Page MenuHomePhabricator
People vpykhtin

vpykhtin (Valery Pykhtin)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 28 2016, 8:30 AM (207 w, 2 d)

Recent Activity
View All

Dec 12 2019

vpykhtin accepted D71132: PostRA Machine Sink should take care of COPY defining register that is a sub-register by another COPY source operand.

LGTM.

Dec 12 2019, 9:09 AM · Restricted Project
vpykhtin added a comment to D71132: PostRA Machine Sink should take care of COPY defining register that is a sub-register by another COPY source operand.

Almost LGTM. Do you need those liveins reorderings?

Dec 12 2019, 8:04 AM · Restricted Project

Dec 6 2019

vpykhtin added a comment to D71089: [AMDGPU] Optimizing unnecessary copies for REG_SEQUENCE PHI operand. Also fixes rocBLAS error.

Looks good, but the test would be nice to have.

Dec 6 2019, 8:12 AM · Restricted Project

Nov 26 2019

vpykhtin committed rG008e65a7bfb3: [AMDGPU] Fix emitIfBreak CF lowering: use temp reg to make register coalescer… (authored by vpykhtin).
[AMDGPU] Fix emitIfBreak CF lowering: use temp reg to make register coalescer…
Nov 26 2019, 8:07 AM
vpykhtin closed D70405: [AMDGPU] Fix emitIfBreak CF lowering: use a temp register to make register coalescer life easier..
Nov 26 2019, 8:07 AM · Restricted Project

Nov 19 2019

vpykhtin accepted D70400: [AMDGPU][GFX10] Disabled v_movrel*[sdwa|dpp] opcodes in codegen.

LGTM.

Nov 19 2019, 8:34 AM · Restricted Project
vpykhtin added inline comments to D70405: [AMDGPU] Fix emitIfBreak CF lowering: use a temp register to make register coalescer life easier..
Nov 19 2019, 6:54 AM · Restricted Project
vpykhtin updated the diff for D70405: [AMDGPU] Fix emitIfBreak CF lowering: use a temp register to make register coalescer life easier..

updated the diff per comment. Used utils/update_llc_test_checks.py tool to update autogenerated test.

Nov 19 2019, 6:54 AM · Restricted Project

Nov 18 2019

vpykhtin updated the diff for D70405: [AMDGPU] Fix emitIfBreak CF lowering: use a temp register to make register coalescer life easier..

added test fix.

Nov 18 2019, 10:31 AM · Restricted Project
vpykhtin added inline comments to D70400: [AMDGPU][GFX10] Disabled v_movrel*[sdwa|dpp] opcodes in codegen.
Nov 18 2019, 10:05 AM · Restricted Project
vpykhtin created D70405: [AMDGPU] Fix emitIfBreak CF lowering: use a temp register to make register coalescer life easier..
Nov 18 2019, 9:38 AM · Restricted Project
vpykhtin accepted D70402: [AMDGPU][DPP] Corrected DPP combiner.

LGTM.

Nov 18 2019, 8:44 AM · Restricted Project
vpykhtin added a comment to D70400: [AMDGPU][GFX10] Disabled v_movrel*[sdwa|dpp] opcodes in codegen.

Look mostly good, but can you split this change into one that relates to DPP and another that disables asm only instructions?

Nov 18 2019, 8:18 AM · Restricted Project

Oct 25 2019

vpykhtin committed rGc9c18e5a3194: [AMDGPU] Disallow dpp combining for dpp instructions without Src2 operand (when… (authored by vpykhtin).
[AMDGPU] Disallow dpp combining for dpp instructions without Src2 operand (when…
Oct 25 2019, 11:41 AM
vpykhtin closed D69430: Disallow dpp combining for dpp instructions without Src2 operand (when Src2 is required).
Oct 25 2019, 11:40 AM · Restricted Project
vpykhtin accepted D69413: [AMDGPU] Fold AGPR reg_sequence initializers.

LGTM.

Oct 25 2019, 11:40 AM · Restricted Project
vpykhtin created D69430: Disallow dpp combining for dpp instructions without Src2 operand (when Src2 is required).
Oct 25 2019, 4:55 AM · Restricted Project

Oct 23 2019

vpykhtin accepted D69280: [AMDGPU] Allow folding of sgpr to vgpr copy.

LGTM.

