rampitec (Stanislav Mekhanoshin)
User

Projects

User does not belong to any projects.

User Details

User Since
Apr 4 2014, 4:14 AM (189 w, 22 h)

Recent Activity

Yesterday

rampitec added inline comments to D40113: AMDGPU: Fix crash when scheduling DBG_VALUE.
Fri, Nov 17, 4:52 PM
rampitec accepted D40158: AMDGPU: Use gfx9 carry-less add/sub instructions.
Fri, Nov 17, 2:25 PM
rampitec added a comment to D40158: AMDGPU: Use gfx9 carry-less add/sub instructions.

OK, thanks.

Fri, Nov 17, 2:24 PM
rampitec requested changes to D40158: AMDGPU: Use gfx9 carry-less add/sub instructions.
Fri, Nov 17, 1:02 PM
rampitec accepted D40158: AMDGPU: Use gfx9 carry-less add/sub instructions.

LGTM with the assertion added to moveScalarAddSub.

Fri, Nov 17, 12:21 PM
rampitec added inline comments to D39897: AMDGPU: Partial ILP scheduler port from SelectionDAG to SchedulingDAG (experimental).
Fri, Nov 17, 9:42 AM
rampitec accepted D39897: AMDGPU: Partial ILP scheduler port from SelectionDAG to SchedulingDAG (experimental).

LGTM

Fri, Nov 17, 8:50 AM
rampitec accepted D40163: AMDGPU: Move hazard avoidance out of waitcnt pass..

LGTM

Fri, Nov 17, 8:43 AM
rampitec accepted D40172: [AMDGPU] SDWA: remove omod src operand for VOP2b instructions.

Thank you!

Fri, Nov 17, 8:10 AM

Thu, Nov 16

rampitec added inline comments to D40155: AMDGPU: Fix breaking SMEM clauses.
Thu, Nov 16, 7:21 PM
rampitec added inline comments to D40158: AMDGPU: Use gfx9 carry-less add/sub instructions.
Thu, Nov 16, 4:51 PM
rampitec accepted D40155: AMDGPU: Fix breaking SMEM clauses.

LGTM

Thu, Nov 16, 4:02 PM
rampitec added inline comments to D40113: AMDGPU: Fix crash when scheduling DBG_VALUE.
Thu, Nov 16, 3:54 PM
rampitec accepted D40153: AMDGPU: Replace list of SMEM buffer opcodes.

LGTM

Thu, Nov 16, 3:43 PM
rampitec added inline comments to D40088: [AMDGPU][MC][GFX8][GFX9] Corrected names of integer v_{add/addc/sub/subrev/subb/subbrev}.
Thu, Nov 16, 3:41 PM
rampitec accepted D35267: Pass Divergence Analysis data to selection DAG to drive divergence dependent instruction selection.

LGTM

Thu, Nov 16, 1:29 PM

Wed, Nov 15

rampitec accepted D40095: Fix pointer EVT in SelectionDAGBuilder::visitAlloca.

LGTM

Wed, Nov 15, 6:56 PM
rampitec added inline comments to D40113: AMDGPU: Fix crash when scheduling DBG_VALUE.
Wed, Nov 15, 4:56 PM
rampitec added inline comments to D40095: Fix pointer EVT in SelectionDAGBuilder::visitAlloca.
Wed, Nov 15, 12:54 PM
rampitec accepted D39740: CodeGen: Fix pointer info and index type when splitting vector.

LGTM

Wed, Nov 15, 11:50 AM
rampitec accepted D40085: Fix APInt bit size in processDbgDeclares.

LGTM

Wed, Nov 15, 11:47 AM

Tue, Nov 14

rampitec accepted D40059: AMDGPU: Select DS insts without m0 initialization.

LGTM

Tue, Nov 14, 8:17 PM
rampitec accepted D39731: AMDGPU: Don't use MUBUF vaddr if address may overflow.

LGTM

Tue, Nov 14, 3:03 PM
rampitec accepted D39685: AMDGPU: Handle or in multi-use shl ptr combi.

LGTM

Tue, Nov 14, 2:59 PM
rampitec accepted D40040: [AMDGPU] Fix SITargetLowering::LowerCall for pointer info of byval argument.

