Page MenuHomePhabricator

yubing (Bing Yu)
User

Projects

User does not belong to any projects.

User Details

User Since
Aug 22 2019, 5:40 PM (149 w, 4 d)

Recent Activity

Apr 27 2022

yubing committed rG96d3be84438e: [InferAddressSpaces] Check if AS are the same in isNoopPtrIntCastPair (authored by wenju).
[InferAddressSpaces] Check if AS are the same in isNoopPtrIntCastPair
Apr 27 2022, 8:11 PM · Restricted Project, Restricted Project
yubing closed D123573: [InferAddressSpaces] Check if AS are the same in isNoopPtrIntCastPair.
Apr 27 2022, 8:11 PM · Restricted Project, Restricted Project

Dec 22 2021

yubing added inline comments to rG4639461531e7: [DAG][X86] Add TargetLowering::isSplatValueForTargetNode override.
Dec 22 2021, 9:20 PM

Dec 14 2021

yubing added inline comments to D115630: [CodeGen] Require use of Address::invalid() for invalid address.
Dec 14 2021, 4:27 AM · Restricted Project

Dec 8 2021

yubing added a reviewer for D115268: [SLP]Fix comparator for cmp instruction vectorization.: yubing.
Dec 8 2021, 12:58 AM · Restricted Project

Dec 7 2021

yubing added a comment to D115279: [OpenMP][AMDGPU] Switch host-device memory copy to asynchronous version.

It seems hsa_ext_amd.h should define hsa_amd_memory_lock and hsa_amd_memory_unlock according to https://github.com/RadeonOpenCompute/ROCR-Runtime/blob/master/src/inc/hsa_ext_amd.h

Dec 7 2021, 10:32 PM · Restricted Project
yubing added a comment to D115279: [OpenMP][AMDGPU] Switch host-device memory copy to asynchronous version.

hsa_amd_memory_lock and hsa_amd_memory_unlock are missing in hsa.cpp and hsa_ext_amd.h as well under openmp/libomptarget/plugins/amdgpu/dynamic_hsa

Dec 7 2021, 9:58 PM · Restricted Project

Nov 25 2021

yubing added a comment to D113888: [SDAG] Use UnknownSize for masked load/store MMO size.

Hi, @dmgreen With this patch, the llc command crashed. Would you take a look?
llc -mcpu=core-avx2 main.ll

Nov 25 2021, 12:01 AM · Restricted Project

Nov 9 2021

yubing added inline comments to D113035: [InstCombine] enhance vector bitwise select matching.
Nov 9 2021, 8:35 PM · Restricted Project

Oct 19 2021

yubing added a comment to D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.

huh... I guess the problem here was that we interpreted the register numbers as plain unsigned... And virtual registers always have bit 31 set, so I guess the bitset could indeed grow to unreasonable sizes.

  • Please try if DenseSet<Register> works too.
  • If you can please find a shorter more succinct title.

Then we should be good to land this.

SparseBitVector shouldn't be effected by bit 31 being set. It stores 128 bit chunks of bits in a linked list. Insertion does a linear scan forward or backward from the most recently accessed chunk to trying to find the chunk to insert in. Were we accessing it in some pathologically bad way that caused long linear scans?

If the issue is with the inserting function, then the title of this patch is misleading. The number of kill flags in is irrelevant. The place where the insert happens doesn't know how many kill flags exist. Only the later call to MRI->clearKillFlags(I) would know that.

Oct 19 2021, 7:02 PM · Restricted Project

Oct 18 2021

yubing committed rGf383c5331106: [MachineSink] Compile time improvement for large testcases which has many kill… (authored by yubing).
[MachineSink] Compile time improvement for large testcases which has many kill…
Oct 18 2021, 12:46 AM
yubing closed D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.
Oct 18 2021, 12:46 AM · Restricted Project
yubing retitled D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags from [MachineSink] Compile time improvement for large testcases which has many kill flags We did a experiment and observed dramatic decrease on compilation time which spent on clearing kill flags. to [MachineSink] Compile time improvement for large testcases which has many kill flags.
Oct 18 2021, 12:40 AM · Restricted Project

Oct 13 2021

yubing updated the diff for D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.