Oct 23 2019, 9:39 AM · Restricted Project

Oct 16 2019

vpykhtin accepted D69010: [AMDGPU] Supress unused sdwa insts generation.

LGTM.

Oct 16 2019, 9:13 AM · Restricted Project

Oct 15 2019

vpykhtin accepted D68673: [AMDGPU] Support mov dpp with 64 bit operands.

LGTM

Oct 15 2019, 6:31 AM · Restricted Project
vpykhtin accepted D68828: [AMDGPU] Allow DPP combiner to work with REG_SEQUENCE.

LGTM

Oct 15 2019, 6:31 AM · Restricted Project

Oct 14 2019

vpykhtin added inline comments to D68828: [AMDGPU] Allow DPP combiner to work with REG_SEQUENCE.
Oct 14 2019, 10:06 AM · Restricted Project
vpykhtin added inline comments to D68673: [AMDGPU] Support mov dpp with 64 bit operands.
Oct 14 2019, 9:57 AM · Restricted Project

Oct 10 2019

vpykhtin accepted D68813: [AMDGPU] Handle undef old operand in DPP combine.

LGTM.

Oct 10 2019, 2:05 PM · Restricted Project

Oct 2 2019

vpykhtin added a comment to D68338: [AMDGPU] Remove dubious logic in bidirectional list scheduler.

However it would be nice to have a test that shows benefit. I like the reasoning in the description of this change.

Oct 2 2019, 8:33 AM · Restricted Project
vpykhtin added a comment to D68338: [AMDGPU] Remove dubious logic in bidirectional list scheduler.

Sorry, I missed your comment about tests.

Oct 2 2019, 8:22 AM · Restricted Project
vpykhtin added a comment to D68338: [AMDGPU] Remove dubious logic in bidirectional list scheduler.

It would be nice to have a test for this.

Oct 2 2019, 8:22 AM · Restricted Project

Oct 1 2019

vpykhtin accepted D68264: [AMDGPU] Add VerifyScheduling support..

LGTM.

Oct 1 2019, 7:28 AM · Restricted Project
vpykhtin added a reviewer for D68264: [AMDGPU] Add VerifyScheduling support.: rampitec.
Oct 1 2019, 7:28 AM · Restricted Project

Sep 9 2019

vpykhtin accepted D67309: AMDGPU: Move MnemonicAlias out of instruction def hierarchy.

LGTM.

Sep 9 2019, 7:47 AM

Sep 6 2019

vpykhtin committed rGe8ade89bb380: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores (authored by vpykhtin).
[AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores
Sep 6 2019, 8:35 AM
vpykhtin committed rL371214: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.
[AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores
Sep 6 2019, 8:35 AM
vpykhtin closed D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.
Sep 6 2019, 8:34 AM · Restricted Project

Sep 5 2019

vpykhtin accepted D67101: Target hooks for custom COPY insertion..

LGTM.

Sep 5 2019, 7:06 AM · Restricted Project
vpykhtin added inline comments to D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.
Sep 5 2019, 5:22 AM · Restricted Project

Sep 4 2019

vpykhtin added inline comments to D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.
Sep 4 2019, 8:03 AM · Restricted Project
vpykhtin added inline comments to D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.
Sep 4 2019, 6:58 AM · Restricted Project

Sep 3 2019

vpykhtin added inline comments to D67101: Target hooks for custom COPY insertion..
Sep 3 2019, 10:20 AM · Restricted Project
vpykhtin added a comment to D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.

ping

Sep 3 2019, 9:20 AM · Restricted Project

Sep 2 2019

vpykhtin added a comment to D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.

llvm-check is now passing, previous failures were irrelevant to this patch.

Sep 2 2019, 8:01 AM · Restricted Project

Aug 30 2019

vpykhtin updated the diff for D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.

Added test. There're some failures in GlobalISel tests need to check if its connected.

Aug 30 2019, 10:17 AM · Restricted Project

Aug 29 2019

vpykhtin created D66958: [AMDGPU] Enable constant offset promotion to immediate operand for VMEM stores.
Aug 29 2019, 9:41 AM · Restricted Project

Aug 23 2019

vpykhtin accepted D66597: [preprocessor] Add an opportunity to retain excluded conditional blocks.

I'm not an expert in this part of the compiler, but looks good.

