ashutosh.nema (Ashutosh Nema)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 14 2015, 12:49 AM (199 w, 6 d)

Recent Activity

Feb 15 2018

ashutosh.nema added a comment to D43323: [NFC] Consolidate six getPointerOperand() utility functions into one place.

Mostly looks OK, just few minor comment.

Feb 15 2018, 9:47 PM

Nov 27 2017

ashutosh.nema added a comment to D38722: Added Remarks for Loop Versioning LICM Pass.

This patch looks OK to me, except the floating point value support in NV (maybe it can be done as a later enhancement once NV supports floating point values)

Nov 27 2017, 10:17 PM

Nov 5 2017

ashutosh.nema added a comment to D38722: Added Remarks for Loop Versioning LICM Pass.

Thanks for working on this !

Nov 5 2017, 10:56 PM

Sep 7 2017

ashutosh.nema added a comment to D30247: Epilog loop vectorization.

This patch is little old and not updated with latest vectorization changes.

Sep 7 2017, 2:38 AM

Sep 3 2017

ashutosh.nema added a comment to D36878: Inst Combine GEP Flatten.

Sorry for the late reply, I was on vacation.

Sep 3 2017, 9:52 PM

Aug 18 2017

ashutosh.nema updated subscribers of D36878: Inst Combine GEP Flatten.
Aug 18 2017, 9:26 AM
ashutosh.nema edited reviewers for D36878: Inst Combine GEP Flatten, added: sanjoy, nlopes, dberlin; removed: llvm-commits.
Aug 18 2017, 9:25 AM
ashutosh.nema created D36878: Inst Combine GEP Flatten.
Aug 18 2017, 9:23 AM

Jul 27 2017

ashutosh.nema added a comment to D35888: Changed basic cost of Store operation on X86..

Is store is always 2 UOps on x86, independent of the type and register(xmm/ymm) ?

Jul 27 2017, 3:57 AM

Jul 17 2017

ashutosh.nema added a comment to D35411: [SimplifyCFG] Defer folding unconditional branches to LateSimplifyCFG if it can destroy canonical loop structure..

Thanks Balaram for posting this patch, in general idea looks good to preserve the canonical form of the loops.

Jul 17 2017, 2:32 AM

Mar 6 2017

ashutosh.nema added a comment to D30247: Epilog loop vectorization.

Block layout description

Mar 6 2017, 1:05 AM
ashutosh.nema updated the diff for D30247: Epilog loop vectorization.

This change includes:

Mar 6 2017, 12:58 AM
ashutosh.nema added inline comments to D30247: Epilog loop vectorization.
Mar 6 2017, 12:52 AM

Feb 22 2017

ashutosh.nema added a comment to D30247: Epilog loop vectorization.

Test cases are missing well add.

Feb 22 2017, 2:01 AM
ashutosh.nema added a comment to D30247: Epilog loop vectorization.

Block layout:

Feb 22 2017, 1:51 AM
ashutosh.nema added reviewers for D30247: Epilog loop vectorization: Ayal, mkuper, jmolloy, rengolin.
Feb 22 2017, 1:49 AM
ashutosh.nema created D30247: Epilog loop vectorization.
Feb 22 2017, 1:46 AM

Dec 19 2016

ashutosh.nema accepted D27469: [LoopVersioning] Require loop-simplify form for loop versioning..

LGTM, thanks!

Dec 19 2016, 7:20 AM
ashutosh.nema added a comment to D27469: [LoopVersioning] Require loop-simplify form for loop versioning..

It’s good to avoid loop which are not in loop-simplify form.

Dec 19 2016, 3:31 AM

Oct 18 2016

ashutosh.nema added inline comments to D25557: [LAA] Collect pointers with unknown bounds.
Oct 18 2016, 4:25 AM

Oct 17 2016

ashutosh.nema added inline comments to D25557: [LAA] Collect pointers with unknown bounds.
Oct 17 2016, 12:19 AM

Oct 14 2016

ashutosh.nema accepted D25464: [NFC] Loop Versioning for LICM code clean up.

LGTM.

Oct 14 2016, 1:20 AM

Oct 12 2016

ashutosh.nema added a comment to D25464: [NFC] Loop Versioning for LICM code clean up.

Changes looks OK, except one minor formatting comment.

Oct 12 2016, 10:08 PM

Oct 4 2016

ashutosh.nema added inline comments to D24934: [LICM] Add support of a new optimization case to Loop Versioning for LICM + code clean up.
Oct 4 2016, 1:14 AM

Sep 27 2016

ashutosh.nema added a comment to D24934: [LICM] Add support of a new optimization case to Loop Versioning for LICM + code clean up.

Thanks Evgeny for working on this.

Sep 27 2016, 4:33 AM

Jun 27 2016

ashutosh.nema updated subscribers of D21251: [TTI] The cost model should not assume illegal vector casts get completely scalarized.
Jun 27 2016, 3:35 AM

Jun 21 2016

ashutosh.nema added a reviewer for D21363: Strided Memory Access Vectorization: Ayal.
Jun 21 2016, 4:33 AM
ashutosh.nema added a comment to D21363: Strided Memory Access Vectorization.

