Page MenuHomePhabricator

xiangzhangllvm (Xiang Zhang)
User

Projects

User does not belong to any projects.

User Details

User Since
Dec 25 2018, 11:14 PM (184 w, 1 d)

Recent Activity

Thu, Jun 30

xiangzhangllvm committed rG72a23cef7e66: [ISel] Match all bits when merge undefs for DAG combine (authored by xiangzhangllvm).
[ISel] Match all bits when merge undefs for DAG combine
Thu, Jun 30, 6:14 PM · Restricted Project, Restricted Project
xiangzhangllvm closed D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.
Thu, Jun 30, 6:14 PM · Restricted Project, Restricted Project
xiangzhangllvm added a reverting change for rG5fe5aa284efe: [ISel] Match all bits when merge undef(s) for DAG combine: rG64f44a90efb7: Revert "[ISel] Match all bits when merge undef(s) for DAG combine".
Thu, Jun 30, 5:59 PM · Restricted Project, Restricted Project
xiangzhangllvm committed rG64f44a90efb7: Revert "[ISel] Match all bits when merge undef(s) for DAG combine" (authored by xiangzhangllvm).
Revert "[ISel] Match all bits when merge undef(s) for DAG combine"
Thu, Jun 30, 5:59 PM · Restricted Project, Restricted Project
xiangzhangllvm committed rG5fe5aa284efe: [ISel] Match all bits when merge undef(s) for DAG combine (authored by xiangzhangllvm).
[ISel] Match all bits when merge undef(s) for DAG combine
Thu, Jun 30, 5:58 PM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.

Address Simon's comments.

Thu, Jun 30, 4:30 AM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.

Please pre-commit fshl-splat-undef.ll with current trunk codegen so that the patch commit shows the codegen change.

LGTM with a few minors

Thu, Jun 30, 4:01 AM · Restricted Project, Restricted Project
xiangzhangllvm committed rG988901ac7405: [test] Add a lit test fshl-splat-undef.ll (authored by xiangzhangllvm).
[test] Add a lit test fshl-splat-undef.ll
Thu, Jun 30, 4:00 AM · Restricted Project, Restricted Project

Wed, Jun 29

xiangzhangllvm accepted D128804: [greedyalloc] Return early when there is no register to allocate..

LGTM

Wed, Jun 29, 6:36 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.

Hi Friends, PLS can you accept this fix ? This is related to one of our Must Fix bug. Many thanks!

Wed, Jun 29, 5:55 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D128804: [greedyalloc] Return early when there is no register to allocate..
Wed, Jun 29, 5:44 PM · Restricted Project, Restricted Project

Tue, Jun 28

xiangzhangllvm updated the diff for D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.
Tue, Jun 28, 8:41 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.

OK, let me add test coverage in APIntTest.cpp, many thanks!

Tue, Jun 28, 7:12 PM · Restricted Project, Restricted Project
xiangzhangllvm accepted D128584: [X86][AMX] Split greedy RA for tile register.
Tue, Jun 28, 5:39 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.
Tue, Jun 28, 12:44 AM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D128584: [X86][AMX] Split greedy RA for tile register.

In previous code, the tile register and shape register are allocated in the same pass. The shape config is processed before "virtregrewriter". Though when filling shape, it is still virtual register, but the virtual register has been split or spillied.

Tue, Jun 28, 12:06 AM · Restricted Project, Restricted Project

Mon, Jun 27

xiangzhangllvm added inline comments to D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.
Mon, Jun 27, 11:36 PM · Restricted Project, Restricted Project
xiangzhangllvm added a reviewer for D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine: craig.topper.
Mon, Jun 27, 11:33 PM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.

Add a test

Mon, Jun 27, 11:30 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D128584: [X86][AMX] Split greedy RA for tile register.
Mon, Jun 27, 7:31 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D128584: [X86][AMX] Split greedy RA for tile register.

I think the spill/split should still cover the shape regs:

Mon, Jun 27, 7:16 PM · Restricted Project, Restricted Project

Fri, Jun 24

xiangzhangllvm added a comment to D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.

Let me try create a small lit test for it. Thanks

Fri, Jun 24, 7:33 PM · Restricted Project, Restricted Project
xiangzhangllvm requested review of D128570: [ISel] Round down mask bit when merge undef(s) for DAG combine.
Fri, Jun 24, 7:32 PM · Restricted Project, Restricted Project

Wed, Jun 22

xiangzhangllvm accepted D126771: [fastalloc] Support allocate specific register class in fastalloc..