Aug 23 2019, 8:12 AM · Restricted Project, Restricted Project

Aug 1 2019

vpykhtin accepted D65496: AMDGPU/SILoadStoreOptimizer: Add helper functions for working with CombineInfo.

LGTM, with tips :)

Aug 1 2019, 8:12 AM · Restricted Project

Jul 22 2019

vpykhtin added a comment to D65081: AMDGPU: Start redefining atomic PatFrags.

I'm not sure I can follow all the consequences of this change, but as for tablegen usage it looks good.

Jul 22 2019, 6:18 AM

Jul 12 2019

vpykhtin added a comment to D64393: [AMDGPU] Fix DPP combiner check for exec modification.

Looks good, thank you!

Jul 12 2019, 8:52 AM · Restricted Project
vpykhtin added inline comments to D64393: [AMDGPU] Fix DPP combiner check for exec modification.
Jul 12 2019, 8:41 AM · Restricted Project
vpykhtin added inline comments to D64393: [AMDGPU] Fix DPP combiner check for exec modification.
Jul 12 2019, 8:20 AM · Restricted Project
vpykhtin accepted D64393: [AMDGPU] Fix DPP combiner check for exec modification.

I think we can submit this.

Jul 12 2019, 7:06 AM · Restricted Project

Jul 11 2019

vpykhtin added inline comments to D64393: [AMDGPU] Fix DPP combiner check for exec modification.
Jul 11 2019, 5:43 AM · Restricted Project
vpykhtin added inline comments to D64393: [AMDGPU] Fix DPP combiner check for exec modification.
Jul 11 2019, 4:39 AM · Restricted Project
vpykhtin added inline comments to D64393: [AMDGPU] Fix DPP combiner check for exec modification.
Jul 11 2019, 4:36 AM · Restricted Project

Jul 10 2019

vpykhtin added a comment to D64393: [AMDGPU] Fix DPP combiner check for exec modification.

But is it guaranteed that multiple uses by a single instruction will be adjacent in the list?

Jul 10 2019, 6:18 AM · Restricted Project
vpykhtin added a comment to D64393: [AMDGPU] Fix DPP combiner check for exec modification.

Sorry, there is a code for returnining unique instr, so we can use the count:

Jul 10 2019, 6:00 AM · Restricted Project
vpykhtin added a comment to D64393: [AMDGPU] Fix DPP combiner check for exec modification.

Yea, I thought there is a guaranty on unique instructions, but looking into defusechain_instr_iterator I don't see how this can be true, so let's continue using the set.

Jul 10 2019, 5:49 AM · Restricted Project
vpykhtin added a comment to D64393: [AMDGPU] Fix DPP combiner check for exec modification.

Matt, I'm not against using scan limiter, just about aborting the scan after the last use.

Jul 10 2019, 5:29 AM · Restricted Project

Jul 9 2019

vpykhtin added a comment to D64393: [AMDGPU] Fix DPP combiner check for exec modification.

I missed r363675 commit and I think the original semantics of isEXECMaskConstantBetweenDefAndUses should be restored: there should be no scan after the last use.

Jul 9 2019, 6:48 AM · Restricted Project

Jul 5 2019

vpykhtin accepted D64207: [AMDGPU] DPP combiner: recognize identities for more opcodes.

LGTM. Thank you!

Jul 5 2019, 4:16 AM · Restricted Project

Jul 4 2019

vpykhtin added a comment to D64207: [AMDGPU] DPP combiner: recognize identities for more opcodes.

Right, this is hard to follow even for me :). 3rd operand is src1_modifiers, you can use a junk value for this to check whether the DPP combiner don't crash and don't combine it.

Jul 4 2019, 9:19 AM · Restricted Project
vpykhtin added a reviewer for D64207: [AMDGPU] DPP combiner: recognize identities for more opcodes: cwabbott.
Jul 4 2019, 8:19 AM · Restricted Project
vpykhtin added a comment to D64207: [AMDGPU] DPP combiner: recognize identities for more opcodes.

I think modifiers are checked correctly by the existing code, but can you add a test for e64 encodings into dpp_combine.mir similar to what is under "check for floating point modifiers" comment?

Jul 4 2019, 8:06 AM · Restricted Project
vpykhtin added a comment to D64207: [AMDGPU] DPP combiner: recognize identities for more opcodes.

