Page MenuHomePhabricator

wmi (Wei Mi)
User

Projects

User does not belong to any projects.

User Details

User Since
Feb 20 2015, 10:57 AM (383 w, 5 d)

Recent Activity

Sep 24 2021

wmi committed rG80865f7579d6: Add "REQUIRES: zlib" in forward-compatible.test since it handles compressed… (authored by wmi).
Add "REQUIRES: zlib" in forward-compatible.test since it handles compressed…
Sep 24 2021, 3:37 PM
wmi committed rGe8b376547b4b: Fixed a bug in https://reviews.llvm.org/rG8eb617d719bdc6a4ed7773925d2421b9bbdd4… (authored by wmi).
Fixed a bug in https://reviews.llvm.org/rG8eb617d719bdc6a4ed7773925d2421b9bbdd4…
Sep 24 2021, 3:24 PM
wmi accepted D110374: [ThinLTO] Update combined index for SamplePGO indirect calls to locals.

LGTM.

Sep 24 2021, 9:00 AM · Restricted Project

Sep 23 2021

wmi added inline comments to D110374: [ThinLTO] Update combined index for SamplePGO indirect calls to locals.
Sep 23 2021, 9:12 PM · Restricted Project
wmi accepted D110296: [ThinLTO] Don't emit original GUID for locals to distributed indexes.
Sep 23 2021, 9:05 AM · Restricted Project
wmi added a comment to D109860: profi - a flow-based profile inference algorithm: Part I (out of 3).

Thanks Sergey. That looks like a good work to improve the consistency of the profile. Have you checked whether the new algorithm can infer the missing parts based on equivalence relationship on the CFG? If it can, could you add a test for it?

Do you mind to elaborate more on this? What kind of a test would you like to see? (i may not be familiar with the terminology)

Sep 23 2021, 9:02 AM · Restricted Project

Sep 22 2021

wmi added inline comments to D110296: [ThinLTO] Don't emit original GUID for locals to distributed indexes.
Sep 22 2021, 8:00 PM · Restricted Project

Sep 20 2021

wmi added a comment to D109860: profi - a flow-based profile inference algorithm: Part I (out of 3).

Thanks Sergey. That looks like a good work to improve the consistency of the profile. Have you checked whether the new algorithm can infer the missing parts based on equivalence relationship on the CFG? If it can, could you add a test for it?

Sep 20 2021, 10:37 AM · Restricted Project

Sep 10 2021

wmi accepted D109531: [CSSPGO] Enable pseudo probe instrumentation in O0 mode..

LGTM.

Sep 10 2021, 10:13 AM · Restricted Project, Restricted Project
wmi added a comment to D109531: [CSSPGO] Enable pseudo probe instrumentation in O0 mode..

The change makes sense given instr PGO also happens for O0. But practically, if a file is being built with O0, do we care about its profile given we're not really optimizing it anyways? Functions from O0 modules are not supposed to be inlined into O1+ modules either.

We probably don't care about performance for O0 build. The change is for consistency, also makes the compiler happy which otherwise will complain about "Pseudo-probe-based profile requires SampleProfileProbePass" for O0 modules that don't have probes.

The complain message is emitted in SampleProfileLoader::doInitialization. llvm will not run SampleProfileLoader pass for O0 module. Why there is the complain?

Good question. It could happen in lto postlink which by default optimizes in -O2 mode. More specifically, with the following command, both cc1 and lld will run in default mode, which is -O0 for cc1 and -O2 for lld.

clang -flto 1.cpp -v -fuse-ld=lld
Sep 10 2021, 9:24 AM · Restricted Project, Restricted Project

Sep 9 2021

wmi added a comment to D109531: [CSSPGO] Enable pseudo probe instrumentation in O0 mode..

The change makes sense given instr PGO also happens for O0. But practically, if a file is being built with O0, do we care about its profile given we're not really optimizing it anyways? Functions from O0 modules are not supposed to be inlined into O1+ modules either.

We probably don't care about performance for O0 build. The change is for consistency, also makes the compiler happy which otherwise will complain about "Pseudo-probe-based profile requires SampleProfileProbePass" for O0 modules that don't have probes.

