Page MenuHomePhabricator

kzhuravl (Konstantin Zhuravlyov)
User

Projects

User does not belong to any projects.

User Details

User Since
Feb 16 2016, 12:13 PM (161 w, 4 d)

Recent Activity

Fri, Mar 22

kzhuravl created D59718: AMDGPU: Make sram-ecc off by default for Vega20.
Fri, Mar 22, 2:26 PM
kzhuravl created D59717: AMDGPU: Ignore SRAM ECC feature when inlining.
Fri, Mar 22, 2:08 PM

Wed, Mar 20

kzhuravl added a comment to D59614: [AMDGPU] Add missing dependency on BinaryFormat.

Dup of D59610 ?

Wed, Mar 20, 2:30 PM · Restricted Project
kzhuravl accepted D59613: [AMDGPU] Do not generate spurious PAL metadata.

LGTM

Wed, Mar 20, 2:14 PM · Restricted Project
kzhuravl committed rG88268e3e3607: AMDHSA: Fix COMPUTE_PGM_RSRC2.USER_SGPR calculation when parsing ISA assembly (authored by kzhuravl).
AMDHSA: Fix COMPUTE_PGM_RSRC2.USER_SGPR calculation when parsing ISA assembly
Wed, Mar 20, 12:44 PM
kzhuravl committed rL356603: AMDHSA: Fix COMPUTE_PGM_RSRC2.USER_SGPR calculation when parsing ISA assembly.
AMDHSA: Fix COMPUTE_PGM_RSRC2.USER_SGPR calculation when parsing ISA assembly
Wed, Mar 20, 12:43 PM
kzhuravl closed D59570: AMDHSA: Fix COMPUTE_PGM_RSRC2.USER_SGPR calculation when parsing ISA assembly.
Wed, Mar 20, 12:43 PM · Restricted Project

Tue, Mar 19

kzhuravl created D59570: AMDHSA: Fix COMPUTE_PGM_RSRC2.USER_SGPR calculation when parsing ISA assembly.
Tue, Mar 19, 5:14 PM · Restricted Project
kzhuravl updated the diff for D59517: AMDGPU: Add support for cross address space synchronization scopes.

Added docs and tests. Addressed review feedback.

Tue, Mar 19, 1:34 PM

Mon, Mar 18

kzhuravl added a child revision for D59517: AMDGPU: Add support for cross address space synchronization scopes: D59494: AMDGPU: Add support for cross address space synchronization scopes (clang).
Mon, Mar 18, 3:38 PM
kzhuravl added a parent revision for D59494: AMDGPU: Add support for cross address space synchronization scopes (clang): D59517: AMDGPU: Add support for cross address space synchronization scopes.
Mon, Mar 18, 3:38 PM
kzhuravl created D59517: AMDGPU: Add support for cross address space synchronization scopes.
Mon, Mar 18, 3:38 PM
kzhuravl added a comment to D59517: AMDGPU: Add support for cross address space synchronization scopes.

This still requires some more tests to be added (which I am working on), and docs updated. I just wanted to get this out here for some early code reviews and testing on hw.

Mon, Mar 18, 3:38 PM
kzhuravl retitled D59494: AMDGPU: Add support for cross address space synchronization scopes (clang) from AMDGPU: Add support for cross address space synchronization scopes to AMDGPU: Add support for cross address space synchronization scopes (clang).
Mon, Mar 18, 3:38 PM
kzhuravl added inline comments to D59494: AMDGPU: Add support for cross address space synchronization scopes (clang).
Mon, Mar 18, 3:38 PM
kzhuravl updated the diff for D59494: AMDGPU: Add support for cross address space synchronization scopes (clang).

Address review feedback.

Mon, Mar 18, 3:38 PM
kzhuravl added a comment to D59494: AMDGPU: Add support for cross address space synchronization scopes (clang).

Backend change will be posted later today. I am currently working on adding backend tests.

Mon, Mar 18, 8:29 AM
kzhuravl created D59494: AMDGPU: Add support for cross address space synchronization scopes (clang).
Mon, Mar 18, 8:29 AM

Wed, Mar 13

kzhuravl accepted D59330: [AMDGPU] Silence gcc 7 warnings.

