wdng (Wei Ding)
User

Projects

User does not belong to any projects.

User Details

User Since
Feb 23 2016, 2:17 PM (86 w, 5 d)

Recent Activity

Tue, Oct 17

wdng committed rL316037: AMDGPU : Fix an error for the llvm.cttz implementation..
AMDGPU : Fix an error for the llvm.cttz implementation.
Tue, Oct 17, 2:50 PM
wdng closed D39014: AMDGPU : Fix an error for the llvm.cttz implementation. by committing rL316037: AMDGPU : Fix an error for the llvm.cttz implementation..
Tue, Oct 17, 2:50 PM
wdng created D39014: AMDGPU : Fix an error for the llvm.cttz implementation..
Tue, Oct 17, 11:40 AM

Mon, Oct 16

wdng retitled D38634: AMDGPU : Custom lowering constrained fps. from AMDGPU : Expand SDNodeFlags APIs & custom lowering constrained fps. to AMDGPU : Custom lowering constrained fps..
Mon, Oct 16, 11:37 AM
wdng updated the diff for D38634: AMDGPU : Custom lowering constrained fps..

Instead of using SDNodeFlags to store metadata information, this patch directly appends an extra operand for rounding mode during the DAG build phase based on Stats's suggestion. This patch currently implements strict fps for fadd, fsub, fma, and sqrt. Thanks a lot for Andy and Stas's comments for this!

Mon, Oct 16, 11:35 AM

Thu, Oct 12

wdng committed rL315610: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..
Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ.
Thu, Oct 12, 12:37 PM
wdng closed D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ. by committing rL315610: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..
Thu, Oct 12, 12:37 PM

Wed, Oct 11

wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Removed -DAG checks completely.

Wed, Oct 11, 5:45 PM
wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Remove duplicate check lines.

Wed, Oct 11, 5:08 PM
wdng added inline comments to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..
Wed, Oct 11, 4:27 PM
wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Address code reviews.

Wed, Oct 11, 4:14 PM

Tue, Oct 10

wdng added a reviewer for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ.: rampitec.
Tue, Oct 10, 11:10 AM
wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Ping.

Tue, Oct 10, 8:18 AM

Mon, Oct 9

wdng added a reviewer for D38634: AMDGPU : Custom lowering constrained fps.: rampitec.
Mon, Oct 9, 2:58 PM
wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Address code reivews.

Mon, Oct 9, 2:53 PM

Fri, Oct 6

wdng updated the diff for D38634: AMDGPU : Custom lowering constrained fps..

Fixed format issue.

Fri, Oct 6, 9:56 AM
wdng created D38634: AMDGPU : Custom lowering constrained fps..
Fri, Oct 6, 9:38 AM

Mon, Oct 2

wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Ping.

Mon, Oct 2, 12:10 PM

Tue, Sep 26

wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Ping.

Tue, Sep 26, 3:30 PM

Sep 22 2017

wdng edited reviewers for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ., added: kzhuravl; removed: craig.topper.
Sep 22 2017, 12:29 PM

Sep 20 2017

wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Ping.

Sep 20 2017, 9:57 PM

Sep 19 2017

wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Will create another separate ticket to fix the v_ffbl_sdwa instruction generation.

Sep 19 2017, 2:54 PM

Sep 15 2017

wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Ping.

Sep 15 2017, 8:47 AM

Sep 14 2017

wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Fix the issues that variables are not capitalized.

Sep 14 2017, 2:47 PM
wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Address code reviews.

Sep 14 2017, 2:41 PM
wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Upload a full diff.

Sep 14 2017, 8:42 AM
wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Changes based on code review feedback.

Sep 14 2017, 8:40 AM

Sep 11 2017

wdng added inline comments to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..
Sep 11 2017, 9:07 AM

Sep 8 2017

wdng added a reviewer for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ.: t-tye.
Sep 8 2017, 12:10 PM
wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Ping.

Sep 8 2017, 8:40 AM

Sep 7 2017

wdng added a reviewer for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ.: craig.topper.
Sep 7 2017, 11:37 AM
wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..
Sep 7 2017, 11:13 AM

Sep 1 2017

wdng updated the diff for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Just add a custom lowering ISD:CTTZ to ISD::CTTZ_ZERO_UNDEF

Sep 1 2017, 12:45 PM
wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

Ping.

Sep 1 2017, 9:29 AM
wdng added a comment to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..

If we don't want to change the original way of implementation, we may want to do a custom lowering for ISD::CTTZ at AMDGPU backend to ISD::CTTZ_ZERO_UNDE?

Sep 1 2017, 9:02 AM

Aug 31 2017

wdng added inline comments to D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..
Aug 31 2017, 1:54 PM
wdng created D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ..
Aug 31 2017, 12:28 PM

Aug 23 2017

wdng committed rL311629: Add ‘llvm.experimental.constrained.fma‘ Intrinsic..
Add ‘llvm.experimental.constrained.fma‘ Intrinsic.
Aug 23 2017, 9:19 PM
wdng closed D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic by committing rL311629: Add ‘llvm.experimental.constrained.fma‘ Intrinsic..
Aug 23 2017, 9:19 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Address code reviews. Thanks!