Sep 9 2021, 8:14 PM · Restricted Project, Restricted Project

Sep 7 2021

wmi committed rG8eb617d719bd: [SampleFDO] Allow forward compatibility when adding a new section for extbinary (authored by wmi).
[SampleFDO] Allow forward compatibility when adding a new section for extbinary
Sep 7 2021, 7:40 PM
wmi closed D109398: [SampleFDO] Allow forward compatibility when adding a new section for extbinary format. .
Sep 7 2021, 7:40 PM · Restricted Project
wmi requested review of D109398: [SampleFDO] Allow forward compatibility when adding a new section for extbinary format. .
Sep 7 2021, 3:25 PM · Restricted Project

Sep 1 2021

wmi accepted D109036: [CSSPGO] Sort function offset table to speed up profile loading..

LGTM.

Sep 1 2021, 11:37 AM · Restricted Project
wmi accepted D108342: [CSSPGO] Enable loading MD5 CS profile..

LGTM.

Sep 1 2021, 8:56 AM · Restricted Project

Aug 31 2021

wmi added inline comments to D108342: [CSSPGO] Enable loading MD5 CS profile..
Aug 31 2021, 10:49 AM · Restricted Project
wmi added inline comments to D108342: [CSSPGO] Enable loading MD5 CS profile..
Aug 31 2021, 8:58 AM · Restricted Project

Aug 25 2021

wmi accepted D108437: [CSSPGO] split context string III - llvm-profgen changes.

LGTM.

Aug 25 2021, 11:13 AM · Restricted Project
wmi accepted D108350: [CSSPGO] Use probe inline tree to track zero size fully optimized context for pre-inliner.

LGTM.

Aug 25 2021, 8:45 AM · Restricted Project

Aug 23 2021

wmi accepted D108435: [CSSPGO] split context string II - reader/writer changes.

LGTM.

Aug 23 2021, 8:36 PM · Restricted Project
wmi accepted D108433: [CSSPGO] split context string - compiler changes.

LGTM, thanks.

Aug 23 2021, 8:35 PM · Restricted Project
wmi added inline comments to D108437: [CSSPGO] split context string III - llvm-profgen changes.
Aug 23 2021, 2:18 PM · Restricted Project
wmi added inline comments to D108437: [CSSPGO] split context string III - llvm-profgen changes.
Aug 23 2021, 12:10 PM · Restricted Project
wmi added inline comments to D108437: [CSSPGO] split context string III - llvm-profgen changes.
Aug 23 2021, 12:09 PM · Restricted Project
wmi added inline comments to D108435: [CSSPGO] split context string II - reader/writer changes.
Aug 23 2021, 9:50 AM · Restricted Project

Aug 20 2021

wmi added inline comments to D108435: [CSSPGO] split context string II - reader/writer changes.
Aug 20 2021, 3:09 PM · Restricted Project
wmi added inline comments to D108433: [CSSPGO] split context string - compiler changes.
Aug 20 2021, 2:06 PM · Restricted Project
wmi added inline comments to D108433: [CSSPGO] split context string - compiler changes.
Aug 20 2021, 11:18 AM · Restricted Project

Aug 19 2021

wmi added a comment to D107299: [CSSPGO] Split context string to deduplicate function name used in the context..

Thanks Hongtao. The compile time result looks great. The patch is large. Is it possible to split it into smaller ones, SampleContext related change and SecCSNameTable related change for example?

Thanks for the suggestion. The separation may not be complete. The point is that the previous StringRef context is used in both reader/writer, and now we are changing it to be an ArrayRef which relies on an underlying array buffer (i.e., the CSName the reader builds up). Separating SecCSNameTable from this patch will still require the reader to build the underlying CSName table to bridge full string contexts in profile to ArrayRef contexts used elsewhere. Does this sound good to you?

Aug 19 2021, 2:28 PM · Restricted Project
wmi added a comment to D107299: [CSSPGO] Split context string to deduplicate function name used in the context..

Thanks Hongtao. The compile time result looks great. The patch is large. Is it possible to split it into smaller ones, SampleContext related change and SecCSNameTable related change for example?