LGTM

Tue, Nov 14, 10:33 AM
rampitec added inline comments to D39998: AMDGPU: Directly implement computeKnownBits for workitem intrinsics .
Tue, Nov 14, 9:52 AM
rampitec accepted D40000: AMDGPU: Error on stack size overflow.

LGTM

Tue, Nov 14, 9:52 AM
rampitec added inline comments to D39685: AMDGPU: Handle or in multi-use shl ptr combi.
Tue, Nov 14, 9:37 AM

Mon, Nov 13

rampitec accepted D39983: AMDGPU: Add separate definitions for DS insts without m0 use.

LGTM

Mon, Nov 13, 3:04 PM
rampitec accepted D39973: Let llvm.invariant.group.barrier accepts pointer to any address space.

LGTM

Mon, Nov 13, 2:44 PM
rampitec added inline comments to D39973: Let llvm.invariant.group.barrier accepts pointer to any address space.
Mon, Nov 13, 1:58 PM
rampitec added inline comments to D39973: Let llvm.invariant.group.barrier accepts pointer to any address space.
Mon, Nov 13, 1:53 PM
rampitec added inline comments to D39973: Let llvm.invariant.group.barrier accepts pointer to any address space.
Mon, Nov 13, 1:00 PM
rampitec accepted D39970: AMDGPU: Fix producing saveexec when the copy is spilled.

LGTM

Mon, Nov 13, 12:12 PM
rampitec accepted D39945: AMDGPU: Fix not converting d16 load/stores to offset.

LGTM

Mon, Nov 13, 9:23 AM
rampitec accepted D39951: AMDGPU: Implement computeKnownBitsForTargetNode for mbcnt.

LGTM with spacing fixed.

Mon, Nov 13, 9:19 AM
rampitec added inline comments to D39897: AMDGPU: Partial ILP scheduler port from SelectionDAG to SchedulingDAG (experimental).
Mon, Nov 13, 9:17 AM

Fri, Nov 10

rampitec added inline comments to D39897: AMDGPU: Partial ILP scheduler port from SelectionDAG to SchedulingDAG (experimental).
Fri, Nov 10, 12:30 PM
rampitec added a comment to D39897: AMDGPU: Partial ILP scheduler port from SelectionDAG to SchedulingDAG (experimental).

Can you add some tests just to show it does not crash? Maybe add run-lines to schedule-regpressure-limit.ll, schedule-regpressure-limit2.ll

Fri, Nov 10, 12:24 PM

Thu, Nov 9

rampitec added a comment to D35267: Pass Divergence Analysis data to selection DAG to drive divergence dependent instruction selection.

This actually looks clean to me, thank you!

Thu, Nov 9, 12:06 PM

Tue, Nov 7

rampitec added inline comments to D39758: CodeGen: Fix pointer info in SplitVecOp_EXTRACT_VECTOR_ELT.
Tue, Nov 7, 6:07 PM
rampitec added inline comments to D39758: CodeGen: Fix pointer info in SplitVecOp_EXTRACT_VECTOR_ELT.
Tue, Nov 7, 6:01 PM
rampitec added inline comments to D39758: CodeGen: Fix pointer info in SplitVecOp_EXTRACT_VECTOR_ELT.
Tue, Nov 7, 5:48 PM
rampitec accepted D39758: CodeGen: Fix pointer info in SplitVecOp_EXTRACT_VECTOR_ELT.

LGTM

Tue, Nov 7, 2:01 PM
rampitec added inline comments to D39731: AMDGPU: Don't use MUBUF vaddr if address may overflow.
Tue, Nov 7, 10:00 AM

Mon, Nov 6

rampitec accepted D39674: AMDGPU: Remove redundant combine.

LGTM

Mon, Nov 6, 10:52 AM
rampitec accepted D39677: AMDGPU: Fix multi-use shl/add combine.

LGTM

Mon, Nov 6, 10:33 AM
rampitec added inline comments to D39685: AMDGPU: Handle or in multi-use shl ptr combi.
Mon, Nov 6, 10:29 AM
rampitec accepted D39686: AMDGPU: Preserve nuw in shl add ptr combine.

LGTM

Mon, Nov 6, 10:23 AM

Fri, Nov 3

rampitec accepted D39616: [AMDGPU] Remove hardcoded address space value from AMDGPULibFunc.