Replace DenseSet<unsigned> with DenseSet<Register>

Oct 13 2021, 7:31 PM · Restricted Project
yubing added inline comments to D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.
Oct 13 2021, 7:28 PM · Restricted Project

Oct 12 2021

yubing added a comment to D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.

How do you calculate the kill flags handling time?

I am using std::chrono::high_resolution_clock::now()

std::chrono::high_resolution_clock::time_point start = std::chrono::high_resolution_clock::now();
RegsToClearKillFlags.insert(MO.getReg()); // Remember to clear kill flags.
std::chrono::duration<double, std::milli> duration = std::chrono::high_resolution_clock::now() - start;

Do you also calculate the time of all the access to RegsToClearKillFlags?

Oct 12 2021, 11:25 PM · Restricted Project
yubing added a comment to D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.

How do you calculate the kill flags handling time?

I am using std::chrono::high_resolution_clock::now()

Oct 12 2021, 10:20 PM · Restricted Project
yubing updated the diff for D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.

use DenseSet<unsigned> instead, please ignore previous patch

Oct 12 2021, 7:40 PM · Restricted Project
yubing added reviewers for D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags: LuoYuanke, LiuChen3, pengfei, MatzeB.
Oct 12 2021, 7:32 PM · Restricted Project
yubing requested review of D111688: [MachineSink] Compile time improvement for large testcases which has many kill flags.
Oct 12 2021, 7:29 PM · Restricted Project

Sep 17 2021

yubing updated the diff for D109949: [x86][AMX] Support undef and zeroinitializer for x86_amx.

Solve the lit testcases fail

Sep 17 2021, 1:36 AM · Restricted Project

Sep 16 2021

yubing updated the summary of D109949: [x86][AMX] Support undef and zeroinitializer for x86_amx.
Sep 16 2021, 11:21 PM · Restricted Project
yubing requested review of D109949: [x86][AMX] Support undef and zeroinitializer for x86_amx.
Sep 16 2021, 11:17 PM · Restricted Project

Sep 8 2021

yubing abandoned D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps.
Sep 8 2021, 8:24 PM · Restricted Project
yubing accepted D109434: [X86][AVX] Add missing X86ISD::VBROADCAST(128->256) isel patterns for AVX1 targets.

LGTM

Sep 8 2021, 8:23 PM · Restricted Project
yubing added a comment to D109434: [X86][AVX] Add missing X86ISD::VBROADCAST(128->256) isel patterns for AVX1 targets.
Sep 8 2021, 8:22 PM · Restricted Project
yubing added inline comments to D109434: [X86][AVX] Add missing X86ISD::VBROADCAST(128->256) isel patterns for AVX1 targets.
Sep 8 2021, 8:22 PM · Restricted Project
yubing added inline comments to D109434: [X86][AVX] Add missing X86ISD::VBROADCAST(128->256) isel patterns for AVX1 targets.
Sep 8 2021, 6:09 AM · Restricted Project
yubing added inline comments to D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps.
Sep 8 2021, 6:07 AM · Restricted Project

Sep 7 2021

yubing added inline comments to D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps.
Sep 7 2021, 11:27 PM · Restricted Project
yubing added inline comments to D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps.
Sep 7 2021, 7:52 AM · Restricted Project
yubing added a comment to D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps.

Hi, @lebedev.ri .

