Page MenuHomePhabricator

dfukalov (Daniil Fukalov)
Compiler Engineer at AMD

Projects

User does not belong to any projects.

User Details

User Since
Mar 27 2014, 8:40 AM (362 w, 4 d)

Recent Activity

Sat, Mar 6

dfukalov added inline comments to D98027: [NFC][AA] Convert AliasResult to scoped enumeration..
Sat, Mar 6, 7:42 AM · Restricted Project

Fri, Mar 5

dfukalov added a comment to D98027: [NFC][AA] Convert AliasResult to scoped enumeration..

No, I don't have a patch at the moment, but will upload one since you're generally ok with this step.

Fri, Mar 5, 9:46 AM · Restricted Project
dfukalov updated subscribers of D98027: [NFC][AA] Convert AliasResult to scoped enumeration..
Fri, Mar 5, 3:10 AM · Restricted Project
dfukalov requested review of D98027: [NFC][AA] Convert AliasResult to scoped enumeration..
Fri, Mar 5, 3:06 AM · Restricted Project

Thu, Mar 4

dfukalov committed rG9899427174b4: [NFC][AliasSetTracker] Remove implicit conversion AliasResult to integer. (authored by dfukalov).
[NFC][AliasSetTracker] Remove implicit conversion AliasResult to integer.
Thu, Mar 4, 1:54 PM
dfukalov closed D97973: [NFC][AliasSetTracker] Remove implicit conversion AliasResult to integer..
Thu, Mar 4, 1:53 PM · Restricted Project
dfukalov requested review of D97973: [NFC][AliasSetTracker] Remove implicit conversion AliasResult to integer..
Thu, Mar 4, 12:18 PM · Restricted Project

Wed, Mar 3

dfukalov updated subscribers of D95543: [GVN] Clobber partially aliased loads..
Wed, Mar 3, 8:40 AM · Restricted Project
dfukalov retitled D95543: [GVN] Clobber partially aliased loads. from [WIP][GVN] Clobber partially aliased loads. to [GVN] Clobber partially aliased loads..
Wed, Mar 3, 8:39 AM · Restricted Project
dfukalov updated the diff for D95543: [GVN] Clobber partially aliased loads..

Implementation.

Wed, Mar 3, 8:33 AM · Restricted Project

Tue, Mar 2

dfukalov added a comment to D97676: [DSE] Extending isOverwrite to support offsetted fully overlapping stores.

I noticed @dfukalov planned to fix the DSE after their patch, so I'm not sure if I should delete mine, as their fixes/improvements may be covering more cases compared to my isolated test file.

I planned to work on DSE after GVN (eliminating redundant loads).
It seems these two fixes may be independent, but you can check my GVN approach in the diff: https://reviews.llvm.org/D93529?id=313082.
My test case for DSE fix is

define void @foo(float* %arg, i32 %i) {
bb:
  %i7 = add nuw nsw i32 %i, 1
  %i8 = zext i32 %i7 to i64
  %i9 = getelementptr inbounds float, float* %arg, i64 %i8
  store float undef, float* %i9, align 4
Tue, Mar 2, 8:22 AM · Restricted Project
dfukalov committed rG6e967834b9fa: [AA] Cache (optionally) estimated PartialAlias offsets. (authored by dfukalov).
[AA] Cache (optionally) estimated PartialAlias offsets.
Tue, Mar 2, 8:04 AM
dfukalov closed D93529: [AA] Cache (optionally) estimated PartialAlias offsets..
Tue, Mar 2, 8:04 AM · Restricted Project
dfukalov updated the diff for D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Rebased, addressed comments.

Tue, Mar 2, 7:39 AM · Restricted Project

Mon, Mar 1

dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

@asbirlea, @nikic, would you please take a look at the updated patch?

Mon, Mar 1, 10:00 AM · Restricted Project

Wed, Feb 17

dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Ping...

Wed, Feb 17, 11:44 PM · Restricted Project

Tue, Feb 16

dfukalov added a comment to D96805: [AMDGPU][CostModel] Refine cost model for control-flow instructions..