LGTM

Wed, Jun 22, 8:33 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D126771: [fastalloc] Support allocate specific register class in fastalloc..
Wed, Jun 22, 5:47 PM · Restricted Project, Restricted Project

Mon, Jun 20

xiangzhangllvm added inline comments to D126771: [fastalloc] Support allocate specific register class in fastalloc..
Mon, Jun 20, 5:53 PM · Restricted Project, Restricted Project

Sun, Jun 19

xiangzhangllvm added inline comments to D126771: [fastalloc] Support allocate specific register class in fastalloc..
Sun, Jun 19, 11:43 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D126771: [fastalloc] Support allocate specific register class in fastalloc..
Sun, Jun 19, 7:29 PM · Restricted Project, Restricted Project

Fri, Jun 17

xiangzhangllvm added inline comments to D126771: [fastalloc] Support allocate specific register class in fastalloc..
Fri, Jun 17, 1:37 AM · Restricted Project, Restricted Project

Sun, Jun 12

xiangzhangllvm added a comment to D126771: [fastalloc] Support allocate specific register class in fastalloc..

I think the idea of "split" special registers RA pass is good (especially for the registers who need to be config).
I notice all the "isVirtual<Register>" places need carefully to "exclude" the special registers.
Seems we may no need to do it if we can make sure that we handle these special registers before normal fast RA.
(because they have be allocated to physic registers)

Sun, Jun 12, 10:24 PM · Restricted Project, Restricted Project

May 30 2022

xiangzhangllvm committed rG5d5aba78dbbe: [X86][NFC] Refine X86 Domain Reassignment for compiling time (authored by xiangzhangllvm).
[X86][NFC] Refine X86 Domain Reassignment for compiling time
May 30 2022, 7:11 PM · Restricted Project, Restricted Project
xiangzhangllvm closed D126622: [X86][NFC] Refine X86 Domain Reassignment for compiling time.
May 30 2022, 7:11 PM · Restricted Project, Restricted Project

May 29 2022

xiangzhangllvm requested review of D126622: [X86][NFC] Refine X86 Domain Reassignment for compiling time.
May 29 2022, 5:51 PM · Restricted Project, Restricted Project

May 27 2022

xiangzhangllvm accepted D126519: [X86][AMX] Replace LDTILECFG with PLDTILECFGV on auto-config..

LGTM

May 27 2022, 1:04 AM · Restricted Project, Restricted Project

May 26 2022

xiangzhangllvm added inline comments to D126519: [X86][AMX] Replace LDTILECFG with PLDTILECFGV on auto-config..
May 26 2022, 11:03 PM · Restricted Project, Restricted Project

May 23 2022

xiangzhangllvm accepted D125075: [X86][AMX] Multiple configure for AMX register..

Mark some TODO for your planning.

May 23 2022, 7:41 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D125075: [X86][AMX] Multiple configure for AMX register..
May 23 2022, 6:47 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D125075: [X86][AMX] Multiple configure for AMX register..
May 23 2022, 6:15 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D125075: [X86][AMX] Multiple configure for AMX register..
May 23 2022, 5:46 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D125075: [X86][AMX] Multiple configure for AMX register..
May 23 2022, 2:10 AM · Restricted Project, Restricted Project

May 20 2022

xiangzhangllvm added inline comments to D125075: [X86][AMX] Multiple configure for AMX register..
May 20 2022, 7:43 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D125075: [X86][AMX] Multiple configure for AMX register..
May 20 2022, 2:59 AM · Restricted Project, Restricted Project

May 10 2022

xiangzhangllvm committed rG2ea8f203cd9a: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI (authored by xiangzhangllvm).
[CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI
May 10 2022, 5:59 PM · Restricted Project, Restricted Project
xiangzhangllvm closed D125159: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI.
May 10 2022, 5:58 PM · Restricted Project, Restricted Project

May 9 2022

xiangzhangllvm updated the diff for D125159: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI.

Address pengfei's comment

May 9 2022, 8:09 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D125159: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI.
May 9 2022, 2:02 AM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D125075: [X86][AMX] Multiple configure for AMX register..
May 9 2022, 1:22 AM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D125159: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI.

Address pengfei's comment.

