Page MenuHomePhabricator

wenlei (Wenlei He)
User

Projects

User does not belong to any projects.

User Details

User Since
Apr 3 2019, 5:55 PM (77 w, 5 d)

Recent Activity

Sun, Sep 20

wenlei added a comment to D87740: [llvm-profdata]Fix llvm-profdata crash on compact binary profile.

Landed the change on behalf of @wlei.

Sun, Sep 20, 5:00 PM · Restricted Project
wenlei committed rGa8b8a9374a3c: [llvm-profdata]Fix llvm-profdata crash on compact binary profile (authored by wlei).
[llvm-profdata]Fix llvm-profdata crash on compact binary profile
Sun, Sep 20, 5:00 PM
wenlei closed D87740: [llvm-profdata]Fix llvm-profdata crash on compact binary profile.
Sun, Sep 20, 4:59 PM · Restricted Project

Wed, Sep 16

wenlei added inline comments to D87777: [ASAN] Properly deal with musttail calls in ASAN.
Wed, Sep 16, 5:53 PM · Restricted Project
wenlei accepted D87777: [ASAN] Properly deal with musttail calls in ASAN.

LGTM except a minor comment.

Wed, Sep 16, 5:25 PM · Restricted Project
wenlei accepted D87740: [llvm-profdata]Fix llvm-profdata crash on compact binary profile.

LGTM. A small nit: the code snippet in the description could be removed as it's indeed in the codebase already. :)

Wed, Sep 16, 5:02 PM · Restricted Project

Tue, Sep 15

wenlei updated subscribers of D86156: [BFI] Make BFI information available through loop passes inside LoopStandardAnalysisResults.

Just saw this. Thanks @Alina Sbirlea for fixing the tests, much appreciated!

Tue, Sep 15, 9:26 PM · Restricted Project, Restricted Project
wenlei added a comment to D87169: SVML support for log10, sqrt.

Committed on behalf of @ayermolo.

Tue, Sep 15, 5:44 PM · Restricted Project
wenlei committed rG056534dc2b15: SVML support for log10, sqrt (authored by wenlei).
SVML support for log10, sqrt
Tue, Sep 15, 5:41 PM
wenlei closed D87169: SVML support for log10, sqrt.
Tue, Sep 15, 5:41 PM · Restricted Project
wenlei added a comment to D87551: [LICM] Make Loop ICM profile aware again.

Committed on behalf of @modimo.

Tue, Sep 15, 5:28 PM · Restricted Project
wenlei committed rG2c391a5a14ae: [LICM] Make Loop ICM profile aware again (authored by wenlei).
[LICM] Make Loop ICM profile aware again
Tue, Sep 15, 5:26 PM
wenlei closed D87551: [LICM] Make Loop ICM profile aware again.
Tue, Sep 15, 5:25 PM · Restricted Project
wenlei added a comment to D86156: [BFI] Make BFI information available through loop passes inside LoopStandardAnalysisResults.

Committed on behalf of @modimo.

Tue, Sep 15, 4:18 PM · Restricted Project, Restricted Project
wenlei committed rG2ea4c2c598b7: [BFI] Make BFI information available through loop passes inside… (authored by wenlei).
[BFI] Make BFI information available through loop passes inside…
Tue, Sep 15, 4:17 PM
wenlei closed D86156: [BFI] Make BFI information available through loop passes inside LoopStandardAnalysisResults.
Tue, Sep 15, 4:16 PM · Restricted Project, Restricted Project

Mon, Sep 14

wenlei accepted D87620: [TSAN] Handle musttail call properly in EscapeEnumerator (and TSAN).
Mon, Sep 14, 3:43 PM · Restricted Project
wenlei added inline comments to D87620: [TSAN] Handle musttail call properly in EscapeEnumerator (and TSAN).
Mon, Sep 14, 3:43 PM · Restricted Project
wenlei added inline comments to D87620: [TSAN] Handle musttail call properly in EscapeEnumerator (and TSAN).
Mon, Sep 14, 12:50 PM · Restricted Project
wenlei added inline comments to D87620: [TSAN] Handle musttail call properly in EscapeEnumerator (and TSAN).
Mon, Sep 14, 12:08 PM · Restricted Project

Wed, Sep 9

wenlei added a comment to D73739: Exception support for basic block sections.