Aug 19 2021, 9:59 AM · Restricted Project
wmi added a comment to D106450: [InstCombine] Fold (gep (oneuse(gep Ptr, Idx0)), Idx1) -> (gep Ptr, (add Idx0, Idx1)) (PR51069).

Got a testcase and put it here: https://bugs.llvm.org/show_bug.cgi?id=51540

Aug 19 2021, 9:13 AM · Restricted Project

Aug 18 2021

wmi accepted D107878: [SampleFDO] Flow Sensitive Sample FDO (FSAFDO) profile loader.

LGTM. Thanks.

Aug 18 2021, 12:58 PM · Restricted Project, Restricted Project

Aug 17 2021

wmi added inline comments to D107878: [SampleFDO] Flow Sensitive Sample FDO (FSAFDO) profile loader.
Aug 17 2021, 5:52 PM · Restricted Project, Restricted Project
wmi accepted D108180: [CSSPGO] Track and use context-sensitive post-optimization function size to drive global pre-inliner in llvm-profgen.

LGTM.

Aug 17 2021, 12:09 PM · Restricted Project
wmi added inline comments to D107878: [SampleFDO] Flow Sensitive Sample FDO (FSAFDO) profile loader.
Aug 17 2021, 10:47 AM · Restricted Project, Restricted Project
wmi added inline comments to D108180: [CSSPGO] Track and use context-sensitive post-optimization function size to drive global pre-inliner in llvm-profgen.
Aug 17 2021, 9:38 AM · Restricted Project

Aug 16 2021

wmi accepted D108142: [SamplePGO] Fixing a memory issue when creating profiles on-demand.

LGTM.

Aug 16 2021, 2:24 PM · Restricted Project
wmi added inline comments to D108142: [SamplePGO] Fixing a memory issue when creating profiles on-demand.
Aug 16 2021, 12:24 PM · Restricted Project

Aug 12 2021

wmi added inline comments to D107878: [SampleFDO] Flow Sensitive Sample FDO (FSAFDO) profile loader.
Aug 12 2021, 2:56 PM · Restricted Project, Restricted Project
wmi added inline comments to D107878: [SampleFDO] Flow Sensitive Sample FDO (FSAFDO) profile loader.
Aug 12 2021, 12:40 PM · Restricted Project, Restricted Project

Aug 11 2021

wmi accepted D107876: [CSSPGO] Allow the use of debug-info-for-profiling and pseudo-probe-for-profiling together.

Could you remind me the discriminator difference between debug info based AFDO and pseudo probe based AFDO? I forgot it. I am sure it is described in some patch but I havn't found it. Give me a pointer is good enough.

In short, unlike block pseudo probe, a callsite probe relies on the calliste discriminator to be tracked. This has a couple usage: 1. to track direct call and indirect callsite 2. to model inline contexts with a list of callsite probes.

A probe discriminator is treated as empty or null for AutoFDO since it has lowest three bits as "111".

Aug 11 2021, 10:48 PM · Restricted Project, Restricted Project
wmi added a comment to D107876: [CSSPGO] Allow the use of debug-info-for-profiling and pseudo-probe-for-profiling together.

Could you remind me the discriminator difference between debug info based AFDO and pseudo probe based AFDO? I forgot it. I am sure it is described in some patch but I havn't found it. Give me a pointer is good enough.

Aug 11 2021, 9:11 AM · Restricted Project, Restricted Project
wmi accepted D104579: [SampleFDO] Add two passes of MIRAddFSDiscriminatorsPass.
Aug 11 2021, 8:55 AM · Restricted Project

Aug 10 2021

wmi added a comment to D104579: [SampleFDO] Add two passes of MIRAddFSDiscriminatorsPass.

This extra pass turns to be positive in performance -- I'm seeing additional 0.6% - 0.8% improvement.

Aug 10 2021, 11:54 AM · Restricted Project

Aug 5 2021

wmi added a comment to D107173: [CSSPGO] Introduce MD5-based context-sensitive profile.