May 9 2022, 1:02 AM · Restricted Project, Restricted Project

May 8 2022

xiangzhangllvm added inline comments to D125159: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI.
May 8 2022, 6:29 PM · Restricted Project, Restricted Project

May 7 2022

xiangzhangllvm added inline comments to D125159: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI.
May 7 2022, 1:55 AM · Restricted Project, Restricted Project
xiangzhangllvm requested review of D125159: [CodeGen] Fix ConvertNodeToLibcall for STRICT_FPOWI.
May 7 2022, 1:30 AM · Restricted Project, Restricted Project

May 6 2022

xiangzhangllvm added inline comments to D118355: Add -mmanual-endbr switch to allow manual selection of control-flow protection.
May 6 2022, 5:43 PM · Restricted Project, Restricted Project, Restricted Project

May 5 2022

xiangzhangllvm added inline comments to D118355: Add -mmanual-endbr switch to allow manual selection of control-flow protection.
May 5 2022, 6:07 PM · Restricted Project, Restricted Project, Restricted Project

May 4 2022

xiangzhangllvm added a comment to D124903: [X86][AMX] Replace PXOR instruction with SET0 in AMX pre config..

To generate zero value, ... may convert

>%0:vr128 = PXORrr undef %0, undef %0
>to %1:vr128 = PXORrr undef %1:vr128(tied-def 0), undef %0:vr128.

It is not expected.

May 4 2022, 6:20 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124903: [X86][AMX] Replace PXOR instruction with SET0 in AMX pre config..
May 4 2022, 6:09 PM · Restricted Project, Restricted Project

Apr 26 2022

xiangzhangllvm accepted D124378: [X86][AMX] combine tile cast and load/store instruction..

LGTM

Apr 26 2022, 5:43 PM · Restricted Project, Restricted Project
xiangzhangllvm committed rGc430f0f532de: [X86] Add use condition for combineSetCCMOVMSK (authored by xiangzhangllvm).
[X86] Add use condition for combineSetCCMOVMSK
Apr 26 2022, 1:44 AM · Restricted Project, Restricted Project
xiangzhangllvm closed D123652: Add use condition for combine SetCCMOVMSK.
Apr 26 2022, 1:44 AM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124378: [X86][AMX] combine tile cast and load/store instruction..
Apr 26 2022, 1:15 AM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

Many thanks for your careful reviewing !!

Apr 26 2022, 1:00 AM · Restricted Project, Restricted Project

Apr 25 2022

xiangzhangllvm added a comment to D124378: [X86][AMX] combine tile cast and load/store instruction..

Combine usually make sense. Not sure the case load/store has special attributes, e.g. volatile. The new amx load/store may miss them.

Apr 25 2022, 7:01 PM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D123652: Add use condition for combine SetCCMOVMSK.

Use IsOneUse, let it more readable.

Apr 25 2022, 6:33 PM · Restricted Project, Restricted Project

Apr 24 2022

xiangzhangllvm added inline comments to D123652: Add use condition for combine SetCCMOVMSK.
Apr 24 2022, 8:39 PM · Restricted Project, Restricted Project
xiangzhangllvm accepted D124342: [X86][AMX] Report error when shapes are not pre-defined..
Apr 24 2022, 7:00 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D124342: [X86][AMX] Report error when shapes are not pre-defined..

LGTM

Apr 24 2022, 7:00 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124342: [X86][AMX] Report error when shapes are not pre-defined..
Apr 24 2022, 5:38 PM · Restricted Project, Restricted Project

Apr 22 2022

xiangzhangllvm updated the diff for D123652: Add use condition for combine SetCCMOVMSK.

Rebase

Apr 22 2022, 12:09 AM · Restricted Project, Restricted Project

Apr 21 2022

xiangzhangllvm added a comment to D124067: [x86] Support 3 builtin functions for 32-bits targets.

close with clang format at
commit 6454ff35e0e7b0c0762c640031aa6c2b5d1f16ec
[Clang Format] emmintrin.h smmintrin.h (NFC)

Apr 21 2022, 8:31 PM · Restricted Project, Restricted Project
xiangzhangllvm committed rGafa536e33e10: [x86] Support 3 builtin functions for 32-bits mode (authored by xiangzhangllvm).
[x86] Support 3 builtin functions for 32-bits mode
Apr 21 2022, 8:29 PM · Restricted Project, Restricted Project
xiangzhangllvm closed D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 21 2022, 8:29 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

Thank you for your effort !!
Hi @RKSimon , a small question: do your local lit test can be checked by “make check-all” these 2 days ?
I ask it because I try to change the code and do make check-all to find out which tests will be affect. But I find we (out team) all meet “make check-all” problem these days. (Can not finish in very long time)