LGTM

Wed, Mar 13, 1:59 PM · Restricted Project

Thu, Mar 7

kzhuravl committed rG47f0bf8f1fa0: AMDHSA: Code object v3 updates (authored by kzhuravl).
AMDHSA: Code object v3 updates
Thu, Mar 7, 11:59 AM
kzhuravl committed rL355630: AMDHSA: Code object v3 updates.
AMDHSA: Code object v3 updates
Thu, Mar 7, 11:59 AM
kzhuravl closed D59057: AMDHSA: Code object v3 updates.
Thu, Mar 7, 11:59 AM · Restricted Project
kzhuravl updated the diff for D59057: AMDHSA: Code object v3 updates.

Address review feedback.

Thu, Mar 7, 9:48 AM · Restricted Project

Wed, Mar 6

kzhuravl created D59057: AMDHSA: Code object v3 updates.
Wed, Mar 6, 3:07 PM · Restricted Project
kzhuravl committed rG3161c89a22f9: AMDGPU: Fix the mapping of sub group sync scope (authored by kzhuravl).
AMDGPU: Fix the mapping of sub group sync scope
Wed, Mar 6, 12:54 PM
kzhuravl committed rL355549: AMDGPU: Fix the mapping of sub group sync scope.
AMDGPU: Fix the mapping of sub group sync scope
Wed, Mar 6, 12:54 PM
kzhuravl committed rC355549: AMDGPU: Fix the mapping of sub group sync scope.
AMDGPU: Fix the mapping of sub group sync scope
Wed, Mar 6, 12:54 PM
kzhuravl closed D58847: AMDGPU: Fix the mapping of sub group sync scope.
Wed, Mar 6, 12:54 PM · Restricted Project

Fri, Mar 1

kzhuravl created D58847: AMDGPU: Fix the mapping of sub group sync scope.
Fri, Mar 1, 2:06 PM · Restricted Project

Fri, Feb 22

kzhuravl committed rG9a278bf6b5f4: Revert "AMDGPU/NFC: Cleanup subtarget predicates" (authored by kzhuravl).
Revert "AMDGPU/NFC: Cleanup subtarget predicates"
Fri, Feb 22, 3:21 PM
kzhuravl committed rL354700: Revert "AMDGPU/NFC: Cleanup subtarget predicates".
Revert "AMDGPU/NFC: Cleanup subtarget predicates"
Fri, Feb 22, 3:20 PM

Feb 21 2019

kzhuravl accepted D58159: AMDGPU: Remove debugger related subtarget features.

LGTM

Feb 21 2019, 2:00 PM
kzhuravl accepted D58526: AMDGPU: Don't emit debugger subtarget features.

LGTM

Feb 21 2019, 1:04 PM
kzhuravl added a comment to D58159: AMDGPU: Remove debugger related subtarget features.

Can SIProgramInfo::DebuggerWavefrontPrivateSegmentOffsetSGPR and SIProgramInfo::DebuggerPrivateSegmentBufferSGPR be completely deleted?

Feb 21 2019, 12:57 PM
kzhuravl added a comment to D58159: AMDGPU: Remove debugger related subtarget features.

https://github.com/llvm-mirror/llvm/blob/master/lib/Target/AMDGPU/SIProgramInfo.h#L55
https://github.com/llvm-mirror/llvm/blob/master/lib/Target/AMDGPU/SIProgramInfo.h#L61

Feb 21 2019, 12:57 PM
kzhuravl committed rGc2650178a179: AMDGPU/NFC: Cleanup subtarget predicates (authored by kzhuravl).
AMDGPU/NFC: Cleanup subtarget predicates
Feb 21 2019, 12:44 PM
kzhuravl added inline comments to D58526: AMDGPU: Don't emit debugger subtarget features.
Feb 21 2019, 12:44 PM
kzhuravl committed rL354620: AMDGPU/NFC: Cleanup subtarget predicates.
AMDGPU/NFC: Cleanup subtarget predicates
Feb 21 2019, 12:43 PM
kzhuravl closed D58522: AMDGPU/NFC: Cleanup subtarget predicates.
Feb 21 2019, 12:43 PM · Restricted Project
kzhuravl added a comment to D58159: AMDGPU: Remove debugger related subtarget features.