It seems to me this threshold bump partially compensated by cbr cost increase in all cases of unroll loops with ifs, where it is multiplicated by trip count.
This threshold bumped because of test/CodeGen/AMDGPU/unroll.ll, where started to fail

Tue, Feb 16, 3:38 PM · Restricted Project
dfukalov requested review of D96805: [AMDGPU][CostModel] Refine cost model for control-flow instructions..
Tue, Feb 16, 11:21 AM · Restricted Project

Wed, Feb 10

dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Ping...

Wed, Feb 10, 9:23 AM · Restricted Project

Feb 3 2021

dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Gentle ping...

Feb 3 2021, 3:08 AM · Restricted Project

Feb 1 2021

dfukalov added inline comments to D94363: [AA] Pass AAResults through AAQueryInfo.
Feb 1 2021, 5:01 PM · Restricted Project

Jan 31 2021

dfukalov added inline comments to D95748: AMDGPU: Fix dbg_value handling when forming soft clause bundles.
Jan 31 2021, 3:52 PM · Restricted Project

Jan 27 2021

dfukalov retitled D93529: [AA] Cache (optionally) estimated PartialAlias offsets. from [AA] Store and return estimated PartialAlias offsets. to [AA] Cache (optionally) estimated PartialAlias offsets..
Jan 27 2021, 11:32 AM · Restricted Project
dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

... Could you (re-)upload a concrete use case as dependent patch?

I've just created D95543 with the test for upcoming changes in GVN. It is actually the same code as one used in unit test in this patch.

Jan 27 2021, 11:30 AM · Restricted Project
dfukalov updated the diff for D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Added memory sizes to map key, BatchAA can be created without caching.

Jan 27 2021, 11:21 AM · Restricted Project
dfukalov planned changes to D95543: [GVN] Clobber partially aliased loads..
Jan 27 2021, 10:18 AM · Restricted Project
dfukalov requested review of D95543: [GVN] Clobber partially aliased loads..
Jan 27 2021, 10:17 AM · Restricted Project

Jan 22 2021

dfukalov added inline comments to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..
Jan 22 2021, 10:49 AM · Restricted Project
dfukalov updated the diff for D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Comments addressed.

Jan 22 2021, 10:47 AM · Restricted Project

Jan 20 2021

dfukalov committed rG560d7e04113b: [NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets (authored by dfukalov).
[NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets
Jan 20 2021, 11:23 AM
dfukalov closed D95036: [NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets.
Jan 20 2021, 11:23 AM · Restricted Project
dfukalov retitled D95036: [NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets from [NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets. to [NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets.
Jan 20 2021, 7:55 AM · Restricted Project
dfukalov added a comment to D95036: [NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets.

The main reason was reducing include dependencies and so source files re-builds. At the moment both GCN and R600 targets rebuild if one just changes one option in GCN related code. Actually this is the first part of split, I was thinking about splitting AMDGPUSubtarget.cpp then, and following AMDGPUMCTargetDesc.h (it currently contains both GCN+R600 tablegen output).

Jan 20 2021, 7:52 AM · Restricted Project
dfukalov requested review of D95036: [NFC][AMDGPU] Split AMDGPUSubtarget.h to R600 and GCN subtargets.
Jan 20 2021, 4:59 AM · Restricted Project

Jan 19 2021

dfukalov retitled D93529: [AA] Cache (optionally) estimated PartialAlias offsets. from [GVN][BasicAA] Enable clobbering in GVN. to [AA] Store and return estimated PartialAlias offsets..
Jan 19 2021, 2:55 PM · Restricted Project
dfukalov updated the summary of D93529: [AA] Cache (optionally) estimated PartialAlias offsets..
Jan 19 2021, 2:53 PM · Restricted Project
dfukalov updated the diff for D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

AA part of the splitted patch.

Jan 19 2021, 2:49 PM · Restricted Project

Jan 18 2021

dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

IIUC this is similar to the handling of partial overwrites in DSE (https://github.com/llvm/llvm-project/blob/master/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp#L548)?