LGTM

Fri, Nov 3, 9:02 PM
rampitec added inline comments to D39616: [AMDGPU] Remove hardcoded address space value from AMDGPULibFunc.
Fri, Nov 3, 2:10 PM

Mon, Oct 30

rampitec accepted D39413: AMDGPU: Select v_mad_u64_u32 and v_mad_i64_i32.

LGTM

Mon, Oct 30, 8:02 PM
rampitec accepted D39432: InferAddressSpaces: Fix bug about replacing addrspacecast.

LGTM

Mon, Oct 30, 2:01 PM
rampitec added inline comments to D39413: AMDGPU: Select v_mad_u64_u32 and v_mad_i64_i32.
Mon, Oct 30, 9:49 AM

Wed, Oct 25

rampitec accepted D39306: Fix CodeGen/AMDGPU/fcanonicalize-elimination.ll on FreeBSD 11.0.

LGTM

Wed, Oct 25, 2:41 PM

Mon, Oct 23

rampitec accepted D39205: AMDGPU: Initialize WavefrontSize from TD files.

LGTM

Mon, Oct 23, 2:08 PM
rampitec accepted D39187: AMDGPU: Cleanup local atomic node names.

LGTM

Mon, Oct 23, 9:57 AM

Fri, Oct 20

rampitec added inline comments to D39125: [inlineasm] Fix crash when number of matched input constraint operands overflows signed char.
Fri, Oct 20, 9:39 AM · Restricted Project

Oct 16 2017

rampitec added inline comments to D38634: AMDGPU : Custom lowering constrained fps..
Oct 16 2017, 2:35 PM

Oct 10 2017

rampitec accepted D38754: Prevent Machine Copy Propagation from replacing live copy with the dead one.

LGTM, but please also wait for @MatzeB.
In general I believe that "Rename Disconnected Subregister Components" should either eliminate what it has marked dead or have a pass running after it to do so, as MachineDCE will not handle it.

Oct 10 2017, 12:07 PM
rampitec accepted D38710: AMDGPU: Fix failure to select branch with optnone.

LGTM

Oct 10 2017, 10:41 AM
rampitec added inline comments to D38710: AMDGPU: Fix failure to select branch with optnone.
Oct 10 2017, 10:28 AM
rampitec accepted D38709: AMDGPU: Fix incorrect selection of pseudo-branches.

LGTM

Oct 10 2017, 10:25 AM

Oct 9 2017

rampitec added inline comments to D38634: AMDGPU : Custom lowering constrained fps..
Oct 9 2017, 3:19 PM

Oct 6 2017

rampitec accepted D38610: [AMDGPU] Lower enqueued blocks and generate runtime metadata.

LGTM

Oct 6 2017, 12:19 PM
rampitec added inline comments to D38610: [AMDGPU] Lower enqueued blocks and generate runtime metadata.
Oct 6 2017, 10:55 AM
rampitec committed rL315081: [AMDGPU] New 64 bit div/rem expansion.
[AMDGPU] New 64 bit div/rem expansion
Oct 6 2017, 10:26 AM
rampitec closed D38607: [AMDGPU] New 64 bit div/rem expansion by committing rL315081: [AMDGPU] New 64 bit div/rem expansion.
Oct 6 2017, 10:26 AM
rampitec updated the diff for D38607: [AMDGPU] New 64 bit div/rem expansion.

Added comment.

Oct 6 2017, 10:13 AM
rampitec updated the diff for D38607: [AMDGPU] New 64 bit div/rem expansion.

Added message to assert.

Oct 6 2017, 10:08 AM
rampitec added a comment to D38607: [AMDGPU] New 64 bit div/rem expansion.

Is this a port of what SC emits? I thought it emitted control flow for this.

Oct 6 2017, 10:02 AM

Oct 5 2017

rampitec added inline comments to D38607: [AMDGPU] New 64 bit div/rem expansion.
Oct 5 2017, 7:05 PM
rampitec created D38607: [AMDGPU] New 64 bit div/rem expansion.
Oct 5 2017, 4:23 PM

Oct 4 2017

rampitec accepted D38562: AMDGPU: Fix not accounting for instruction size in bundles.

