Page MenuHomePhabricator

nhaustov (Nikolay Haustov)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 19 2016, 1:59 AM (165 w, 5 d)

Recent Activity

Jul 21 2017

nhaustov committed rL308741: Test commit.
Test commit
Jul 21 2017, 6:59 AM

Dec 15 2016

nhaustov accepted D27619: AMDGPU: [AMDGPU] Assembler: add .hsa_code_object_metadata directive for functime metadata V2.0.
Dec 15 2016, 2:53 AM

Dec 13 2016

nhaustov added a comment to D27619: AMDGPU: [AMDGPU] Assembler: add .hsa_code_object_metadata directive for functime metadata V2.0.

Could you change directive name to .amdgpu_runtime_metadata? I think it better reflects the purpose as metadata has nothing to do with HSA.

Dec 13 2016, 5:44 AM

Dec 8 2016

nhaustov accepted D27225: [AMDGPU] Fix number of reserved SGPRs on CI to reflect flat scratch use.

Looks good as short term solution.

Dec 8 2016, 11:17 AM
nhaustov accepted D25046: AMDGPU: Emit runtime metadata version 2 as YAML.

Looks good.

Dec 8 2016, 3:30 AM

Dec 5 2016

nhaustov added inline comments to D25046: AMDGPU: Emit runtime metadata version 2 as YAML.
Dec 5 2016, 6:11 AM

Oct 28 2016

nhaustov added a comment to D26077: [AMDGPU] Expand vector mulhu/mulhs .

Can you also add test on mulhs?

Oct 28 2016, 8:42 AM

Sep 20 2016

nhaustov committed rL281962: AMDGPU: Improve documentation..
AMDGPU: Improve documentation.
Sep 20 2016, 2:13 AM
nhaustov closed D24724: AMDGPU: Improve documentation. by committing rL281962: AMDGPU: Improve documentation..
Sep 20 2016, 2:13 AM

Sep 19 2016

nhaustov updated the diff for D24724: AMDGPU: Improve documentation..

Add generated html file for easy review (will be removed in the commit).

Sep 19 2016, 5:43 AM
nhaustov retitled D24724: AMDGPU: Improve documentation. from to AMDGPU: Improve documentation..
Sep 19 2016, 5:41 AM

Sep 1 2016

nhaustov accepted D24145: AMDGPU: Add runtime metadata for pointee alignment of argument.
Sep 1 2016, 11:20 AM

Aug 31 2016

nhaustov added a comment to D24023: AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass.

LGTM. Can you also add a test which errors on an alias that is a constantexpr?

Aug 31 2016, 4:38 AM
nhaustov committed rL280221: AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass.
AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass
Aug 31 2016, 4:27 AM
nhaustov closed D24023: AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass by committing rL280221: AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass.
Aug 31 2016, 4:26 AM

Aug 30 2016

nhaustov updated subscribers of D24023: AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass.
Aug 30 2016, 3:15 AM
nhaustov retitled D24023: AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass from to AMDGPU/SI: Handle aliases in AMDGPUAlwaysInlinePass.
Aug 30 2016, 3:15 AM

Aug 2 2016

nhaustov accepted D23071: [OpenCL] Remove extra native_ functions from opencl-c.h.

Could you also note in commit message that online HTML docs are not very clear, but pdf of OpenCL specification has it right.

Aug 2 2016, 11:59 AM

Aug 1 2016

nhaustov accepted D22940: [OpenCL] Handle -cl-fp32-correctly-rounded-divide-sqrt.

Ok.

Aug 1 2016, 2:21 AM
nhaustov abandoned D22751: AMDGPU Device Libs pass..

Due to difficulties, it was decided to pursue different approach - create control functions in Runtime and link them as bitcode libraries.

Aug 1 2016, 2:11 AM

Jul 27 2016

nhaustov added inline comments to D22751: AMDGPU Device Libs pass..
Jul 27 2016, 6:55 AM
nhaustov updated the diff for D22751: AMDGPU Device Libs pass..
  • Update after Matt's review.
Jul 27 2016, 6:55 AM
nhaustov added inline comments to D22838: AMDGPU/SI: Implement amdgcn image intrinsics.
Jul 27 2016, 6:28 AM

Jul 26 2016

nhaustov added inline comments to D22751: AMDGPU Device Libs pass..
Jul 26 2016, 3:22 AM
nhaustov updated the diff for D22751: AMDGPU Device Libs pass..

Update after Tom's review.

Jul 26 2016, 3:21 AM

Jul 25 2016

nhaustov added a comment to D22751: AMDGPU Device Libs pass..