I expect static and dynamic counts in general will require rounds of tuning and will still hit pathological cases on certain scenarios.

Wed, Sep 9, 1:46 PM · Restricted Project

Tue, Sep 8

wenlei added a comment to D86859: [Coroutine] Make dealing with alloca spills more robust.

Chatted with @lxfind off patch in the morning. A quick summary:

Tue, Sep 8, 11:13 AM · Restricted Project
wenlei added inline comments to D86859: [Coroutine] Make dealing with alloca spills more robust.
Tue, Sep 8, 8:50 AM · Restricted Project

Fri, Sep 4

wenlei committed rG4eef14f9780d: [OpenMPOpt] Assume indirect call always changes ICV (authored by weiwang).
[OpenMPOpt] Assume indirect call always changes ICV
Fri, Sep 4, 9:06 AM
wenlei closed D87104: [OpenMPOpt] Assume indirect call always changes ICV.
Fri, Sep 4, 9:06 AM · Restricted Project
wenlei added inline comments to D86859: [Coroutine] Make dealing with alloca spills more robust.
Fri, Sep 4, 12:11 AM · Restricted Project

Thu, Sep 3

wenlei added a comment to D86730: SVML support for log2.

Landed on behalf of @ayermolo ..

Thu, Sep 3, 12:02 PM · Restricted Project
wenlei committed rGd1be928d23fe: SVML support for log2 (authored by wenlei).
SVML support for log2
Thu, Sep 3, 11:59 AM
wenlei closed D86730: SVML support for log2.
Thu, Sep 3, 11:59 AM · Restricted Project

Wed, Sep 2

wenlei accepted D86730: SVML support for log2.

LGTM. I'd recommend wait a few days for other reviewers.

Wed, Sep 2, 11:56 AM · Restricted Project
wenlei added a comment to D85812: [coroutine] should disable inline before calling coro split.

Took another look, AlwaysInline is part of LLVMipo too. :(

Wed, Sep 2, 11:27 AM · Restricted Project
wenlei accepted D86675: [ThinLTO] Fix a metadata lost issue with DICompileUnit import..

LGTM. @tejohnson appreciate if you could take a look too, thanks. (this is blocking an internal usage for us)

Wed, Sep 2, 8:52 AM · Restricted Project

Aug 28 2020

wenlei added a comment to D85812: [coroutine] should disable inline before calling coro split.

I want to point out a strange layering issue. Note that the test llvm/test/Transforms/Coroutines/coro-inline.ll uses -sample-profile: this implies that Coroutines should be a higher level library than sample-profile (a higher level library can use all facilities available at lower layers). However, LLVMipo (the library which contains sample-profile) has a dependency on LLVMCoroutines (the library which contains Coroutines.h)....... This is a conflict of library layering.

edit: LLVMCoroutines actually depends on LLVMipo, so circular dependency

You're right about the circular dependency on headers. Unlike actual circular dependency on actual libs though, such layering violation from header is not obvious, curious how did you catch this?

@dongAxis1944 to avoid the layering issues, I think you may need to revert back to your original version that prohibits inlining of pre-split coroutine in the general inline cost computation. The down side of that is we might cover up pass ordering issue in the future, but that's minor comparing the layering issue.

if i put the attribute CORO_PRESPLIT_ATTR to the attribute.td, it seem it might not have the layering issue(Because it will not depend coroutine lib itself), do you think it might be better?

Aug 28 2020, 12:10 AM · Restricted Project

Aug 27 2020

wenlei added a comment to D77925: Revert "[TLI] Per-function fveclib for math library used for vectorization".

Overall that would likely work for XLA. Something I'd like to mention though in response to:

The veclib type is also tied to the accepted values for -fveclib, which is a list of supported lib,

-fveclib is a Clang thing, it shouldn't limit what LLVM does. Of course LLVM needs to support Clang, but does not have to inherit the limitation of map 1:1 to Clang UI.
In particular as a library, it isn't clear why we would make the choice to write LLVM VecLib support this way.