I am in favor of removing these. Clang will also need to be updated around here:
https://github.com/llvm-mirror/clang/blob/master/lib/Driver/ToolChains/AMDGPU.cpp#L41

Feb 21 2019, 11:19 AM
kzhuravl created D58522: AMDGPU/NFC: Cleanup subtarget predicates.
Feb 21 2019, 11:09 AM · Restricted Project

Feb 14 2019

kzhuravl committed rG87498153aa12: LLD/AMDGPU: Preserve ABI version during linking ELF for AMDGPU (authored by kzhuravl).
LLD/AMDGPU: Preserve ABI version during linking ELF for AMDGPU
Feb 14 2019, 4:00 PM
kzhuravl committed rL354086: LLD/AMDGPU: Preserve ABI version during linking ELF for AMDGPU.
LLD/AMDGPU: Preserve ABI version during linking ELF for AMDGPU
Feb 14 2019, 3:59 PM
kzhuravl committed rLLD354086: LLD/AMDGPU: Preserve ABI version during linking ELF for AMDGPU.
LLD/AMDGPU: Preserve ABI version during linking ELF for AMDGPU
Feb 14 2019, 3:59 PM
kzhuravl closed D58026: LLD: Preserve ABI version during linking ELF.
Feb 14 2019, 3:59 PM · Restricted Project
kzhuravl committed rG1e126c503b8b: AMDGPU: Set ABI version to 1 for code object v3 (authored by kzhuravl).
AMDGPU: Set ABI version to 1 for code object v3
Feb 14 2019, 3:56 PM
kzhuravl committed rL354085: AMDGPU: Set ABI version to 1 for code object v3.
AMDGPU: Set ABI version to 1 for code object v3
Feb 14 2019, 3:55 PM
kzhuravl closed D57811: AMDGPU: Set ABI version to 1 for code object v3.
Feb 14 2019, 3:55 PM · Restricted Project
kzhuravl committed rGe0484eb2f2d0: MC/ELF: Allow targets to set ABI version (authored by kzhuravl).
MC/ELF: Allow targets to set ABI version
Feb 14 2019, 2:43 PM
kzhuravl committed rL354072: MC/ELF: Allow targets to set ABI version.
MC/ELF: Allow targets to set ABI version
Feb 14 2019, 2:43 PM
kzhuravl closed D57810: MC/ELF: Allow targets to set ABI version.
Feb 14 2019, 2:43 PM · Restricted Project

Feb 13 2019

kzhuravl added a comment to D57811: AMDGPU: Set ABI version to 1 for code object v3.

Just want to know a bit more about background of this change. Are you making an ABI-breaking change, so you are defining a new ABI version number for the AMDGPU target. Is this understanding correct? Is the AMDGPU target migrating to ABI version 2, or will version 1 and 2 co-exist in the future?

Feb 13 2019, 2:21 PM · Restricted Project

Feb 12 2019

kzhuravl added a comment to D57811: AMDGPU: Set ABI version to 1 for code object v3.

ping

Feb 12 2019, 4:45 PM · Restricted Project
kzhuravl added a comment to D57810: MC/ELF: Allow targets to set ABI version.

ping

Feb 12 2019, 4:45 PM · Restricted Project
kzhuravl updated the diff for D58026: LLD: Preserve ABI version during linking ELF.

Address review feedback.

Feb 12 2019, 2:51 PM · Restricted Project
kzhuravl committed rG6220d62e5c9c: AMDGPU/NFC: Remove SubtargetFeatureISAVersion since it is not used anywhere (authored by kzhuravl).
AMDGPU/NFC: Remove SubtargetFeatureISAVersion since it is not used anywhere
Feb 12 2019, 2:50 PM
kzhuravl committed rL353892: AMDGPU/NFC: Remove SubtargetFeatureISAVersion since it is not used anywhere.
AMDGPU/NFC: Remove SubtargetFeatureISAVersion since it is not used anywhere
Feb 12 2019, 2:49 PM
kzhuravl committed rGacb231c8d8f2: AMDGPU: Remove duplicate processor (gfx900) (authored by kzhuravl).
AMDGPU: Remove duplicate processor (gfx900)
Feb 12 2019, 2:30 PM
kzhuravl committed rL353889: AMDGPU: Remove duplicate processor (gfx900).
AMDGPU: Remove duplicate processor (gfx900)
Feb 12 2019, 2:30 PM
kzhuravl accepted D57977: [HIP] compile option code-object-v3 propagate to llc.

