov added inline comments to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..
Nov 9 2022, 1:05 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

Avoid vectorization if there is reduction/recurrence induced overhead

Nov 9 2022, 1:01 AM · Restricted Project, Restricted Project

Nov 7 2022

ebrevnov added a comment to D137149: Use PassGate from LLVMContext if any otherwise global one.

chasing pointers can be expensive, and the current compile time regressions are unacceptable so requesting changes in phabricator

perhaps we can only register the callback if either -opt-bisect-limit is set or if some other new cl::opt (-force-pass-gate-instrumentation?), then you'd specify that option when debugging?

That looks a really hacky to me ... I would choose that option as a last resort.
Once realized we can't afford requesting pass gate from the module on each invocation of callback (due to CT) I'm inclined to think that passing LLVMContext (and FAM as well) during construction of StandardInstrumentations is the right way to go. This way we explicitly set expectations that StandardInstrumentations is tied to a single LLVMContext. Are you OK with such approach? If not then I will go with the caching approach (but I think it's less expressive and more complicated).

Nov 7 2022, 10:23 PM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov added inline comments to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..
Nov 7 2022, 4:45 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

I've been taking a look at the example that is getting a lot worse. There are certainly some issues with the code generation being non-optimal, but even after a lot of optimizations it looks like it will always be worse than the scalar version. There is a lot of predications and fairly efficient scalar instructions like BFI, which makes accurate cost modelling difficult.

I would like to understand you case better. Let's take current non optimality of code generation out of the consideration. Is it a problem of inaccurate cost estimation for some particular instructions or not taking overhead which comes with the vectorization into account?

Nov 7 2022, 4:32 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D137149: Use PassGate from LLVMContext if any otherwise global one.

thanks for the context, agreed that using the LLVMContext's pass gate is what's intended

the cleanest way IMO is to get the IR's LLVMContext (e.g. via unwrapModule(IR, /*Force*/ true).getContext()) and use the pass gate from there, rather than collect the pass gate at the beginning. WDYT of something like https://reviews.llvm.org/D137358?

Totally agree. Just didn't realize we already can access the module. Integrated suggested approach except caching part. I think caching is unnecessary complication in this case.

then you start running into compile time issues: https://llvm-compile-time-tracker.com/compare.php?from=66b830889d9bc4bfc58aa1ea47d1b3a7cbee7c5c&to=b7972184a59c0aa0b634de653be1c1363e980f4c&stat=instructions:u
previously we didn't call OptPassGateInstrumentation::registerCallbacks and run these callbacks before every pass if -opt-bisect-limit was enabled, with this patch we do.

That's kind of unexpected (at least for me). Do we known why such lightweight operation noticeably impacts compile time? Is it getting a module from a loop or something else?

Nov 7 2022, 1:56 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Nov 4 2022

ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Minor update

Nov 4 2022, 6:50 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.

Added DominatorTree verification. Do not preserve BPI/BFI after JumpThreading.

Nov 4 2022, 6:44 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D137149: Use PassGate from LLVMContext if any otherwise global one.

thanks for the context, agreed that using the LLVMContext's pass gate is what's intended

the cleanest way IMO is to get the IR's LLVMContext (e.g. via unwrapModule(IR, /*Force*/ true).getContext()) and use the pass gate from there, rather than collect the pass gate at the beginning. WDYT of something like https://reviews.llvm.org/D137358?

Nov 4 2022, 2:13 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov updated the diff for D137149: Use PassGate from LLVMContext if any otherwise global one.

Integreated "unwrapModule" approach

Nov 4 2022, 2:10 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Nov 3 2022

ebrevnov added a comment to D137149: Use PassGate from LLVMContext if any otherwise global one.

It's sort of strange to have a pass gate reference in the LLVMContext and not use it. But you could just have a StandardInstrumentations::setPassGate method to store a reference locally.

IMHO, we should make StandardInstrumentations context aware... Alternative to StandardInstrumentations::setPassGate could be passing LLVMContext during StandardInstrumentations construction.

Nov 3 2022, 3:25 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov added inline comments to D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Nov 3 2022, 1:43 AM · Restricted Project, Restricted Project
ebrevnov added inline comments to D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Nov 3 2022, 1:42 AM · Restricted Project, Restricted Project
ebrevnov added inline comments to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..
Nov 3 2022, 1:13 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

I tried running the original benchmarks again with this patch series, it still sees the same decrease in performance I'm afraid. If it was a small change it would be understandable, we accidentally end up on the wrong side of the scalar cost and choose to vectorize where we shouldn't. But the new code is 40% slower than the scalar version, so it's quite a difference. I haven't had a chance to look into the costs it produces, there is a chance we are underestimating the cost of predication or overestimating the cost of scalar. At worst, providing getMinTripCountTailFoldingThreshold works like it should we can always put a limit on the tripcount, providing we can find a decent minimum that works in general for MVE.

It sounds like current cost model works really bad for this case and it's essentially the reason why getMinTripCountTailFoldingThreshold had been introduced. If this is the case would you mind trying updated version and see if it helps. Otherwise, I'd probably need a test case I could analyze.