I tried the idea of using vector<Callsite> (Callsite is a struct containing function name and LineLocation) in SampleContext and changed the ProfileMap to be map<SampleContext, FunctionSamples> in the last two days. I believed the idea should work but I found the required change was much larger than I thought since ProfileMap has been used in many interfaces. I am worried that for the moment I cannot devote enough time to finish the whole change, so now I tend to use Hongtao's md5 patch and leave the further improvement to the future since the current patch has already significantly improved the build process and unblocked our experiment. We can always come back to revisit it. What is your opinion?

Thanks for the heads-up, Wei. Sounds good to stick with this patch for now, if you need more time to complete your idea which I think will get us a bigger win and be more aligned with the non-CS implementation. Will you continue working on that?

Aug 5 2021, 12:29 PM · Restricted Project

Aug 4 2021

wmi added a comment to D107173: [CSSPGO] Introduce MD5-based context-sensitive profile.

I tried the idea

Aug 4 2021, 9:35 PM · Restricted Project

Aug 3 2021

wmi added a comment to D107173: [CSSPGO] Introduce MD5-based context-sensitive profile.

Thanks for working on this.

Decomposing context strings into two levels is natural choice for deduplication. However, that means we would need to reconstruct strings in profile reader. So it's more like space vs speed thing. If compiler speed is the same, then storing decomposed strings in file and reconstruct them on profile loading might be a better choice as it's more compact and also more consistent with non-CS profile. Perhaps would be good to measure both.. Right now with md5 profile, plus some context trimming through cold threshold or pre-inliner tuning, we hope to bring e2e build time on par with AutoFDO for large workloads.

Right, D107173 in its own can already let me build our search benchmark in distributed way which I cannot do that before because of OOM issue, and that is really helpful. https://reviews.llvm.org/D107299 will still run into OOM in distributed build because we will still have the full context strings after profile reading, and that consumes excessive memory. For D107299, I believe that issue can also be solved after I add md5 support to it.

Like wenlei said, decomposing context strings can give us more compact profile (deduplicating: 418M vs md5: 1.8G for the profile collected from our search test), but reconstructing takes more time (2x). I am thinking about whether reconstructing the context string is necessary since the context represented in ContextTrieNode is also decomposed into multiple levels?

By the way, the context reconstructing time may also be reduced if D107299 can be coupled with md5.

@wmi Thanks for sharing your change and the numbers! And thanks for trying my implementation. Interested have you seen build time improvement with it?

I didn't compare it with the case without md5 because I cannot build it successfully in our build system without md5. I tried building it locally but it took one week to finish.

I tried dumping the profile to text format and the md5 version took about a half time to finish compared with the case without md5, so that is an indication we should see build time improvement with it.

Curious were you trying probe-based CS profile or line-based CS profile?

Aug 3 2021, 10:02 AM · Restricted Project

Aug 2 2021

wmi added a comment to D107173: [CSSPGO] Introduce MD5-based context-sensitive profile.

Thanks for working on this.

Decomposing context strings into two levels is natural choice for deduplication. However, that means we would need to reconstruct strings in profile reader. So it's more like space vs speed thing. If compiler speed is the same, then storing decomposed strings in file and reconstruct them on profile loading might be a better choice as it's more compact and also more consistent with non-CS profile. Perhaps would be good to measure both.. Right now with md5 profile, plus some context trimming through cold threshold or pre-inliner tuning, we hope to bring e2e build time on par with AutoFDO for large workloads.

Right, D107173 in its own can already let me build our search benchmark in distributed way which I cannot do that before because of OOM issue, and that is really helpful. https://reviews.llvm.org/D107299 will still run into OOM in distributed build because we will still have the full context strings after profile reading, and that consumes excessive memory. For D107299, I believe that issue can also be solved after I add md5 support to it.

Like wenlei said, decomposing context strings can give us more compact profile (deduplicating: 418M vs md5: 1.8G for the profile collected from our search test), but reconstructing takes more time (2x). I am thinking about whether reconstructing the context string is necessary since the context represented in ContextTrieNode is also decomposed into multiple levels?

By the way, the context reconstructing time may also be reduced if D107299 can be coupled with md5.

@wmi Thanks for sharing your change and the numbers! And thanks for trying my implementation. Interested have you seen build time improvement with it?