Should this really be in the backend? I also thought we were going to use a global constant variable instead of a function call

Yes, it needs to be. Global variable approach doesn't support functions in different modes.
Any specific concerns with the patch?

The idea was that would be controlled by multiple globals with private visibility in different modules linked together.
Why does it need to be in the backend? This seems more like a pass that belongs with the runtime/compiler library when linking the AMD specific library

Jul 25 2016, 10:58 AM
nhaustov added a comment to D22751: AMDGPU Device Libs pass..

Should this really be in the backend? I also thought we were going to use a global constant variable instead of a function call

Jul 25 2016, 10:53 AM
nhaustov added a reviewer for D22751: AMDGPU Device Libs pass.: yaxunl.
Jul 25 2016, 8:08 AM
nhaustov retitled D22751: AMDGPU Device Libs pass. from to AMDGPU Device Libs pass..
Jul 25 2016, 5:45 AM

Jul 20 2016

nhaustov accepted D22526: AMDGPU: Fix bug causing crash due to invalid opencl version metadata..

Ok.

Jul 20 2016, 2:59 AM

Jul 15 2016

nhaustov accepted D21849: [AMDGPU] Add metadata for runtime.
Jul 15 2016, 3:27 AM

Jul 14 2016

nhaustov requested changes to D21849: [AMDGPU] Add metadata for runtime.
Jul 14 2016, 5:58 AM

Jul 1 2016

nhaustov committed rL274341: Resubmit r268719 - AMDGPU/SI: Add amdgpu_kernel calling convention. Part 2..
Resubmit r268719 - AMDGPU/SI: Add amdgpu_kernel calling convention. Part 2.
Jul 1 2016, 3:08 AM
nhaustov closed D19917: AMDGPU/SI: Add amdgpu_kernel calling convention. by committing rL274341: Resubmit r268719 - AMDGPU/SI: Add amdgpu_kernel calling convention. Part 2..
Jul 1 2016, 3:08 AM

Jun 30 2016

nhaustov committed rL274220: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..
AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels.
Jun 30 2016, 2:13 AM
nhaustov closed D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels. by committing rL274220: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..
Jun 30 2016, 2:13 AM

Jun 29 2016

nhaustov updated D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..
Jun 29 2016, 3:18 AM
nhaustov updated the diff for D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..

Update after discussion with Sam and review from Anastasia.

Jun 29 2016, 3:17 AM

Jun 28 2016

nhaustov committed rL273997: [Driver] Add method to redirect output of Compilation..
[Driver] Add method to redirect output of Compilation.
Jun 28 2016, 1:07 AM
nhaustov closed D21224: [Driver] Add method to redirect output of Compilation. by committing rL273997: [Driver] Add method to redirect output of Compilation..
Jun 28 2016, 1:07 AM

Jun 27 2016

nhaustov added a comment to D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..

I am wondering what's the difference between this calling convention and spir_kernel.

Jun 27 2016, 11:08 AM
nhaustov added a comment to D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..

I am wondering what's the difference between this calling convention and spir_kernel.

Jun 27 2016, 9:20 AM

Jun 23 2016

nhaustov accepted D21647: AMDGPU/SI: Make sure not to fold offsets into local address space globals.

Looks good.

Jun 23 2016, 9:26 AM

Jun 22 2016

nhaustov added reviewers for D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels.: yaxunl, Anastasia.
Jun 22 2016, 8:52 AM

Jun 20 2016

nhaustov added a comment to D21224: [Driver] Add method to redirect output of Compilation..

Ping.

Jun 20 2016, 4:27 AM
nhaustov updated the diff for D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..

Add test for calling OpenCL kernel from kernel.

Jun 20 2016, 4:26 AM
nhaustov added a comment to D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..

Does this new patch fix the OpenCL regression?

Jun 20 2016, 4:25 AM

Jun 15 2016

nhaustov retitled D21367: AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels. from to AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels..
Jun 15 2016, 3:42 AM

Jun 10 2016

nhaustov updated the diff for D21224: [Driver] Add method to redirect output of Compilation..

Fix summary.

Jun 10 2016, 4:22 AM
nhaustov retitled D21224: [Driver] Add method to redirect output of Compilation. from to [Driver] Add method to redirect output of Compilation..
Jun 10 2016, 4:20 AM

May 23 2016

nhaustov accepted D20166: [AMDGPU] Assembler: refactor parsing of modifiers and immediates. Allow modifiers for imms..

Ok.