Nov 3 2022, 1:09 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

Make sure MinTripCountTailFoldingThreshold is honord even for loops with short compile tine known TC.

Nov 3 2022, 1:02 AM · Restricted Project, Restricted Project

Nov 2 2022

ebrevnov added a comment to D137149: Use PassGate from LLVMContext if any otherwise global one.

ideally we wouldn't have to pass LLVMContext if possible, it makes it easier for users of this API to mess up. the optional FAM parameter to StandardInstrumentations already bothers me

I agree. Let's see if we can find something better...

Nov 2 2022, 9:58 PM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov added a comment to D137165: [JumpThreading] Copy profile metadata on select unfolding.

We should update BPI\BFI in a similar way as it is done in updateBlockFreqAndEdgeWeight.

Nov 2 2022, 4:53 AM · Restricted Project, Restricted Project

Nov 1 2022

ebrevnov updated the summary of D137149: Use PassGate from LLVMContext if any otherwise global one.
Nov 1 2022, 10:31 PM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
ebrevnov requested review of D137149: Use PassGate from LLVMContext if any otherwise global one.
Nov 1 2022, 3:10 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Oct 31 2022

ebrevnov added a comment to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

Ping!

Oct 31 2022, 7:51 AM · Restricted Project, Restricted Project

Oct 27 2022

ebrevnov added a reviewer for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading: mkazantsev.
Oct 27 2022, 2:46 AM · Restricted Project, Restricted Project
ebrevnov retitled D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading from [JT] Preserve exisiting BPI/BFI during JumpThreading to [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Oct 27 2022, 2:44 AM · Restricted Project, Restricted Project
ebrevnov added reviewers for D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading: nikic, fhahn, jyknight.
Oct 27 2022, 2:44 AM · Restricted Project, Restricted Project
ebrevnov requested review of D136827: [JT][CT] Preserve exisiting BPI/BFI during JumpThreading.
Oct 27 2022, 2:39 AM · Restricted Project, Restricted Project

Oct 17 2022

ebrevnov added inline comments to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..
Oct 17 2022, 6:34 AM · Restricted Project, Restricted Project

Oct 14 2022

ebrevnov added a comment to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

Thanks for feedback!
The problem with mve-known-trip-count.ll was caused by 2 things 1) Outdated code base 2) Bug in the implementation caused to skip "preferPredicateOverEpilogue" check and as a result scalar epilogue have been selected. Now both items are fixed mve-known-trip-count.ll works as previously (cost model says vectorization with tail folding is not profitable).

Oct 14 2022, 10:30 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

Rebase + fix problem for platforms which prefer tail folding over scalar epilogue.

Oct 14 2022, 10:12 AM · Restricted Project, Restricted Project
ebrevnov requested review of D135971: [LV][NFC][WIP] Restructure getScalarEpilogueLowering.
Oct 14 2022, 10:11 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D115712: [LV] Make sure prefer-predicate-over-epilogue works for short TC loops.

rebase

Oct 14 2022, 10:09 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D115711: [LV][NFC] New test case to check prefer-predicate-over-epilogue for short TC loops.

rebase

Oct 14 2022, 10:05 AM · Restricted Project, Restricted Project
ebrevnov updated the diff for D115710: [LV][NFC] Update test checks using utils/update_test_checks.py.

rebase

Oct 14 2022, 10:03 AM · Restricted Project, Restricted Project

Oct 10 2022

ebrevnov updated the summary of D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..
Oct 10 2022, 2:07 AM · Restricted Project, Restricted Project

Oct 9 2022

ebrevnov added a comment to D115712: [LV] Make sure prefer-predicate-over-epilogue works for short TC loops.

ping

Oct 9 2022, 7:24 AM · Restricted Project, Restricted Project
ebrevnov added reviewers for D115712: [LV] Make sure prefer-predicate-over-epilogue works for short TC loops: anna, nikic, lebedev.ri.
Oct 9 2022, 7:24 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D115711: [LV][NFC] New test case to check prefer-predicate-over-epilogue for short TC loops.

ping

Oct 9 2022, 7:24 AM · Restricted Project, Restricted Project
ebrevnov added reviewers for D115711: [LV][NFC] New test case to check prefer-predicate-over-epilogue for short TC loops: anna, lebedev.ri, nikic.
Oct 9 2022, 7:23 AM · Restricted Project, Restricted Project
ebrevnov added a comment to D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC..

PING!

Oct 9 2022, 7:23 AM · Restricted Project, Restricted Project
ebrevnov added a reviewer for D115713: [LV] Don't apply "TinyTripCountVectorThreshold" for loops with compile time known TC.: anna.
Oct 9 2022, 7:22 AM · Restricted Project, Restricted Project

Sep 14 2022

ebrevnov added inline comments to D130896: [AA] Tracking per-location ModRef info in FunctionModRefBehavior (NFCI).
Sep 14 2022, 10:02 PM · Restricted Project, Restricted Project

Sep 13 2022

ebrevnov accepted D130896: [AA] Tracking per-location ModRef info in FunctionModRefBehavior (NFCI).

I've left some NITs which are not blockers and can be addressed later. LGTM.

Sep 13 2022, 3:06 AM · Restricted Project, Restricted Project