Aug 2 2021, 2:03 PM · Restricted Project
wmi added a comment to D107173: [CSSPGO] Introduce MD5-based context-sensitive profile.

Thanks for working on this.

Decomposing context strings into two levels is natural choice for deduplication. However, that means we would need to reconstruct strings in profile reader. So it's more like space vs speed thing. If compiler speed is the same, then storing decomposed strings in file and reconstruct them on profile loading might be a better choice as it's more compact and also more consistent with non-CS profile. Perhaps would be good to measure both.. Right now with md5 profile, plus some context trimming through cold threshold or pre-inliner tuning, we hope to bring e2e build time on par with AutoFDO for large workloads.

Aug 2 2021, 11:25 AM · Restricted Project
wmi added a comment to D107173: [CSSPGO] Introduce MD5-based context-sensitive profile.

Thanks for the work to reduce the CS profile size! It is something we really need.

I am also trying it in a slightly different way.
For "[main:3.1 @ _Z5funcBi]:120:19" in your example, I split the context string into multiple tuples of {string, line, discriminator}:
{main, 3, 1} {_Z5funcBi, 0, 0}, and for each name in the tuple, we will only save the index to the name table. So we will not have new entry in the name table for different contexts.

In this way, we won't have any increase in the name table section compared with non-CS profile even when we uses string based name. We do need to have a new section called CSNameTable to store the tuples. When we read the section, we will recontruct the context string from the tuple and the rest of the profile handling will have no change.

In this way, on top of it we can also apply md5 to the nametable using existing md5 mechanism, to further compress the name table section.

I havn't finished the implemention, but I am close to it. It will be good to discuss how we converge the effort here.

@wmi good to know you are also working on improving the efficiency. Thanks!
The string split design sounds interesting. I'm curious to know more about it, like how context strings are split in the profile, and how they are constructed and represented in the compiler.

Aug 2 2021, 11:03 AM · Restricted Project
wmi requested review of D107299: [CSSPGO] Split context string to deduplicate function name used in the context..
Aug 2 2021, 10:58 AM · Restricted Project

Jul 30 2021

wmi added a comment to D107173: [CSSPGO] Introduce MD5-based context-sensitive profile.

Thanks for the work to reduce the CS profile size! It is something we really need.

Jul 30 2021, 12:28 PM · Restricted Project

Jul 28 2021

wmi added inline comments to D106930: [ThinLTO] Disallow importing for functions with indir branch to block address.
Jul 28 2021, 9:19 AM · Restricted Project

Jul 26 2021

wmi accepted D106588: [CSSPGO] Tweak ICP threshold in top-down inliner.

LGTM.

Jul 26 2021, 10:46 AM · Restricted Project

Jul 22 2021

wmi accepted D106599: [CSSPGO] Fix a typo in SampleContextTracker.

Good catch.

Jul 22 2021, 2:55 PM · Restricted Project

Jul 20 2021

wmi added a comment to rG9c4c2f24725e: [SimplifyCFG] Tail-merging all blocks with `ret` terminator.

We saw a performance regression caused by this change and filed a bug here: https://bugs.llvm.org/show_bug.cgi?id=51149

Jul 20 2021, 5:12 PM

Jul 9 2021

wmi accepted D105729: [AFDO] Merge function attributes after inlining.

LGTM.

Jul 9 2021, 1:46 PM · Restricted Project

Jul 2 2021

wmi added a comment to D105248: [CSSPGO] Do not import pseudo probe desc in thinLTO.

-funique-internal-linkage-names is not enabled by default. Will -fpseudo-probe-for-profiling implicitly enable -funique-internal-linkage-names?

Jul 2 2021, 10:02 AM · Restricted Project

Jun 25 2021

wmi added a comment to D104926: [CSSPGO] Switches to disable pre-link inlining.

Thanks Hongtao, I will try it.