May 23 2016, 1:59 AM · Restricted Project

May 10 2016

nhaustov updated subscribers of D17578: [OpenCL]Allowing explicit conversion of "0" to event_t type.
May 10 2016, 7:41 AM

May 6 2016

nhaustov committed rL268740: Revert "AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels.".
Revert "AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels."
May 6 2016, 8:07 AM
nhaustov committed rL268739: Revert "AMDGPU/SI: Add amdgpu_kernel calling convention. Part 2.".
Revert "AMDGPU/SI: Add amdgpu_kernel calling convention. Part 2."
May 6 2016, 8:05 AM
nhaustov committed rL268719: AMDGPU/SI: Add amdgpu_kernel calling convention. Part 2..
AMDGPU/SI: Add amdgpu_kernel calling convention. Part 2.
May 6 2016, 2:29 AM
nhaustov committed rL268718: AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels..
AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels.
May 6 2016, 2:21 AM
nhaustov closed D19918: AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels. by committing rL268718: AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels..
May 6 2016, 2:21 AM
nhaustov committed rL268717: AMDGPU/SI: Add amdgpu_kernel calling convention. Part 1..
AMDGPU/SI: Add amdgpu_kernel calling convention. Part 1.
May 6 2016, 2:13 AM

May 5 2016

nhaustov added a comment to D19917: AMDGPU/SI: Add amdgpu_kernel calling convention..

I'll also probably split this change into two parts and submit AMDGPUMachineFunction.* changes after corresponding Clang change. This will ensure frontends work.

May 5 2016, 4:55 AM
nhaustov updated the diff for D19917: AMDGPU/SI: Add amdgpu_kernel calling convention..

Also recognize SPIR_KERNEL as kernel for now to make sure other frontent (HCC) works.
Add bitcode compatibility tests for amdgpu and other missing calling conventions.

May 5 2016, 4:54 AM

May 4 2016

nhaustov added a parent revision for D19918: AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels.: D19917: AMDGPU/SI: Add amdgpu_kernel calling convention..
May 4 2016, 6:54 AM
nhaustov added a child revision for D19917: AMDGPU/SI: Add amdgpu_kernel calling convention.: D19918: AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels..
May 4 2016, 6:54 AM
nhaustov retitled D19918: AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels. from to AMDGPU/SI: Use amdgpu_kernel calling convention for OpenCL kernels..
May 4 2016, 6:53 AM
nhaustov retitled D19917: AMDGPU/SI: Add amdgpu_kernel calling convention. from to AMDGPU/SI: Add amdgpu_kernel calling convention..
May 4 2016, 6:51 AM

Apr 29 2016

nhaustov updated subscribers of D18369: [OpenCL] Upstreaming khronos OpenCL header file..
Apr 29 2016, 5:13 AM
nhaustov committed rL268015: AMDGPU/SI: Assembler: Unify parsing/printing of operands..
AMDGPU/SI: Assembler: Unify parsing/printing of operands.
Apr 29 2016, 2:08 AM
nhaustov closed D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands. by committing rL268015: AMDGPU/SI: Assembler: Unify parsing/printing of operands..
Apr 29 2016, 2:08 AM
nhaustov updated the diff for D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands..

Rebase on latest repository (with reverted TTMP quads changes).

Apr 29 2016, 2:05 AM

Apr 28 2016

nhaustov accepted D19602: [AMDGPU][llvm-mc] Add some missing testcases to trap.s.

Ok.

Apr 28 2016, 9:08 AM · Restricted Project

Apr 27 2016

nhaustov updated the diff for D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands..

Merge with latest changes.

Apr 27 2016, 9:50 AM
nhaustov updated the diff for D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands..

Use getLexer.is().
Add comment for omod.

Apr 27 2016, 6:55 AM
nhaustov added a comment to D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands..
Apr 27 2016, 6:37 AM
nhaustov added a comment to D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands..

It's hard to split this change because operand changes depend on parsing changes and vice versa.

Apr 27 2016, 6:11 AM
nhaustov updated the diff for D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands..

Move sdwa_sel up and remove NamedBitMatchClass.
Move removal of prefix to ParseInstruction.

Apr 27 2016, 6:09 AM
nhaustov retitled D19584: AMDGPU/SI: Assembler: Unify parsing/printing of operands. from to AMDGPU/SI: Assembler: Unify parsing/printing of operands..
Apr 27 2016, 3:34 AM

Apr 26 2016

nhaustov accepted D19342: [AMDGPU][llvm-mc] Add support of TTMP quads. Rework M0 exclusion for SMRD..

LGTM

Apr 26 2016, 7:07 AM · Restricted Project

Apr 25 2016

nhaustov accepted D19335: [AMDGPU][llvm-mc] s_getreg/setreg* - Support symbolic names of hardware registers..

Ok.

Apr 25 2016, 6:44 AM · Restricted Project
nhaustov accepted D19329: [AMDGPU][llvm-mc] s_getreg/setreg* - Add hwreg(...) syntax..

Ok.