  1. I think https://reviews.llvm.org/D105390 also has the same issue where you might create X86ISD::VBROADCAST(128->256) in AVX.
  2. Besides, Before https://reviews.llvm.org/D105390, we have nicer asm output for the testcase we added:
Sep 7 2021, 2:24 AM · Restricted Project
yubing added a reviewer for D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps: lebedev.ri.
Sep 7 2021, 2:17 AM · Restricted Project
yubing added reviewers for D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps: RKSimon, pengfei, LuoYuanke.
Sep 7 2021, 2:17 AM · Restricted Project
yubing requested review of D109348: [X86][AVX] Prohibit creating X86ISD::VBROADCAST(128->256) when it is AVX in combineConcatVectorOps.
Sep 7 2021, 2:16 AM · Restricted Project

Aug 18 2021

yubing committed rGffe58de39319: [X86] [AMX] Fix the test case failure caused by D107544. (authored by yubing).
[X86] [AMX] Fix the test case failure caused by D107544.
Aug 18 2021, 7:31 AM
yubing closed D108269: [X86] [AMX] Fix the test case failure caused by D107544..
Aug 18 2021, 7:31 AM · Restricted Project
yubing updated the diff for D108269: [X86] [AMX] Fix the test case failure caused by D107544..

Bring back getRowFromCol

Aug 18 2021, 1:36 AM · Restricted Project
yubing commandeered D108269: [X86] [AMX] Fix the test case failure caused by D107544..
Aug 18 2021, 1:34 AM · Restricted Project

Aug 17 2021

yubing committed rGbcec4ccd04ae: [X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast. (authored by yubing).
[X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast.
Aug 17 2021, 2:04 AM
yubing closed D107544: [X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast..
Aug 17 2021, 2:04 AM · Restricted Project
yubing added inline comments to D107544: [X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast..
Aug 17 2021, 12:07 AM · Restricted Project
yubing updated the diff for D107544: [X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast..

Address flanhn and yuanke's comments

Aug 17 2021, 12:07 AM · Restricted Project

Aug 16 2021

yubing added inline comments to D107544: [X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast..
Aug 16 2021, 7:57 AM · Restricted Project

Aug 5 2021

yubing added a reviewer for D107544: [X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast.: LuoYuanke.
Aug 5 2021, 2:30 AM · Restricted Project
yubing requested review of D107544: [X86] [AMX] Replace bitcast with specific AMX intrinsics with X86 specific cast..
Aug 5 2021, 2:30 AM · Restricted Project

Jun 15 2021

yubing added inline comments to D87981: [X86] AMX programming model..
Jun 15 2021, 7:00 PM · Restricted Project, Restricted Project

Jun 11 2021

yubing committed rG56d5c46b494d: [X86] Support __tile_stream_loadd intrinsic for new AMX interface (authored by yubing).
[X86] Support __tile_stream_loadd intrinsic for new AMX interface
Jun 11 2021, 2:34 AM
yubing closed D103784: [X86] Support __tile_stream_loadd intrinsic for new AMX interface.
Jun 11 2021, 2:33 AM · Restricted Project, Restricted Project

Jun 9 2021

yubing updated the diff for D103784: [X86] Support __tile_stream_loadd intrinsic for new AMX interface.

Rebase

Jun 9 2021, 12:57 AM · Restricted Project, Restricted Project
yubing updated the diff for D103784: [X86] Support __tile_stream_loadd intrinsic for new AMX interface.

Address yuanke's comments

Jun 9 2021, 12:21 AM · Restricted Project, Restricted Project

Jun 6 2021

yubing added reviewers for D103784: [X86] Support __tile_stream_loadd intrinsic for new AMX interface: pengfei, LuoYuanke, LiuChen3.
Jun 6 2021, 7:46 PM · Restricted Project, Restricted Project
yubing requested review of D103784: [X86] Support __tile_stream_loadd intrinsic for new AMX interface.
Jun 6 2021, 7:45 PM · Restricted Project, Restricted Project

Apr 11 2021

yubing committed rG747111ea71f5: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation (authored by yubing).
[X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation
Apr 11 2021, 10:58 PM
yubing closed D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.
Apr 11 2021, 10:58 PM · Restricted Project

Mar 31 2021

yubing updated the diff for D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.

Just a rebase

Mar 31 2021, 7:36 AM · Restricted Project
yubing updated the diff for D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.