Jun 25 2021, 9:51 AM · Restricted Project

Jun 22 2021

wmi accepted D104546: [CSSPGO][llvm-profgen] Handle return to external transition..
Jun 22 2021, 2:33 PM · Restricted Project
wmi added inline comments to D104546: [CSSPGO][llvm-profgen] Handle return to external transition..
Jun 22 2021, 12:08 PM · Restricted Project
wmi added inline comments to D104546: [CSSPGO][llvm-profgen] Handle return to external transition..
Jun 22 2021, 8:49 AM · Restricted Project

Jun 21 2021

wmi added inline comments to D104584: [SampleFDO] Make FSDiscriminator flag part of function parameters.
Jun 21 2021, 8:38 AM · Restricted Project

Jun 16 2021

wmi accepted D104129: [CSSPGO] Report zero-count probe in profile instead of dangling probes..

In "6. Be consistent with non-CS dwarf line number based profile", if you can mention how the compiler treat missing line number and missing pseudo in the description as a record, that will be helpful.

Jun 16 2021, 11:19 AM · Restricted Project
wmi added a comment to D104129: [CSSPGO] Report zero-count probe in profile instead of dangling probes..

Be consistent with dwarf line number based profile.

This is not accurate, see earlier comments. What i meant was not about "non-probe profile -> dwarf line number profile".

Otherwise lgtm. Thanks.

It should be consistent with non-CS dwarf-based profile. CS dwarf-based profile doesn't have either zero or dangling reported, likely needs a fix. Is that what you meant?

Non-cs dwarf-based profile normally doesn't have zero counts in the profile. Any count below a certain threshold is omitted in the profile to save size. This is the default behavior for our production usage (I believe for google as well). This is why the current change that needs explicit zero is not consistent with afdo.

Jun 16 2021, 10:57 AM · Restricted Project

Jun 3 2021

wmi accepted D103550: [SampleFDO] New hierarchical discriminator for FS SampleFDO (llvm-profdata part).

LGTM.

Jun 3 2021, 5:09 PM · Restricted Project
wmi accepted D103650: [CSSPGO][llvm-profgen] Make extended binary the default output format.

LGTM.

Jun 3 2021, 3:10 PM · Restricted Project

Jun 2 2021

wmi added inline comments to D103550: [SampleFDO] New hierarchical discriminator for FS SampleFDO (llvm-profdata part).
Jun 2 2021, 5:21 PM · Restricted Project

May 27 2021

wmi accepted D103041: [SampleFDO] New hierarchical discriminator for FS SampleFDO (ProfileData part).

LGTM.

May 27 2021, 9:45 AM · Restricted Project

May 25 2021

wmi accepted D103073: [LTT] Handle merged llvm.assume when dropping type tests.

LGTM.

May 25 2021, 3:35 PM · Restricted Project, Restricted Project
wmi added inline comments to D103041: [SampleFDO] New hierarchical discriminator for FS SampleFDO (ProfileData part).
May 25 2021, 11:03 AM · Restricted Project
wmi accepted D103071: [CSSPGO][llvm-profgen] Change default cold threshold for context merging.

LGTM.

May 25 2021, 9:42 AM · Restricted Project

May 18 2021

wmi accepted D102537: [CSSPGO] Overwrite branch weight annotated in previous pass..

LGTM.

May 18 2021, 1:27 PM · Restricted Project
wmi accepted D102246: [SampleFDO] New hierarchical discriminator for Flow Sensitive SampleFDO.

LGTM.

May 18 2021, 11:35 AM · Restricted Project

May 17 2021

wmi added a comment to D102537: [CSSPGO] Overwrite branch weight annotated in previous pass..

If we target AutoFDO as well, would be good to mention the impact on AutoFDO too.

I haven't tested for non-CS case. In theory it should help there but with a less-than-ideal counts quality, the change may not help in practice.

May 17 2021, 11:33 PM · Restricted Project

May 14 2021

wmi accepted D102429: [CSSPGO] Update pseudo probe distribution factor based on inline context..

LGTM.

May 14 2021, 6:30 PM · Restricted Project
wmi added inline comments to D102429: [CSSPGO] Update pseudo probe distribution factor based on inline context..
May 14 2021, 2:33 PM · Restricted Project
wmi added inline comments to D102429: [CSSPGO] Update pseudo probe distribution factor based on inline context..
May 14 2021, 9:30 AM · Restricted Project

May 13 2021

wmi added a comment to D102246: [SampleFDO] New hierarchical discriminator for Flow Sensitive SampleFDO.