If so, could we just generalize the helper there, rather than threading this through the different caches from BasicAA to GVN? Yes, we have to do some extra work in GVN, but there would be no need to maintain a cached value not used by other clients, it will work with any AA implementations returning partial alias & there will be no extra work when moving away from MemDepAnalysis.

Jan 18 2021, 3:56 PM · Restricted Project
dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Ping. I'm going to submit the change on Friday if no objections are observed.

Jan 18 2021, 1:28 AM · Restricted Project

Jan 13 2021

dfukalov added reviewers for D93529: [AA] Cache (optionally) estimated PartialAlias offsets.: jdoerfert, hliao, reames.
Jan 13 2021, 4:48 AM · Restricted Project

Jan 7 2021

dfukalov committed rG6a87e9b08bf0: [NFC][AMDGPU] Reduce include files dependency. (authored by dfukalov).
[NFC][AMDGPU] Reduce include files dependency.
Jan 7 2021, 11:22 AM
dfukalov closed D93813: [NFC][AMDGPU] Reduce include files dependency..
Jan 7 2021, 11:22 AM · Restricted Project

Jan 6 2021

dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Ping...

Jan 6 2021, 6:14 AM · Restricted Project
dfukalov added inline comments to D93813: [NFC][AMDGPU] Reduce include files dependency..
Jan 6 2021, 6:14 AM · Restricted Project
dfukalov updated the diff for D93813: [NFC][AMDGPU] Reduce include files dependency..

Fixed include in wrong line, rebased.

Jan 6 2021, 6:13 AM · Restricted Project

Dec 28 2020

dfukalov added a comment to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Ping...

Dec 28 2020, 12:49 AM · Restricted Project

Dec 25 2020

dfukalov requested review of D93813: [NFC][AMDGPU] Reduce include files dependency..
Dec 25 2020, 7:53 AM · Restricted Project

Dec 23 2020

dfukalov added inline comments to D93685: [MIRPrinter] Fix incorrect output of unnamed stack names.
Dec 23 2020, 4:53 PM · Restricted Project
dfukalov added inline comments to D93685: [MIRPrinter] Fix incorrect output of unnamed stack names.
Dec 23 2020, 3:38 AM · Restricted Project

Dec 22 2020

dfukalov added a comment to D93685: [MIRPrinter] Fix incorrect output of unnamed stack names.

It seems patch should include update for alloca-crspill.ll test (failed Unit Tests precheckin).

Dec 22 2020, 7:12 AM · Restricted Project

Dec 21 2020

dfukalov added inline comments to D93529: [AA] Cache (optionally) estimated PartialAlias offsets..
Dec 21 2020, 4:41 AM · Restricted Project
dfukalov updated the diff for D93529: [AA] Cache (optionally) estimated PartialAlias offsets..

Rebased, change reworked as requested.

Dec 21 2020, 4:34 AM · Restricted Project

Dec 18 2020

dfukalov requested review of D93529: [AA] Cache (optionally) estimated PartialAlias offsets..
Dec 18 2020, 2:42 AM · Restricted Project

Dec 17 2020

dfukalov committed rG9ed8e0caab9b: [NFC] Reduce include files dependency and AA header cleanup (part 2). (authored by dfukalov).
[NFC] Reduce include files dependency and AA header cleanup (part 2).
Dec 17 2020, 3:05 AM
dfukalov closed D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..
Dec 17 2020, 3:05 AM · Restricted Project, Restricted Project

Dec 16 2020

dfukalov added a comment to D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..

Ping...

Dec 16 2020, 3:20 AM · Restricted Project, Restricted Project

Dec 9 2020

dfukalov added inline comments to D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..
Dec 9 2020, 7:23 AM · Restricted Project, Restricted Project
dfukalov updated the diff for D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..
Dec 9 2020, 6:08 AM · Restricted Project, Restricted Project
dfukalov updated the diff for D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..

Splitting change as requested.