Aug 27 2020, 11:41 PM · Restricted Project, Restricted Project
wenlei added a comment to D86156: [BFI] Make BFI information available through loop passes inside LoopStandardAnalysisResults.

Diff looks reasonable at this point. Thank you for the patch!
Please wait on @nikic for compile-time impact or additional feedback.

Just out of curiosity, in D65060, it was mentioned that using BFI got you ~7% improvement for a CPU related metric (@wenlei). Are you seeing benefits from this patch? And which pass manager are you using?

Aug 27 2020, 6:23 PM · Restricted Project, Restricted Project

Aug 26 2020

wenlei updated subscribers of D85408: Let -basic-block-sections=labels emit basicblock metadata in a new .bb_addr_map section, instead of emitting special unary-encoded symbols..
Aug 26 2020, 12:40 PM · Restricted Project, Restricted Project

Aug 25 2020

wenlei added a comment to D85812: [coroutine] should disable inline before calling coro split.

I want to point out a strange layering issue. Note that the test llvm/test/Transforms/Coroutines/coro-inline.ll uses -sample-profile: this implies that Coroutines should be a higher level library than sample-profile (a higher level library can use all facilities available at lower layers). However, LLVMipo (the library which contains sample-profile) has a dependency on LLVMCoroutines (the library which contains Coroutines.h)....... This is a conflict of library layering.

edit: LLVMCoroutines actually depends on LLVMipo, so circular dependency

Aug 25 2020, 11:06 PM · Restricted Project
wenlei accepted D86332: [SampleFDO] Enhance profile remapping support for inline instance.

LGTM, thanks!

Aug 25 2020, 10:07 PM · Restricted Project
wenlei added inline comments to D86332: [SampleFDO] Enhance profile remapping support for inline instance.
Aug 25 2020, 9:25 PM · Restricted Project
wenlei added a comment to D86332: [SampleFDO] Enhance profile remapping support for inline instance.

Great - thanks for making the changes. Looks good with two minor comments.

Aug 25 2020, 8:40 PM · Restricted Project

Aug 24 2020

wenlei added inline comments to D86332: [SampleFDO] Enhance profile remapping support for inline instance.
Aug 24 2020, 10:40 AM · Restricted Project

Aug 23 2020

wenlei added inline comments to D86332: [SampleFDO] Enhance profile remapping support for inline instance.
Aug 23 2020, 11:49 PM · Restricted Project

Aug 22 2020

wenlei added a comment to D86156: [BFI] Make BFI information available through loop passes inside LoopStandardAnalysisResults.

If only LICM benefits from it, I'd consider creating a BFI instance for the duration of the LICM pass

Aug 22 2020, 11:17 PM · Restricted Project, Restricted Project
wenlei added inline comments to D86332: [SampleFDO] Enhance profile remapping support for inline instance.
Aug 22 2020, 8:47 PM · Restricted Project

Aug 20 2020

wenlei added a comment to D77925: Revert "[TLI] Per-function fveclib for math library used for vectorization".

I rather have a non closed list of veclib: if you just have a map keyed by string instead of an enum it should just work out?

The veclib type is also tied to the accepted values for -fveclib, which is a list of supported lib, anything outside of the list is rejected with error. If you want to allow arbitrary strings as key, it's inconsistent with the restricted/closed values for -fveclib. So basically there's no openness from clang/llvm tool, while there was some openness through the libraries. I think by introducing a "Custom" veclib type, we can solve the impedance mismatch there. And in the XLA case, conceptually it's an indeed a custom veclib, right? Functionality-wise, Custom should just work for XLA usage too.

Aug 20 2020, 9:49 AM · Restricted Project, Restricted Project

Aug 19 2020

wenlei added a reviewer for D86190: [LICM][Coroutine] Don't sink stores from loops with coro.suspend instructions.: lxfind.
Aug 19 2020, 3:47 PM · Restricted Project

Aug 18 2020

wenlei accepted D85812: [coroutine] should disable inline before calling coro split.

LGTM, thanks for the fix.

Aug 18 2020, 7:21 AM · Restricted Project

Aug 17 2020

wenlei added inline comments to D85812: [coroutine] should disable inline before calling coro split.
Aug 17 2020, 9:14 PM · Restricted Project
wenlei accepted D84715: [FIX] Avoid creating BFI when emitting remarks for dead functions.