Apr 21 2022, 7:05 PM · Restricted Project, Restricted Project
xiangzhangllvm committed rG6454ff35e0e7: [Clang Format] emmintrin.h smmintrin.h (NFC) (authored by xiangzhangllvm).
[Clang Format] emmintrin.h smmintrin.h (NFC)
Apr 21 2022, 6:30 PM · Restricted Project, Restricted Project
xiangzhangllvm added a reverting change for rGa69c219a8c9f: [x86] Support 3 builtin functions for 32-bits mode: rGcaf5ad5da7c2: Revert "[x86] Support 3 builtin functions for 32-bits mode".
Apr 21 2022, 6:12 PM · Restricted Project, Restricted Project
xiangzhangllvm committed rGcaf5ad5da7c2: Revert "[x86] Support 3 builtin functions for 32-bits mode" (authored by xiangzhangllvm).
Revert "[x86] Support 3 builtin functions for 32-bits mode"
Apr 21 2022, 6:12 PM · Restricted Project, Restricted Project
xiangzhangllvm committed rGa69c219a8c9f: [x86] Support 3 builtin functions for 32-bits mode (authored by xiangzhangllvm).
[x86] Support 3 builtin functions for 32-bits mode
Apr 21 2022, 6:07 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 21 2022, 5:32 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 21 2022, 12:33 AM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 21 2022, 12:32 AM · Restricted Project, Restricted Project

Apr 20 2022

xiangzhangllvm added inline comments to D123652: Add use condition for combine SetCCMOVMSK.
Apr 20 2022, 7:04 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

Rebased. Thank you!

Apr 20 2022, 6:36 PM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D123652: Add use condition for combine SetCCMOVMSK.
Apr 20 2022, 6:34 PM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 20 2022, 6:28 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 20 2022, 6:21 PM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 20 2022, 6:19 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

please can you rebase?

Apr 20 2022, 5:56 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D124067: [x86] Support 3 builtin functions for 32-bits targets.

OK - SSE2/SSE41 now have i386 coverage - please can you rebase and update the checks to use CHECK/X64/X86 ?

Apr 20 2022, 5:34 PM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 20 2022, 2:31 AM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 20 2022, 1:45 AM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 20 2022, 1:02 AM · Restricted Project, Restricted Project
xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 20 2022, 12:13 AM · Restricted Project, Restricted Project

Apr 19 2022

xiangzhangllvm added inline comments to D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 19 2022, 11:25 PM · Restricted Project, Restricted Project
xiangzhangllvm requested review of D124067: [x86] Support 3 builtin functions for 32-bits targets.
Apr 19 2022, 10:55 PM · Restricted Project, Restricted Project
xiangzhangllvm updated the diff for D123652: Add use condition for combine SetCCMOVMSK.

Add yuanke's test

Apr 19 2022, 6:54 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

Try this test case?

That is cool!

Apr 19 2022, 5:41 PM · Restricted Project, Restricted Project

Apr 14 2022

xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

The small test case is easy to reproduced on llvm.14 PLS refer EY93Kc77E
it continue to do the combine ( gen vpsubb + vptest ) even there is more uses of the cmp result.

Apr 14 2022, 12:57 AM · Restricted Project, Restricted Project
xiangzhangllvm added reviewers for D123652: Add use condition for combine SetCCMOVMSK: pengfei, wxiao3, LuoYuanke.
Apr 14 2022, 12:51 AM · Restricted Project, Restricted Project

Apr 13 2022

xiangzhangllvm added inline comments to rG6997f4d07fa4: [X86] combineSetCCMOVMSK - fold allof(cmpeq(x,y)) -> ptest(sub(x,y)) (PR53379).
Apr 13 2022, 11:19 PM
xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

Maybe use llvm-reduce to reduce the original IR? You can use llvm-extract --bb=FunctionName:DAGCombineBB YourIR.ll to get the bb first.

Apr 13 2022, 10:21 PM · Restricted Project, Restricted Project
xiangzhangllvm added a comment to D123652: Add use condition for combine SetCCMOVMSK.

Please can you add test coverage - preferably for both cases?

Apr 13 2022, 8:51 PM · Restricted Project, Restricted Project

Apr 12 2022

xiangzhangllvm added inline comments to rG6997f4d07fa4: [X86] combineSetCCMOVMSK - fold allof(cmpeq(x,y)) -> ptest(sub(x,y)) (PR53379).
Apr 12 2022, 10:41 PM