LGTM. Please make sure to update your commit message as your patch processes all m options now.

Feb 12 2019, 12:41 PM · Restricted Project
kzhuravl added a comment to D57977: [HIP] compile option code-object-v3 propagate to llc.

Is it still code object v2 by default?

Feb 12 2019, 12:36 PM · Restricted Project
kzhuravl added a comment to D57977: [HIP] compile option code-object-v3 propagate to llc.

Can you add some tests? Preferably with -mxnack/-mno-xnack, -msram-ecc/-mno-sram-ecc, -mcode-object-v3/-mno-code-object-v3

Feb 12 2019, 11:16 AM · Restricted Project

Feb 11 2019

kzhuravl added a comment to D58026: LLD: Preserve ABI version during linking ELF.

The interpretation of ABI version of an ELF object file depends on the target machine type. For example, we can think of a target that allows a mix of version X and version Y objects let a linker to create a shim between cross-version function call. At least most ABIs don't define anything about the ABI version, so it is perhaps most natural to ignore the field at all, unless you are linking for a specific target in which the ABI version has a special meaning. Therefore, I'd think code that works only for AMDGPU target is better.

Feb 11 2019, 3:03 PM · Restricted Project
kzhuravl updated subscribers of D58026: LLD: Preserve ABI version during linking ELF.
Feb 11 2019, 2:51 PM · Restricted Project
kzhuravl added a comment to D58026: LLD: Preserve ABI version during linking ELF.

I quickly hacked up this patch: https://gist.github.com/rui314/f3cd508b6cbbde763fc13bc6190d7505 Does this look good?

Feb 11 2019, 2:46 PM · Restricted Project
kzhuravl updated the diff for D58026: LLD: Preserve ABI version during linking ELF.

Make sure ABI versions across all object files are the same.

Feb 11 2019, 2:42 PM · Restricted Project
kzhuravl added a child revision for D57811: AMDGPU: Set ABI version to 1 for code object v3: D58026: LLD: Preserve ABI version during linking ELF.
Feb 11 2019, 12:24 AM · Restricted Project
kzhuravl added a parent revision for D58026: LLD: Preserve ABI version during linking ELF: D57811: AMDGPU: Set ABI version to 1 for code object v3.
Feb 11 2019, 12:24 AM · Restricted Project
kzhuravl created D58026: LLD: Preserve ABI version during linking ELF.
Feb 11 2019, 12:24 AM · Restricted Project
kzhuravl added a reviewer for D57811: AMDGPU: Set ABI version to 1 for code object v3: ruiu.
Feb 11 2019, 12:21 AM · Restricted Project
kzhuravl added a reviewer for D57810: MC/ELF: Allow targets to set ABI version: ruiu.
Feb 11 2019, 12:21 AM · Restricted Project

Feb 8 2019

kzhuravl added a comment to D57977: [HIP] compile option code-object-v3 propagate to llc.

https://github.com/llvm-mirror/clang/blob/master/lib/Driver/ToolChains/AMDGPU.cpp#L52

Feb 8 2019, 3:42 PM · Restricted Project
kzhuravl added a comment to D57977: [HIP] compile option code-object-v3 propagate to llc.

Can you handle all features as follows instead of checking a particular option?

Feb 8 2019, 3:37 PM · Restricted Project
kzhuravl accepted D57972: [AMDGPU] Split dot-insts feature.

LGTM

Feb 8 2019, 12:49 PM · Restricted Project
kzhuravl accepted D57971: [AMDGPU] Split dot-insts feature.

LGTM

Feb 8 2019, 12:43 PM · Restricted Project
kzhuravl added inline comments to D57971: [AMDGPU] Split dot-insts feature.
Feb 8 2019, 12:35 PM · Restricted Project
kzhuravl accepted D57971: [AMDGPU] Split dot-insts feature.