Address buildfail in https://lab.llvm.org/buildbot#builders/77/builds/5053

Mar 31 2021, 12:59 AM · Restricted Project
yubing reopened D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.
Mar 31 2021, 12:58 AM · Restricted Project

Mar 30 2021

yubing added a reverting change for rG275df61f043c: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation: rG0c63b862c4d9: Revert "[X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar….
Mar 30 2021, 1:34 AM
yubing committed rG0c63b862c4d9: Revert "[X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar… (authored by yubing).
Revert "[X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar…
Mar 30 2021, 1:34 AM
yubing added a reverting change for D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation: rG0c63b862c4d9: Revert "[X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar….
Mar 30 2021, 1:34 AM · Restricted Project
yubing added a comment to D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.

It seems there is buildfail, I will revert it and fix bugs.
https://lab.llvm.org/buildbot#builders/77/builds/5053

Mar 30 2021, 1:31 AM · Restricted Project
yubing committed rG275df61f043c: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation (authored by yubing).
[X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation
Mar 30 2021, 1:21 AM
yubing closed D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.
Mar 30 2021, 1:21 AM · Restricted Project

Mar 29 2021

yubing updated the diff for D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.

Address Pengfei's comments

Mar 29 2021, 4:33 AM · Restricted Project

Mar 28 2021

yubing added inline comments to D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.
Mar 28 2021, 10:19 PM · Restricted Project

Mar 24 2021

yubing added reviewers for D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation: pengfei, LuoYuanke.
Mar 24 2021, 2:00 AM · Restricted Project
yubing requested review of D99244: [X86] Pass to transform tdpbsud&tdpbusd&tdpbuud intrinsics to scalar operation.
Mar 24 2021, 1:54 AM · Restricted Project

Mar 21 2021

yubing committed rG113f077f808f: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation. (authored by yubing).
[X86] Pass to transform tdpbf16ps intrinsics to scalar operation.
Mar 21 2021, 10:01 PM
yubing closed D96110: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation..
Mar 21 2021, 10:01 PM · Restricted Project, Restricted Project
yubing updated the diff for D96110: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation..

Rebase after https://reviews.llvm.org/D98773 is merged.

Mar 21 2021, 8:24 PM · Restricted Project, Restricted Project

Mar 18 2021

yubing updated the diff for D96110: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation..

address Pengfei's comments

Mar 18 2021, 10:31 PM · Restricted Project, Restricted Project
yubing added inline comments to D96110: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation..
Mar 18 2021, 10:30 PM · Restricted Project, Restricted Project
yubing committed rG0002d4bf3624: [X86][AMX][NFC] Give correct Passname for Tile Register Pre-configure (authored by yubing).
[X86][AMX][NFC] Give correct Passname for Tile Register Pre-configure
Mar 18 2021, 2:16 AM
yubing added inline comments to D87981: [X86] AMX programming model..
Mar 18 2021, 12:58 AM · Restricted Project, Restricted Project

Mar 16 2021

yubing committed rG320b72e9cd77: [X86][AMX] Rename amx-bf16 intrinsic according to correct naming convention (authored by yubing).
[X86][AMX] Rename amx-bf16 intrinsic according to correct naming convention
Mar 16 2021, 8:23 PM
yubing closed D98685: [X86][AMX] Rename amx-bf16 intrinsic according to correct naming convention.
Mar 16 2021, 8:23 PM · Restricted Project
yubing added a comment to D93594: [X86] Pass to transform amx intrinsics to scalar operation..

It looks like this has caused a compile-time regression at O0: https://llvm-compile-time-tracker.com/compare.php?from=9341bcbdc93a251b632ffaa51a84452a7a4a5e4e&to=4f198b0c27b04e830a3069aaf4b39cf203eaae4a&stat=instructions

The cause is probably the computation of DomTree and LoopInfo, even if no AMX intrinsics are present. I think you should be able to easily fix this by not fetching DT/LI from the pass manager, and computing them in the pass instead (only if intrinsics are present).