Aug 23 2017, 3:14 PM
wdng added a comment to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Can you revert the white space changes in the places you aren't otherwise modifying? In general, you shouldn't make formatting changes outside of the parts of the file your patch is modifying. It complicates the version control blame process without adding a lot of benefit.

Also, your latest diffs seem not to have full file context (such as you get with the -x -U99999 switch with diff). This isn't important for the current review, but it is something to keep in mind going forward.

I really appreciate your work on this patch, and I hate to seem like I'm nit-picking a lot. I just want to make sure we do things correctly. Thanks!

Aug 23 2017, 12:12 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Fixed a format issue.

Aug 23 2017, 11:46 AM
wdng added a reviewer for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic: craig.topper.
Aug 23 2017, 11:28 AM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Patch update after the patch [X86] Remove X86ISD::FMADD in favor ISD::FMA has been upstreamed.

Aug 23 2017, 11:28 AM

Aug 22 2017

wdng added a comment to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Can you put this off until the patch that Craig submitted in D36983 either lands or gets rejected? If that change goes through, you should be able to remove your modifications to X86ISelDAGToDAG.cpp.

Aug 22 2017, 9:10 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Address code reviews. Thanks a lot!

Aug 22 2017, 1:14 PM

Aug 18 2017

wdng added a comment to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Ping.

Aug 18 2017, 2:45 PM
wdng added inline comments to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.
Aug 18 2017, 11:57 AM

Aug 17 2017

wdng added a comment to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Ping.

Aug 17 2017, 9:22 AM

Aug 16 2017

wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Remove extra period. Thanks!

Aug 16 2017, 3:24 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Update LangRef.rst: put more accurate descriptions into the constrained.fma semantic section.

Aug 16 2017, 2:12 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Update LangRef.rst based on comments.

Aug 16 2017, 1:31 PM
wdng added inline comments to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.
Aug 16 2017, 9:10 AM

Aug 15 2017

wdng added inline comments to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.
Aug 15 2017, 3:44 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Address code reviews.

Aug 15 2017, 3:37 PM

Aug 11 2017

wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Code changes based on Brian's comments.

Aug 11 2017, 3:30 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Upload correct diff.

Aug 11 2017, 1:54 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Address code reviews.

Aug 11 2017, 1:46 PM

Aug 4 2017

wdng added a comment to D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

An update to docs/LangRef.rst is needed.

Aug 4 2017, 2:13 PM
wdng updated the diff for D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.

Add missing lit tests.

Aug 4 2017, 2:04 PM
wdng created D36335: Add ‘llvm.experimental.constrained.fma‘ Intrinsic.
Aug 4 2017, 12:10 PM

Jul 26 2017

wdng committed rL309178: AMDGPU : Widen extending scalar loads to 32-bits..
AMDGPU : Widen extending scalar loads to 32-bits.
Jul 26 2017, 2:09 PM
wdng closed D35146: AMDGPU : Widen extending scalar loads to 32-bits by committing rL309178: AMDGPU : Widen extending scalar loads to 32-bits..
Jul 26 2017, 2:09 PM
wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Modified function name.

Jul 26 2017, 1:43 PM
wdng added a comment to D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Ping.

Jul 26 2017, 8:28 AM

Jul 25 2017

wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Address code reviews.

Jul 25 2017, 10:16 AM
wdng added a comment to D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Ping.

Jul 25 2017, 9:13 AM

Jul 24 2017

wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Upload correct diff file.

Jul 24 2017, 3:47 PM
wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Fixed a mistake for align2_i8 lit test.

Jul 24 2017, 2:38 PM
wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Address code reviews.

Jul 24 2017, 1:52 PM

Jul 21 2017

wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Address code reviews.

Jul 21 2017, 3:42 PM
wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.

Address code reviews.

Jul 21 2017, 9:20 AM

Jul 17 2017

wdng added inline comments to D35146: AMDGPU : Widen extending scalar loads to 32-bits.
Jul 17 2017, 2:51 PM
wdng updated the diff for D35146: AMDGPU : Widen extending scalar loads to 32-bits.
  1. Address code reviews. Looks like adding "isDereferenceableAndAlignedPointer" is too strong to prevent expected code transformations.
  2. Modify related LIT tests.
Jul 17 2017, 9:55 AM

Jul 9 2017

wdng created D35146: AMDGPU : Widen extending scalar loads to 32-bits.
Jul 9 2017, 6:13 AM

Jun 30 2017

wdng added a comment to D34726: AMDGPU/SI: Do not insert an instruction into worklist twice in movetovalu .