LTGM too. Thanks for the fix!

Aug 17 2020, 7:01 PM · Restricted Project

Aug 15 2020

wenlei added a comment to D77925: Revert "[TLI] Per-function fveclib for math library used for vectorization".

I rather have a non closed list of veclib: if you just have a map keyed by string instead of an enum it should just work out?

Aug 15 2020, 10:56 PM · Restricted Project, Restricted Project
wenlei added a comment to D77925: Revert "[TLI] Per-function fveclib for math library used for vectorization".

@mehdi_amini @tejohnson When can we re-land this (with tweaks)? I'm under the impression that a test case demonstrating the 3rd party usage will be added very soon after this revert, then we can tweak the original patch to accommodate that usage, and re-land asap. Or am I missing something there? I'd like to get this unblocked asap. Currently we have to keep this as a private patch on our end which is a bit cumbersome, and I think this one can be useful for others too. Thanks..

@bkramer can you work with Wenlei on this (original patch is D77632).

@wenlei, in the meantime you can see the use case here:
https://github.com/tensorflow/tensorflow/blob/master/tensorflow/compiler/xla/service/cpu/compiler_functor.cc#L198
for revising the patch.

Aug 15 2020, 9:50 PM · Restricted Project, Restricted Project
wenlei committed rG577e58bcc754: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks (authored by wenlei).
[InlineAdvisor] New inliner advisor to replay inlining from optimization remarks
Aug 15 2020, 8:18 PM
wenlei added a comment to D85812: [coroutine] should disable inline before calling coro split.

Looking at sample profile loader inline and your test case (either always inline or with sample profile), I'm guessing your change in the general llvm::getInlineCost was for sample profile loader inline, not CGSCC inline? If that's the case, it might be better to find a way to do this targeting just always inline and sample profile loader inline, then assert this shouldn't happen for CGSCC inline (would be a pass ordering bug).

Aug 15 2020, 6:15 PM · Restricted Project
wenlei updated subscribers of D85812: [coroutine] should disable inline before calling coro split.
Aug 15 2020, 6:11 PM · Restricted Project
wenlei added a comment to D85812: [coroutine] should disable inline before calling coro split.

You'd also want to prevent inlining presplit coroutine in sample profile loader (SampleProfileLoader::inlineHotFunctions), which is before CGSCC pipeline and does top-down inlining now.

Aug 15 2020, 6:03 PM · Restricted Project
wenlei updated subscribers of D85812: [coroutine] should disable inline before calling coro split.
Aug 15 2020, 5:51 PM · Restricted Project

Aug 12 2020

wenlei added a comment to D85628: [HotColdSplitting] Add command line options for supplying cold function names via user input..

For my 2nd question, this patch is what I think how this should be done: https://reviews.llvm.org/D79485

Aug 12 2020, 10:02 PM · Restricted Project
wenlei updated subscribers of D79485: [BPI] Improve static heuristics for "cold" paths..
Aug 12 2020, 9:58 PM · Restricted Project
wenlei added a comment to D85628: [HotColdSplitting] Add command line options for supplying cold function names via user input..

I have two questions for this patch:

Aug 12 2020, 9:48 PM · Restricted Project
wenlei added a reviewer for D85810: [clang] Pass-through remarks options to linker: bruno.
Aug 12 2020, 6:59 PM · Restricted Project

Aug 11 2020

wenlei added a comment to D85745: emit int32 for each version part.

Looks like some CV/COFF tests need to be updated too?

Aug 11 2020, 6:10 PM · Restricted Project

Aug 10 2020

wenlei accepted D85664: [SampleFDO] Stop letting findCalleeFunctionSamples return unrelated profiles for invoke instructions .

Thanks. LGTM.

Aug 10 2020, 10:44 AM · Restricted Project
wenlei added inline comments to D85664: [SampleFDO] Stop letting findCalleeFunctionSamples return unrelated profiles for invoke instructions .
Aug 10 2020, 10:16 AM · Restricted Project

Aug 8 2020

wenlei committed rG540489de6816: [llvm-profdata] Implement llvm-profdata overlap for sample profiles (authored by weihe).
[llvm-profdata] Implement llvm-profdata overlap for sample profiles
Aug 8 2020, 6:07 PM
wenlei closed D83852: [llvm-profdata] Implement llvm-profdata overlap for sample profiles.
Aug 8 2020, 6:02 PM · Restricted Project
wenlei accepted D83852: [llvm-profdata] Implement llvm-profdata overlap for sample profiles.