Mar 16 2021, 4:38 AM · Restricted Project, Restricted Project
yubing updated the diff for D96110: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation..

just do a rebase

Mar 16 2021, 1:17 AM · Restricted Project, Restricted Project
yubing retitled D98685: [X86][AMX] Rename amx-bf16 intrinsic according to correct naming convention from [X86][AMX] Rename amx-bf16 intrinsic according to correct naming convention __tile_tdpbf16ps should be renamed with __tile_dpbf16ps to [X86][AMX] Rename amx-bf16 intrinsic according to correct naming convention.
Mar 16 2021, 1:13 AM · Restricted Project
yubing requested review of D98685: [X86][AMX] Rename amx-bf16 intrinsic according to correct naming convention.
Mar 16 2021, 1:12 AM · Restricted Project
yubing added inline comments to D97358: [X86] Support amx-bf16 intrinsic..
Mar 16 2021, 12:37 AM · Restricted Project, Restricted Project

Mar 15 2021

yubing committed rG4f198b0c27b0: [X86] Pass to transform amx intrinsics to scalar operation. (authored by yubing).
[X86] Pass to transform amx intrinsics to scalar operation.
Mar 15 2021, 7:41 PM
yubing closed D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Mar 15 2021, 7:41 PM · Restricted Project, Restricted Project

Mar 8 2021

yubing updated the diff for D93594: [X86] Pass to transform amx intrinsics to scalar operation..

Fix buildfail when it is -DBUILD_SHARED_LIBS=ON

Mar 8 2021, 9:08 PM · Restricted Project, Restricted Project
yubing reopened D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Mar 8 2021, 7:30 PM · Restricted Project, Restricted Project
yubing added a comment to D93594: [X86] Pass to transform amx intrinsics to scalar operation..

@yubing I've reverted this as it was failing on a lot of buildbots: http://lab.llvm.org:8011/#/builders/109/builds/9867

Mar 8 2021, 6:03 PM · Restricted Project, Restricted Project

Mar 6 2021

yubing added a comment to D93594: [X86] Pass to transform amx intrinsics to scalar operation..

Thanks all for reporting and reverting this. I will do bugfix asap.

Mar 6 2021, 5:13 AM · Restricted Project, Restricted Project

Mar 5 2021

yubing committed rG8198d83965ba: [X86] Pass to transform amx intrinsics to scalar operation. (authored by LuoYuanke).
[X86] Pass to transform amx intrinsics to scalar operation.
Mar 5 2021, 12:03 AM
yubing closed D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Mar 5 2021, 12:02 AM · Restricted Project, Restricted Project

Mar 4 2021

yubing updated the diff for D93594: [X86] Pass to transform amx intrinsics to scalar operation..

Address pengfei's comments

Mar 4 2021, 11:48 PM · Restricted Project, Restricted Project

Mar 2 2021

yubing added inline comments to D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Mar 2 2021, 7:30 PM · Restricted Project, Restricted Project

Mar 1 2021

yubing added inline comments to D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Mar 1 2021, 11:21 PM · Restricted Project, Restricted Project
yubing updated the diff for D93594: [X86] Pass to transform amx intrinsics to scalar operation..

address comments above

Mar 1 2021, 11:13 PM · Restricted Project, Restricted Project

Feb 28 2021

yubing added inline comments to D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Feb 28 2021, 9:09 PM · Restricted Project, Restricted Project

Feb 24 2021

yubing added inline comments to D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Feb 24 2021, 6:50 PM · Restricted Project, Restricted Project

Feb 23 2021

yubing updated the summary of D93594: [X86] Pass to transform amx intrinsics to scalar operation..
Feb 23 2021, 7:22 PM · Restricted Project, Restricted Project
yubing updated the diff for D93594: [X86] Pass to transform amx intrinsics to scalar operation..

Fix some comments and commit message

Feb 23 2021, 7:18 PM · Restricted Project, Restricted Project

Feb 22 2021

yubing updated the diff for D96110: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation..

Modify some comments

Feb 22 2021, 11:06 PM · Restricted Project, Restricted Project
yubing updated the diff for D96110: [X86] Pass to transform tdpbf16ps intrinsics to scalar operation..

Fix incorrect naming for dpbf16's bb

Feb 22 2021, 10:57 PM · Restricted Project, Restricted Project