Dec 9 2020, 5:57 AM · Restricted Project, Restricted Project

Dec 8 2020

dfukalov added inline comments to D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..
Dec 8 2020, 11:33 AM · Restricted Project, Restricted Project
dfukalov added inline comments to D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..
Dec 8 2020, 8:19 AM · Restricted Project, Restricted Project
dfukalov requested review of D92852: [NFC] Reduce include files dependency and AA header cleanup (part 2)..
Dec 8 2020, 8:14 AM · Restricted Project, Restricted Project

Dec 3 2020

dfukalov committed rG2ce38b3f0384: [NFC] Reduce include files dependency. (authored by dfukalov).
[NFC] Reduce include files dependency.
Dec 3 2020, 7:25 AM
dfukalov closed D92489: [NFC] Reduce include files dependency..
Dec 3 2020, 7:25 AM · Restricted Project

Dec 2 2020

dfukalov updated the diff for D92489: [NFC] Reduce include files dependency..

Addressed comment, fixed polly build, fixed clang-tidy comments.

Dec 2 2020, 12:07 PM · Restricted Project
dfukalov requested review of D92489: [NFC] Reduce include files dependency..
Dec 2 2020, 9:07 AM · Restricted Project

Oct 29 2020

dfukalov committed rGb3cdaef518ad: [MIR] Fix out of bounds access in MIRPrinter. (authored by dfukalov).
[MIR] Fix out of bounds access in MIRPrinter.
Oct 29 2020, 4:35 AM
dfukalov closed D90239: [MIR] Fix out of bounds access in MIRPrinter..
Oct 29 2020, 4:35 AM · Restricted Project
dfukalov updated the diff for D90239: [MIR] Fix out of bounds access in MIRPrinter..

Addressed comments.

Oct 29 2020, 4:34 AM · Restricted Project

Oct 28 2020

dfukalov added inline comments to D90239: [MIR] Fix out of bounds access in MIRPrinter..
Oct 28 2020, 11:29 AM · Restricted Project
dfukalov updated the diff for D90239: [MIR] Fix out of bounds access in MIRPrinter..

Addressed comments.

Oct 28 2020, 11:22 AM · Restricted Project

Oct 27 2020

dfukalov requested review of D90239: [MIR] Fix out of bounds access in MIRPrinter..
Oct 27 2020, 8:23 AM · Restricted Project

Oct 24 2020

dfukalov committed rG9068c209655e: [AMDGPU][CostModel] Refine cost model for half- and quarter-rate instructions. (authored by dfukalov).
[AMDGPU][CostModel] Refine cost model for half- and quarter-rate instructions.
Oct 24 2020, 9:53 AM
dfukalov closed D89973: [AMDGPU][CostModel] Refine cost model for half- and quarter-rate instructions..
Oct 24 2020, 9:53 AM · Restricted Project

Oct 22 2020

dfukalov requested review of D89973: [AMDGPU][CostModel] Refine cost model for half- and quarter-rate instructions..
Oct 22 2020, 10:24 AM · Restricted Project

Sep 9 2020

dfukalov committed rGc259d3a061c8: [AMDGPU] Fix for folding v2.16 literals. (authored by dfukalov).
[AMDGPU] Fix for folding v2.16 literals.
Sep 9 2020, 3:40 PM
dfukalov closed D87158: [AMDGPU] Fix for folding v2.16 literals..
Sep 9 2020, 3:39 PM · Restricted Project
dfukalov added inline comments to D87158: [AMDGPU] Fix for folding v2.16 literals..
Sep 9 2020, 7:29 AM · Restricted Project

Sep 8 2020

dfukalov added inline comments to D87158: [AMDGPU] Fix for folding v2.16 literals..
Sep 8 2020, 11:45 AM · Restricted Project

Sep 5 2020

dfukalov added inline comments to D87158: [AMDGPU] Fix for folding v2.16 literals..
Sep 5 2020, 3:29 PM · Restricted Project

Sep 4 2020

dfukalov requested review of D87158: [AMDGPU] Fix for folding v2.16 literals..
Sep 4 2020, 12:55 PM · Restricted Project

Aug 20 2020

dfukalov added a comment to D86262: [LoopIdiomRecognizePass] Options to disable part or the entire Loop Idiom Recognize Pass.