LGTM

Oct 4 2017, 3:22 PM

Oct 3 2017

rampitec accepted D38519: AMDGPU: Implement hasBitPreservingFPLogic.

LGTM

Oct 3 2017, 5:11 PM
rampitec accepted D38518: AMDGPU: Look for src mods before fp_extend.

LGTM

Oct 3 2017, 5:07 PM
rampitec accepted D38510: AMDGPU: Implement isFPExtFree.

LGTM

Oct 3 2017, 12:43 PM
rampitec accepted D38293: Avoid predicated execution of the basic blocks containing scalar instructions.

LGTM

Oct 3 2017, 10:20 AM

Oct 2 2017

rampitec added a comment to D38293: Avoid predicated execution of the basic blocks containing scalar instructions.

I would just bail on any of these instructions not trying to optimize the case, just like you did in the beginning.

Oct 2 2017, 2:16 PM
rampitec added inline comments to D38293: Avoid predicated execution of the basic blocks containing scalar instructions.
Oct 2 2017, 2:08 PM
rampitec accepted D38467: AMDGPU: Remove global isGCN predicates.

LGTM

Oct 2 2017, 12:43 PM
rampitec added inline comments to D38467: AMDGPU: Remove global isGCN predicates.
Oct 2 2017, 10:31 AM
rampitec added inline comments to D38293: Avoid predicated execution of the basic blocks containing scalar instructions.
Oct 2 2017, 10:22 AM
rampitec committed rL314688: Eliminate ftrunc if source is know to be rounded.
Eliminate ftrunc if source is know to be rounded
Oct 2 2017, 9:59 AM
rampitec closed D38421: Eliminate ftrunc if source is know to be rounded by committing rL314688: Eliminate ftrunc if source is know to be rounded.
Oct 2 2017, 9:58 AM

Sep 29 2017

rampitec committed rL314568: [AMDGPU] Set fast-math flags on functions given the options.
[AMDGPU] Set fast-math flags on functions given the options
Sep 29 2017, 4:42 PM
rampitec closed D38325: [AMDGPU] Set fast-math flags on functions given the options by committing rL314568: [AMDGPU] Set fast-math flags on functions given the options.
Sep 29 2017, 4:41 PM
rampitec updated the diff for D38325: [AMDGPU] Set fast-math flags on functions given the options.

Created reference to Options to capture as requested.

Sep 29 2017, 3:47 PM
rampitec added inline comments to D38325: [AMDGPU] Set fast-math flags on functions given the options.
Sep 29 2017, 3:31 PM
rampitec added inline comments to D38325: [AMDGPU] Set fast-math flags on functions given the options.
Sep 29 2017, 3:28 PM
rampitec updated the diff for D38421: Eliminate ftrunc if source is know to be rounded.

Added comment as requested.

Sep 29 2017, 2:24 PM
rampitec added a comment to D38421: Eliminate ftrunc if source is know to be rounded.

We could potentially update visitCEIL and visitFLOOR as well, and use the same opcode test in each, although I don't think such combinations are very likely.

Sep 29 2017, 1:36 PM
rampitec added inline comments to D38421: Eliminate ftrunc if source is know to be rounded.
Sep 29 2017, 1:27 PM
rampitec created D38421: Eliminate ftrunc if source is know to be rounded.
Sep 29 2017, 12:48 PM
rampitec accepted D35361: CodeGen: Fix pointer info in expandUnalignedLoad/Store.

LGTM

Sep 29 2017, 12:17 PM
rampitec added inline comments to D35361: CodeGen: Fix pointer info in expandUnalignedLoad/Store.
Sep 29 2017, 11:44 AM

Sep 28 2017

rampitec added a reviewer for D38325: [AMDGPU] Set fast-math flags on functions given the options: dfukalov.
Sep 28 2017, 11:54 AM

Sep 27 2017

rampitec created D38325: [AMDGPU] Set fast-math flags on functions given the options.
Sep 27 2017, 11:37 AM

Sep 26 2017

rampitec added a comment to D38293: Avoid predicated execution of the basic blocks containing scalar instructions.

Test is needed.

Sep 26 2017, 9:00 PM

Sep 22 2017

rampitec accepted D38166: AMDGPU: Select d16 loads into low component of register.

LGTM

Sep 22 2017, 7:58 AM