LGTM

Feb 8 2019, 12:33 PM · Restricted Project

Feb 6 2019

kzhuravl accepted D57397: AMDGPU/GlobalISel: Legalize addrspacecast.

LGTM

Feb 6 2019, 11:27 AM
kzhuravl accepted D57826: [AMDGPU][MC][CODEOBJECT] Added predefined symbols to access GPU minor and stepping numbers.

LGTM

Feb 6 2019, 9:10 AM · Restricted Project
kzhuravl added a comment to D57811: AMDGPU: Set ABI version to 1 for code object v3.

This review fell through the cracks and never got published.

Feb 6 2019, 1:22 AM · Restricted Project
kzhuravl added a parent revision for D57811: AMDGPU: Set ABI version to 1 for code object v3: D57810: MC/ELF: Allow targets to set ABI version.
Feb 6 2019, 1:21 AM · Restricted Project
kzhuravl added a child revision for D57810: MC/ELF: Allow targets to set ABI version: D57811: AMDGPU: Set ABI version to 1 for code object v3.
Feb 6 2019, 1:21 AM · Restricted Project
kzhuravl created D57811: AMDGPU: Set ABI version to 1 for code object v3.
Feb 6 2019, 1:20 AM · Restricted Project
kzhuravl created D57810: MC/ELF: Allow targets to set ABI version.
Feb 6 2019, 1:19 AM · Restricted Project

Feb 5 2019

kzhuravl accepted D57756: AMDGPU: Fix assert on trunc from bitcast of build_vector.

LGTM

Feb 5 2019, 9:04 AM

Jan 22 2019

kzhuravl committed rL351861: Add missing include (cstdlib) to Demangle.h.
Add missing include (cstdlib) to Demangle.h
Jan 22 2019, 11:18 AM
kzhuravl closed D57035: Add missing include (cstdlib) to Demangle.h.
Jan 22 2019, 11:18 AM
kzhuravl added a comment to D57035: Add missing include (cstdlib) to Demangle.h.

Will fix during committing. Thanks.

Jan 22 2019, 10:44 AM

Jan 21 2019

kzhuravl created D57035: Add missing include (cstdlib) to Demangle.h.
Jan 21 2019, 1:33 PM

Jan 9 2019

kzhuravl accepted D56525: [AMDGPU] Separate feature dot-insts.

LGTM

Jan 9 2019, 6:15 PM
kzhuravl accepted D56524: [AMDGPU] Separate feature dot-insts.

LGTM

Jan 9 2019, 6:14 PM

Jan 2 2019

kzhuravl accepted D56197: [AMDGPU] Change section name with metadata access.

LGTM

Jan 2 2019, 6:24 AM

Dec 11 2018

kzhuravl accepted D48179: [AMDGPU] Emit MessagePack HSA Metadata for v3 code object.

LGTM

Dec 11 2018, 9:27 PM
kzhuravl accepted D55267: [AMDGPU] Set metadata access for explicit section.

LGTM

Dec 11 2018, 5:23 AM

Nov 19 2018

kzhuravl added a comment to D54545: AMDGPU: Fix V_FMA_F16 selection on GFX9.

Is this the one that zeros the high bits vs. leaves them intact?

Nov 19 2018, 1:14 PM
kzhuravl committed rL347265: AMDGPU: Fix V_FMA_F16 selection on GFX9.
AMDGPU: Fix V_FMA_F16 selection on GFX9
Nov 19 2018, 1:13 PM
kzhuravl closed D54545: AMDGPU: Fix V_FMA_F16 selection on GFX9.
Nov 19 2018, 1:12 PM
kzhuravl accepted D54718: Restored selection of scalar_to_vector (v2x16).

LGTM

Nov 19 2018, 11:59 AM

Nov 16 2018

kzhuravl accepted D54301: [AMDGPU] Derive GCNSubtarget from MF to get overridden target features.

LGTM

Nov 16 2018, 10:06 AM

Nov 15 2018

kzhuravl committed rL347001: AMDHSA: More code object v3 fixes:.
AMDHSA: More code object v3 fixes:
Nov 15 2018, 3:16 PM