Why is this the correct course of action?
For example, if the IR already had memcpy/memset, the DA will also be pessimized,
so it really seems like a workaround, not a fix.
Why not just enhance DA?

Aug 20 2020, 1:00 AM · Restricted Project
dfukalov committed rG33e2f69a24b4: [AMDGPU][LoopUnroll] Increase BB size to analyze for complete unroll. (authored by dfukalov).
[AMDGPU][LoopUnroll] Increase BB size to analyze for complete unroll.
Aug 20 2020, 12:42 AM
dfukalov closed D86248: [AMDGPU][LoopUnroll] Increase BB size to analyze for complete unroll..
Aug 20 2020, 12:42 AM · Restricted Project

Aug 19 2020

dfukalov requested review of D86248: [AMDGPU][LoopUnroll] Increase BB size to analyze for complete unroll..
Aug 19 2020, 2:15 PM · Restricted Project

Aug 13 2020

dfukalov abandoned D37438: Fix segfault in FlattenCFG.
Aug 13 2020, 9:12 AM
dfukalov accepted D85829: [NFCI][SimpleLoopUnswitch] Adjust CostKind query.

Please clang-format it.

Aug 13 2020, 7:11 AM · Restricted Project

Aug 6 2020

dfukalov committed rG4ccc38813eb7: [AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation. (authored by dfukalov).
[AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation.
Aug 6 2020, 11:44 AM
dfukalov closed D84995: [AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation..
Aug 6 2020, 11:43 AM · Restricted Project
dfukalov added inline comments to D84995: [AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation..
Aug 6 2020, 9:20 AM · Restricted Project
dfukalov updated the diff for D84995: [AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation..

Check for hasMadMacF32Insts() added.

Aug 6 2020, 2:35 AM · Restricted Project

Aug 5 2020

dfukalov updated the diff for D84995: [AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation..

Change updated with addressed comments.

Aug 5 2020, 3:09 PM · Restricted Project

Jul 31 2020

dfukalov added inline comments to D84995: [AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation..
Jul 31 2020, 9:02 AM · Restricted Project

Jul 30 2020

dfukalov committed rGaa77232a63e3: [NFC][AMDGPU] Improve fused fmul+fadd tests. (authored by dfukalov).
[NFC][AMDGPU] Improve fused fmul+fadd tests.
Jul 30 2020, 6:00 PM
dfukalov closed D84903: [NFC][AMDGPU] Improve fused fmul+fadd tests..
Jul 30 2020, 6:00 PM · Restricted Project
dfukalov requested review of D84995: [AMDGPU][CostModel] Add f16, f64 and contract cases to fused costs estimation..
Jul 30 2020, 5:57 PM · Restricted Project
dfukalov added inline comments to D84903: [NFC][AMDGPU] Improve fused fmul+fadd tests..
Jul 30 2020, 3:53 AM · Restricted Project

Jul 29 2020

dfukalov requested review of D84903: [NFC][AMDGPU] Improve fused fmul+fadd tests..
Jul 29 2020, 6:13 PM · Restricted Project
dfukalov added a comment to D84069: [NFC][PPC][AIX] Add test coverage for _Complex return values.

Hi @cebowleratibm, it seems test is located in wrong place: test/CodeGen/PowerPC instead of llvm/test/CodeGen/PowerPC.

Jul 29 2020, 3:38 PM · Restricted Project

Jul 15 2020

dfukalov committed rG76a0c0ee6ffa: [AMDGPU][CostModel] Improve cost estimation for fused {fadd|fsub}(a,fmul(b,c)) (authored by dfukalov).
[AMDGPU][CostModel] Improve cost estimation for fused {fadd|fsub}(a,fmul(b,c))
Jul 15 2020, 5:07 PM