Thank you for working on this during internship, @weihe! The extra tweaks here on top of our internal version look good to me as well. The test failure doesn't seem related, I will rebase and land this on your behalf.

Aug 8 2020, 5:46 PM · Restricted Project

Jul 30 2020

wenlei added a comment to D84997: [AutoFDO] Avoid merging inlinee samples multiple times.
In D84997#2186553, @wmi wrote:

The multiple inline instance copies generated from loop unrolling or jump threading should have their own FunctionSamples objects. Why merging multiple inline instance to the outline profile will trigger assertion? Could you have a testcase for it?

It's a slightly difference case. In the training build, we didn't unroll, so there's a single inline site in the loop, but in the PGO optimizing build, the loop got unrolled with call site duplicated. And the duplicated call sites all point to the same FunctionSamples, in which case we should merge them only once. Agree that we should add a test case.

Thanks. So it is a combined version of normal PGO and sampleFDO, or the PGO here means the instrument hook for sampleFDO?

Jul 30 2020, 9:16 PM · Restricted Project
wenlei added a comment to D84997: [AutoFDO] Avoid merging inlinee samples multiple times.
In D84997#2186518, @wmi wrote:

The multiple inline instance copies generated from loop unrolling or jump threading should have their own FunctionSamples objects. Why merging multiple inline instance to the outline profile will trigger assertion? Could you have a testcase for it?

Jul 30 2020, 8:35 PM · Restricted Project
wenlei updated subscribers of D84994: [SampleFDO] Fix a crash when the sample profile uses md5 and -sample-profile-merge-inlinee is enabled.

Just realized this is fixing the same issue as @hoyFB's D84997 - we ran into it immediately after -sample-profile-merge-inlinee is turned on too.

Jul 30 2020, 8:21 PM · Restricted Project
wenlei accepted D84997: [AutoFDO] Avoid merging inlinee samples multiple times.

LGTM. Thanks!

Jul 30 2020, 8:15 PM · Restricted Project
wenlei added a comment to D84715: [FIX] Avoid creating BFI when emitting remarks for dead functions.

I'm guessing these are not the only passes that will choke with empty/dead functions. If this is because ORE is being used for a special case, then we might want to fix the special case on the user side.

If we want passes to absorb such empty functions, then there needs to be consistency for all passes. I don't think passes need to absorb such cases though, as this is really a "limbo state" of functions when being removed (a normal empty function should still have entry block and ret instruction).

Agree. The proposed checks are only placed along the particular call path that leads to the failure. If we are going cover all the analyses/passes, the overall effort could be substantial.

If the big assumption is that there shouldn't be an empty body function to deal with, we can probably change how ORE is created at the beginning. Instead of creating a heavy-weight version of ORE (with all dependent analyses invoked), using the light-weight version OptimizationRemarkEmitter ORE(F, nullptr) at LTO.cpp:801.

Can we bail out for dead/empty functions right there before launching ORE on them?

Jul 30 2020, 8:06 PM · Restricted Project

Jul 28 2020

wenlei added a comment to D84569: [ELF] --reproduce should include lto sample profile.

Committed on behalf of @christylee while she tries to recover commit access.

Jul 28 2020, 9:52 AM · Restricted Project
wenlei committed rGbd4757cc4e40: [ELF] --reproduce should include lto sample profile (authored by christylee).
[ELF] --reproduce should include lto sample profile
Jul 28 2020, 9:48 AM
wenlei closed D84569: [ELF] --reproduce should include lto sample profile.
Jul 28 2020, 9:48 AM · Restricted Project

Jul 27 2020

wenlei added a comment to D84715: [FIX] Avoid creating BFI when emitting remarks for dead functions.

I'm guessing these are not the only passes that will choke with empty/dead functions. If this is because ORE is being used for a special case, then we might want to fix the special case on the user side.