I'm not sure if e64 instructions have modifiers that cannot be encoded into DPP version, need to check. Otherwise looks good, though I would split typo corrections into separate patch and submit without review.

Jul 4 2019, 7:45 AM · Restricted Project

Jun 18 2019

vpykhtin committed rG7e854e1cdd23: [AMDGPU] Speed up live-in virtual register set computaion in… (authored by vpykhtin).
[AMDGPU] Speed up live-in virtual register set computaion in…
Jun 18 2019, 4:41 AM
vpykhtin committed rL363661: [AMDGPU] Speed up live-in virtual register set computaion in….
[AMDGPU] Speed up live-in virtual register set computaion in…
Jun 18 2019, 4:40 AM
vpykhtin closed D62401: [AMDGPU] Speed up live-in virtual register set computaion in GCNScheduleDAGMILive.
Jun 18 2019, 4:40 AM · Restricted Project

Jun 17 2019

vpykhtin committed rGbf3c59f79bcc: LiveInterval.h: add LiveRange::findIndexesLiveAt function - return a list of… (authored by vpykhtin).
LiveInterval.h: add LiveRange::findIndexesLiveAt function - return a list of…
Jun 17 2019, 11:27 AM
vpykhtin committed rL363593: LiveInterval.h: add LiveRange::findIndexesLiveAt function - return a list of….
LiveInterval.h: add LiveRange::findIndexesLiveAt function - return a list of…
Jun 17 2019, 11:27 AM
vpykhtin closed D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..
Jun 17 2019, 11:27 AM · Restricted Project

Jun 14 2019

vpykhtin committed rGffeb01c113a5: [AMDGPU] Don't constrain callees with inlinehint from inlining on MaxBB check (authored by vpykhtin).
[AMDGPU] Don't constrain callees with inlinehint from inlining on MaxBB check
Jun 14 2019, 9:35 AM
vpykhtin committed rL363418: [AMDGPU] Don't constrain callees with inlinehint from inlining on MaxBB check.
[AMDGPU] Don't constrain callees with inlinehint from inlining on MaxBB check
Jun 14 2019, 9:34 AM
vpykhtin closed D63337: [AMDGPU] Don't constrain callees with inlinehint from inlining on MaxBB check to prevent link failure.
Jun 14 2019, 9:34 AM · Restricted Project
vpykhtin updated the diff for D63337: [AMDGPU] Don't constrain callees with inlinehint from inlining on MaxBB check to prevent link failure.

updated description

Jun 14 2019, 9:13 AM · Restricted Project
vpykhtin created D63337: [AMDGPU] Don't constrain callees with inlinehint from inlining on MaxBB check to prevent link failure.
Jun 14 2019, 8:54 AM · Restricted Project

Jun 7 2019

vpykhtin added a comment to D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..

ping

Jun 7 2019, 11:24 AM · Restricted Project
vpykhtin committed rGcb8de55f4749: [AMDGPU] Constrain the AMDGPU inliner on maximum number of basic blocks in a… (authored by vpykhtin).
[AMDGPU] Constrain the AMDGPU inliner on maximum number of basic blocks in a…
Jun 7 2019, 5:14 AM
vpykhtin committed rL362789: [AMDGPU] Constrain the AMDGPU inliner on maximum number of basic blocks in a….
[AMDGPU] Constrain the AMDGPU inliner on maximum number of basic blocks in a…
Jun 7 2019, 5:14 AM
vpykhtin closed D62917: [AMDGPU] Constrain the inliner on maximum number of basic blocks in a caller function.
Jun 7 2019, 5:14 AM · Restricted Project

Jun 5 2019

vpykhtin updated the diff for D62917: [AMDGPU] Constrain the inliner on maximum number of basic blocks in a caller function.

updated diff with per review issues done.

Jun 5 2019, 11:33 PM · Restricted Project
vpykhtin created D62917: [AMDGPU] Constrain the inliner on maximum number of basic blocks in a caller function.
Jun 5 2019, 10:04 AM · Restricted Project

May 31 2019

vpykhtin added a comment to D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..

Hi, this is the parent revision for the https://reviews.llvm.org/D62401.