Thanks for splitting the patches.

May 13 2021, 9:52 PM · Restricted Project

May 12 2021

wmi added a comment to D102246: [SampleFDO] New hierarchical discriminator for Flow Sensitive SampleFDO.

Can you split the changes under ProfileData into a separate patch?

May 12 2021, 7:49 AM · Restricted Project

May 7 2021

wmi added inline comments to D102007: [CSSPGO] Fix return value of getProbeWeight.
May 7 2021, 8:56 PM · Restricted Project

May 4 2021

wmi committed rG82956de05f9d: [SampleFDO] Fix a bug when appending function symbol into the Callees set of (authored by wmi).
[SampleFDO] Fix a bug when appending function symbol into the Callees set of
May 4 2021, 10:06 AM
wmi closed D101815: [SampleFDO] Fix a bug when appending function symbol into the Callees set of Root node in ProfiledCallGraph..
May 4 2021, 10:06 AM · Restricted Project

May 3 2021

wmi requested review of D101815: [SampleFDO] Fix a bug when appending function symbol into the Callees set of Root node in ProfiledCallGraph..
May 3 2021, 11:15 PM · Restricted Project

Apr 26 2021

wmi accepted D100075: [CSSPGO] Unblock optimizations with pseudo probe instrumentation part 2..

LGTM.

Apr 26 2021, 3:58 PM · Restricted Project

Apr 21 2021

wmi accepted D100528: [CSSPGO][llvm-profdata] Support trimming cold context when merging profiles.

LGTM, thanks.

Apr 21 2021, 5:23 PM · Restricted Project
wmi added inline comments to D100528: [CSSPGO][llvm-profdata] Support trimming cold context when merging profiles.
Apr 21 2021, 3:18 PM · Restricted Project

Apr 19 2021

wmi accepted D100332: [CSSPGO] Flip SkipPseudoOp to true for MIR APIs..

LGTM, thanks.

Apr 19 2021, 3:05 PM · Restricted Project

Apr 18 2021

wmi accepted D100334: [CSSPGO] Exclude pseudo probes from slot index.

LGTM.

Apr 18 2021, 10:11 PM · Restricted Project
wmi added inline comments to D100528: [CSSPGO][llvm-profdata] Support trimming cold context when merging profiles.
Apr 18 2021, 9:08 PM · Restricted Project
wmi added inline comments to D100332: [CSSPGO] Flip SkipPseudoOp to true for MIR APIs..
Apr 18 2021, 1:11 PM · Restricted Project

Apr 7 2021

wmi accepted D100002: [CSSPGO] Move pseudo probes to the beginning of a block to unblock SelectionDAG combine..

LGTM.

Apr 7 2021, 8:32 PM · Restricted Project
wmi added inline comments to D100002: [CSSPGO] Move pseudo probes to the beginning of a block to unblock SelectionDAG combine..
Apr 7 2021, 5:10 PM · Restricted Project

Mar 30 2021

wmi committed rGd535a05ca1a6: [ThinLTO] During module importing, close one source module before open (authored by wmi).
[ThinLTO] During module importing, close one source module before open
Mar 30 2021, 2:42 PM
wmi closed D99554: [ThinLTO] During module importing, close one source module before open another one for distributed mode..
Mar 30 2021, 2:42 PM · Restricted Project, Restricted Project
wmi added inline comments to D99554: [ThinLTO] During module importing, close one source module before open another one for distributed mode..
Mar 30 2021, 11:38 AM · Restricted Project, Restricted Project
wmi updated the diff for D99554: [ThinLTO] During module importing, close one source module before open another one for distributed mode..

Address Teresa's comment.

Mar 30 2021, 11:37 AM · Restricted Project, Restricted Project
wmi accepted D99550: keeping prof metadata for IndirectBrInst.

LGTM.

Mar 30 2021, 10:32 AM · Restricted Project
wmi accepted D99351: [CSSPGO] Top-down processing order based on full profile..

LGTM.

Mar 30 2021, 9:31 AM · Restricted Project