Hi, is it possible to upload a full diff?

Jun 30 2017, 8:52 PM

Jun 29 2017

wdng abandoned D33375: [InstCombine] Don't fold icmp sgt/slt (add nsw X, C2), C --> icmp sgt/slt X, (C - C2) when X is a PHI node..
Jun 29 2017, 12:50 PM

Jun 9 2017

wdng committed rL305137: AMDGPU : Fix ISA Version Definitions..
AMDGPU : Fix ISA Version Definitions.
Jun 9 2017, 8:54 PM
wdng closed D28531: AMDGPU : Fix ISA Version Definitions. by committing rL305137: AMDGPU : Fix ISA Version Definitions..
Jun 9 2017, 8:53 PM
wdng updated the diff for D28531: AMDGPU : Fix ISA Version Definitions..

Modified lit tests.

Jun 9 2017, 3:47 PM
wdng updated the diff for D28531: AMDGPU : Fix ISA Version Definitions..

Address code reviews.

Jun 9 2017, 9:21 AM

Jun 8 2017

wdng updated the diff for D28531: AMDGPU : Fix ISA Version Definitions..

Address code reviews based on Tony's @t-tye comments

Jun 8 2017, 9:56 AM

Jun 7 2017

wdng added a comment to D28531: AMDGPU : Fix ISA Version Definitions..

Ping.

Jun 7 2017, 2:03 PM
wdng updated the diff for D28531: AMDGPU : Fix ISA Version Definitions..

Address code reviews.

Jun 7 2017, 8:30 AM

Jun 6 2017

wdng updated the diff for D28531: AMDGPU : Fix ISA Version Definitions..

Add gfx600 and gfx601 definitions.

Jun 6 2017, 9:20 AM

Jun 5 2017

wdng updated the diff for D28531: AMDGPU : Fix ISA Version Definitions..

Add new ISA version definitions & fix incorrect ISA versions.

Jun 5 2017, 10:30 PM
wdng removed reviewers for D28531: AMDGPU : Fix ISA Version Definitions.: tstellarAMD, kzhuravl-AMD.
Jun 5 2017, 10:27 PM

May 22 2017

wdng added a comment to D33375: [InstCombine] Don't fold icmp sgt/slt (add nsw X, C2), C --> icmp sgt/slt X, (C - C2) when X is a PHI node..

Sorry, but I still don't understand.

%inc = add nuw nsw i32 %myValue.0, 1
%cmp15 = icmp slt i32 %inc, 255

%myValue.0 will be 0 when we enter the loop. Eventually, it will increment up to 255, and %cmp15 will be false. How is that logically different than:
%cmp15 = icmp slt i32 %myValue.0, 254 ?

%inc is not always one more than %myValue.0?

May 22 2017, 9:56 AM
wdng added a comment to D33375: [InstCombine] Don't fold icmp sgt/slt (add nsw X, C2), C --> icmp sgt/slt X, (C - C2) when X is a PHI node..

I don't understand why this change is needed. Can you explain how the test in this patch can miscompile?

May 22 2017, 9:14 AM
wdng added reviewers for D33375: [InstCombine] Don't fold icmp sgt/slt (add nsw X, C2), C --> icmp sgt/slt X, (C - C2) when X is a PHI node.: spatel, nlopes, efriedma.
May 22 2017, 8:56 AM

May 19 2017

wdng updated the summary of D33375: [InstCombine] Don't fold icmp sgt/slt (add nsw X, C2), C --> icmp sgt/slt X, (C - C2) when X is a PHI node..
May 19 2017, 11:47 PM
wdng created D33375: [InstCombine] Don't fold icmp sgt/slt (add nsw X, C2), C --> icmp sgt/slt X, (C - C2) when X is a PHI node..
May 19 2017, 11:45 PM

Apr 12 2017

wdng committed rL300142: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform….
AMDGPU : Fix common dominator of two incoming blocks terminates with uniform…
Apr 12 2017, 5:04 PM
wdng closed D31350: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform branch issue. by committing rL300142: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform….
Apr 12 2017, 5:04 PM
wdng added inline comments to D31350: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform branch issue..
Apr 12 2017, 4:46 PM
wdng updated the diff for D31350: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform branch issue..

Address code review.

Apr 12 2017, 4:42 PM
wdng updated the diff for D31350: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform branch issue..
  1. Address code reviews.
  2. Will create another patch for optimized searching of divergent terminator.
Apr 12 2017, 4:30 PM
wdng added a comment to D31350: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform branch issue..

Ping.

Apr 12 2017, 10:14 AM
wdng updated the diff for D31350: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform branch issue..

Change function name.

Apr 12 2017, 8:53 AM

Apr 11 2017

wdng added inline comments to D31350: AMDGPU : Fix common dominator of two incoming blocks terminates with uniform branch issue..
Apr 11 2017, 11:47 PM