May 31 2019, 3:21 AM · Restricted Project

May 29 2019

vpykhtin updated the diff for D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..
  • fixed bug when an index was considered in-segment at the segment's end in some cases (lower_bound used instead of upper)
  • added fast (binary search) skip for non-containing segments
  • improved variable naming
May 29 2019, 10:29 AM · Restricted Project

May 28 2019

vpykhtin added a comment to D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..

this implementation contains bug with upper segment boundary and misses opportunity to search on sorted segments too, fixing.

May 28 2019, 5:34 AM · Restricted Project

May 27 2019

vpykhtin updated the diff for D62401: [AMDGPU] Speed up live-in virtual register set computaion in GCNScheduleDAGMILive.

replaced std::vector with SmallVector, moved out of the loop.

May 27 2019, 9:02 AM · Restricted Project
vpykhtin updated the diff for D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..
  • replaced output std::vector with template output iterator, return bool on found
  • added is_sorted assert
  • improved description
May 27 2019, 9:02 AM · Restricted Project

May 24 2019

vpykhtin added inline comments to D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..
May 24 2019, 11:22 AM · Restricted Project
vpykhtin added a reviewer for D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at.: arsenm.
May 24 2019, 11:01 AM · Restricted Project
vpykhtin added reviewers for D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at.: rampitec, qcolombet, MatzeB.
May 24 2019, 11:01 AM · Restricted Project
vpykhtin added a child revision for D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at.: D62401: [AMDGPU] Speed up live-in virtual register set computaion in GCNScheduleDAGMILive.
May 24 2019, 10:52 AM · Restricted Project
vpykhtin added a parent revision for D62401: [AMDGPU] Speed up live-in virtual register set computaion in GCNScheduleDAGMILive: D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..
May 24 2019, 10:52 AM · Restricted Project
vpykhtin created D62411: LiveIntervals: add LiveRange::findIndexesLiveAt function - return a list of SlotIndexes the LiveRange live at..
May 24 2019, 10:52 AM · Restricted Project
vpykhtin updated the diff for D62401: [AMDGPU] Speed up live-in virtual register set computaion in GCNScheduleDAGMILive.

split LiveInterval.h change into different patch

May 24 2019, 10:49 AM · Restricted Project
vpykhtin created D62401: [AMDGPU] Speed up live-in virtual register set computaion in GCNScheduleDAGMILive.
May 24 2019, 9:06 AM · Restricted Project

Mar 14 2019

vpykhtin accepted D59191: [AMDGPU] Asm/disasm v_cndmask_b32_e64 with abs/neg source modifiers.

LGTM, Thanks!

Mar 14 2019, 6:08 AM · Restricted Project

Feb 12 2019

vpykhtin added a comment to D55444: AMDGPU: Fix DPP combiner.

The DPP combiner pass is enabled since rL353691, https://reviews.llvm.org/rGded96df01e95

Feb 12 2019, 2:58 AM · Restricted Project, Restricted Project

Feb 11 2019

vpykhtin committed rGe1c338e5270a: [AMDGPU] fix atomic_optimizations_buffer.ll test after DPP combiner was enabled… (authored by vpykhtin).
[AMDGPU] fix atomic_optimizations_buffer.ll test after DPP combiner was enabled…
Feb 11 2019, 8:29 AM
vpykhtin committed rL353717: [AMDGPU] fix atomic_optimizations_buffer.ll test after DPP combiner was enabled….
[AMDGPU] fix atomic_optimizations_buffer.ll test after DPP combiner was enabled…
Feb 11 2019, 8:28 AM
vpykhtin committed rGded96df01e95: [AMDGPU] Enable DPP combiner pass by default. (authored by vpykhtin).
[AMDGPU] Enable DPP combiner pass by default.
Feb 11 2019, 3:15 AM
vpykhtin committed rL353691: [AMDGPU] Enable DPP combiner pass by default..
[AMDGPU] Enable DPP combiner pass by default.
Feb 11 2019, 3:15 AM

Feb 8 2019

vpykhtin committed rG7fe97f8c7ca1: [AMDGPU] Fix DPP combiner (authored by vpykhtin).
[AMDGPU] Fix DPP combiner
Feb 8 2019, 4:03 AM
vpykhtin added a comment to D55444: AMDGPU: Fix DPP combiner.

Thank you Connor! I really appreciate your effort on this DPP work.

Feb 8 2019, 4:03 AM · Restricted Project, Restricted Project