Apr 25 2016, 6:44 AM · Restricted Project
nhaustov accepted D19360: [AMDGPU] Assembler: basic support for SDWA instructions.

LGTM

Apr 25 2016, 6:29 AM · Restricted Project

Apr 22 2016

nhaustov committed rL267144: AMDGPU/SI: Add test missed in rL266865.
AMDGPU/SI: Add test missed in rL266865
Apr 22 2016, 4:45 AM

Apr 21 2016

nhaustov added a comment to D19342: [AMDGPU][llvm-mc] Add support of TTMP quads. Rework M0 exclusion for SMRD..

Can you also add 1-2 tests to test/MC/AMDGPU/reg-syntax-extra.s?

Apr 21 2016, 6:32 AM · Restricted Project
nhaustov added inline comments to D19360: [AMDGPU] Assembler: basic support for SDWA instructions.
Apr 21 2016, 6:15 AM · Restricted Project
nhaustov accepted D19317: [AMDGPU] Assembler: prevent parseDPPCtrlOps from eating invalid tokens.

LGTM

Apr 21 2016, 6:08 AM

Apr 20 2016

nhaustov committed rL266865: AMDGPU/SI: Assembler: improvements to support trap handlers..
AMDGPU/SI: Assembler: improvements to support trap handlers.
Apr 20 2016, 2:40 AM
nhaustov closed D19163: AMDGPU/SI: Assembler: improvements to support trap handlers. by committing rL266865: AMDGPU/SI: Assembler: improvements to support trap handlers..
Apr 20 2016, 2:40 AM

Apr 18 2016

nhaustov accepted D19116: [AMDGPU] Inline asm: use AsmNames for register constraints..

Looks good.

Apr 18 2016, 2:01 AM · Restricted Project

Apr 15 2016

nhaustov retitled D19163: AMDGPU/SI: Assembler: improvements to support trap handlers. from to AMDGPU/SI: Assembler: improvements to support trap handlers..
Apr 15 2016, 9:31 AM

Apr 5 2016

nhaustov accepted D18552: [AMDGPU] AsmParser: disable DPP for unsupported instructions. New dpp tests. Fix v_nop_dpp..

LGTM.

Apr 5 2016, 9:35 AM · Restricted Project

Mar 18 2016

nhaustov accepted D18267: [AMDGPU] Assembler: Change dpp_ctrl syntax to match sp3.
Mar 18 2016, 8:35 AM · Restricted Project

Mar 15 2016

nhaustov committed rL263540: [AMDGPU] Assembler: Update SOP* tests.
[AMDGPU] Assembler: Update SOP* tests
Mar 15 2016, 12:50 AM
nhaustov closed D18084: [AMDGPU] Assembler: Update SOP* tests by committing rL263540: [AMDGPU] Assembler: Update SOP* tests.
Mar 15 2016, 12:49 AM

Mar 14 2016

nhaustov added a reviewer for D18084: [AMDGPU] Assembler: Update SOP* tests: vpykhtin.
Mar 14 2016, 4:27 AM
nhaustov committed rL263420: [AMDGPU] Assembler: SOP* instruction fixes.
[AMDGPU] Assembler: SOP* instruction fixes
Mar 14 2016, 4:22 AM
nhaustov closed D18040: [AMDGPU] Assembler: SOP* instruction fixes by committing rL263420: [AMDGPU] Assembler: SOP* instruction fixes.
Mar 14 2016, 4:22 AM

Mar 11 2016

nhaustov retitled D18084: [AMDGPU] Assembler: Update SOP* tests from to [AMDGPU] Assembler: Update SOP* tests.
Mar 11 2016, 6:58 AM
nhaustov committed rL263212: [AMDGPU] Assembler: change v_madmk operands to have same order as mad..
[AMDGPU] Assembler: change v_madmk operands to have same order as mad.
Mar 11 2016, 1:32 AM
nhaustov closed D17984: [AMDGPU] Assembler: change v_madmk operands to have same order as mad. by committing rL263212: [AMDGPU] Assembler: change v_madmk operands to have same order as mad..
Mar 11 2016, 1:32 AM

Mar 10 2016

nhaustov retitled D18040: [AMDGPU] Assembler: SOP* instruction fixes from to [AMDGPU] Assembler: SOP* instruction fixes.
Mar 10 2016, 6:15 AM

Mar 9 2016

nhaustov added inline comments to D16998: [AMDGPU] llvm-objdump: disassembling amdgcn object file.
Mar 9 2016, 5:36 AM · Restricted Project
nhaustov committed rL263006: [AMDGPU] Assembler: Support abs() syntax..
[AMDGPU] Assembler: Support abs() syntax.
Mar 9 2016, 3:08 AM