Jul 27 2020, 9:51 PM · Restricted Project

Jul 19 2020

wenlei committed rGd41d952be977: Revert "[InlineAdvisor] New inliner advisor to replay inlining from… (authored by wenlei).
Revert "[InlineAdvisor] New inliner advisor to replay inlining from…
Jul 19 2020, 8:50 AM
wenlei added a reverting change for rG2d6ecfa168c2: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks: rGd41d952be977: Revert "[InlineAdvisor] New inliner advisor to replay inlining from….
Jul 19 2020, 8:50 AM
wenlei added a comment to rG2d6ecfa168c2: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

This is a resubmit for https://reviews.llvm.org/D83743.

Jul 19 2020, 8:43 AM
wenlei abandoned D84086: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
Jul 19 2020, 8:39 AM · Restricted Project
wenlei committed rG2d6ecfa168c2: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks (authored by wenlei).
[InlineAdvisor] New inliner advisor to replay inlining from optimization remarks
Jul 19 2020, 8:37 AM

Jul 18 2020

wenlei updated the diff for D84086: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

test

Jul 18 2020, 10:08 PM · Restricted Project

Jul 17 2020

wenlei created D84086: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
Jul 17 2020, 8:55 PM · Restricted Project
wenlei abandoned D84080: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
Jul 17 2020, 8:54 PM · Restricted Project
wenlei updated the diff for D84080: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

test

Jul 17 2020, 7:13 PM · Restricted Project
wenlei updated the diff for D84080: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

test

Jul 17 2020, 6:35 PM · Restricted Project
Herald added a project to D84080: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks: Restricted Project.
Jul 17 2020, 5:06 PM · Restricted Project
wenlei added a comment to D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

The inline-replay.ll test that this added appears to be broken (as was noted by the pre-merge check https://reviews.llvm.org/harbormaster/unit/view/120108/) and is failing in our build.

Jul 17 2020, 2:59 PM · Restricted Project
wenlei committed rG029946b11268: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks (authored by wenlei).
[InlineAdvisor] New inliner advisor to replay inlining from optimization remarks
Jul 17 2020, 1:31 PM
wenlei closed D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
Jul 17 2020, 1:31 PM · Restricted Project
wenlei added reviewers for D83852: [llvm-profdata] Implement llvm-profdata overlap for sample profiles: davidxl, wmi.
Jul 17 2020, 11:56 AM · Restricted Project

Jul 16 2020

wenlei added a comment to D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

lgtm - a small nit, could you also add in the description the current scope (the top down, SampleProfile case) - for clarity. Thanks!

Jul 16 2020, 12:37 PM · Restricted Project
wenlei updated the summary of D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
Jul 16 2020, 12:36 PM · Restricted Project
wenlei added inline comments to D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
Jul 16 2020, 11:45 AM · Restricted Project
wenlei updated the diff for D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

address feedbacks.

Jul 16 2020, 11:43 AM · Restricted Project
wenlei added inline comments to D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
Jul 16 2020, 11:43 AM · Restricted Project
wenlei updated the diff for D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

rebase

Jul 16 2020, 9:15 AM · Restricted Project

Jul 15 2020

wenlei accepted D83849: [TableGen] Change std::vector to SmallVector.

LGTM. Thanks!

Jul 15 2020, 10:28 PM · Restricted Project
wenlei added a comment to D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.
In D83743#2153532, @wmi wrote:

Thanks for the explanation.

We could also replay CGSCC inline of one build in another. The way it's implemented would work for those scenarios.

When you replay CGSCC, the inline decision in replay file should have order? If it is unordered, I cannot see how it can replay CGSCC inline from another build exactly.

Jul 15 2020, 10:03 PM · Restricted Project
wenlei added a comment to D83849: [TableGen] Change std::vector to SmallVector.

Reserve space in a vector to reduce the number of mallocs. The size of VTList is usually 1, but often 6 or 7.

Jul 15 2020, 9:14 PM · Restricted Project

Jul 14 2020

wenlei updated the diff for D83743: [InlineAdvisor] New inliner advisor to replay inlining from optimization remarks.

Address feedback, update comment.

Jul 14 2020, 6:24 PM · Restricted Project