Thanks Ayal for looking into this RFC.

Jun 21 2016, 4:17 AM

Jun 20 2016

ashutosh.nema added a comment to D21363: Strided Memory Access Vectorization.

Thanks Elena for looking into this RFC.

Jun 20 2016, 4:50 AM

Jun 14 2016

ashutosh.nema updated subscribers of D21363: Strided Memory Access Vectorization.

Adding Shahid as co Author.

Jun 14 2016, 11:41 PM
ashutosh.nema retitled D21363: Strided Memory Access Vectorization from to Strided Memory Access Vectorization.
Jun 14 2016, 11:13 PM

Jun 13 2016

ashutosh.nema accepted D21281: Fix a typo in loop versioning..

LGTM.

Jun 13 2016, 3:46 AM

May 18 2016

ashutosh.nema updated subscribers of D19795: Add new flag and intrinsic support for MWAITX and MONITORX instructions..
May 18 2016, 3:56 AM
ashutosh.nema updated subscribers of D19796: Add new intrinsic support for MONITORX and MWAITX instructions..
May 18 2016, 3:56 AM

May 2 2016

ashutosh.nema added a comment to D19809: Move createReversePostOrderFunctionAttrsPass right after the inliner is done.

Thanks for pointing Mehdi, this change looks OK.

May 2 2016, 9:35 PM

Apr 25 2016

ashutosh.nema closed D19310: X86 TRUNCATE (v16i32 to v16i8) cost change in SSE4.1 mode.

Thanks Simon for review, this change landed at revision 267123.

Apr 25 2016, 12:15 AM

Apr 20 2016

ashutosh.nema updated the diff for D19310: X86 TRUNCATE (v16i32 to v16i8) cost change in SSE4.1 mode.

Incorporated comments from Simon.

Apr 20 2016, 8:00 AM

Apr 19 2016

ashutosh.nema added a reviewer for D19310: X86 TRUNCATE (v16i32 to v16i8) cost change in SSE4.1 mode: delena.
Apr 19 2016, 11:40 PM
ashutosh.nema retitled D19310: X86 TRUNCATE (v16i32 to v16i8) cost change in SSE4.1 mode from to X86 TRUNCATE (v16i32 to v16i8) cost change in SSE4.1 mode.
Apr 19 2016, 11:27 PM

Apr 18 2016

ashutosh.nema added a comment to D15922: [Cloning] cloneLoopWithPreheader(): add assert to ensure no sub-loops.

Indentation is still not as per llvm coding standard, please run clang-format.
You can specifically run it on the code added by you using below command:

Apr 18 2016, 10:20 PM

Apr 17 2016

ashutosh.nema added a comment to D15922: [Cloning] cloneLoopWithPreheader(): add assert to ensure no sub-loops.

I see lot of indentation problems here, please correct them.
Try using clang-format its useful.

Apr 17 2016, 10:50 PM
ashutosh.nema added a comment to D15922: [Cloning] cloneLoopWithPreheader(): add assert to ensure no sub-loops.

Thanks Vaivaswatha, looks OK to me except the name of function.
As I mentioned earlier ‘cloneInnermostLoopWithPreheader’ looks more intuitive
vs ‘cloneInnerLoopWithPreheader’ because we clone the innermost loop only.

Apr 17 2016, 10:42 PM

Apr 13 2016

ashutosh.nema added inline comments to D15922: [Cloning] cloneLoopWithPreheader(): add assert to ensure no sub-loops.
Apr 13 2016, 7:27 AM

Apr 5 2016

ashutosh.nema added inline comments to D18701: [ARM] Adding IEEE-754 SIMD detection to loop vectorizer.
Apr 5 2016, 6:59 AM
ashutosh.nema added inline comments to D18701: [ARM] Adding IEEE-754 SIMD detection to loop vectorizer.
Apr 5 2016, 4:27 AM

Mar 13 2016

ashutosh.nema added a comment to D18094: [x86, AVX] replace masked load with full vector load when possible.

Thanks Sanjay for working on this.

Mar 13 2016, 9:41 PM

Feb 12 2016

ashutosh.nema abandoned D7900: LoopVersioning for LICM.
Feb 12 2016, 3:46 AM

Feb 11 2016

ashutosh.nema added a comment to D17141: [ARM] Adding IEEE-754 SIMD detection to loop vectorizer.

Thanks Renato for working on this.

Feb 11 2016, 9:02 PM

Feb 8 2016

ashutosh.nema added a comment to rL259986: New Loop Versioning LICM Pass.

Thanks Nick, I will incorporate your comments.

Feb 8 2016, 9:48 PM

Feb 5 2016

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Thanks Hal.

Feb 5 2016, 11:55 PM

Feb 3 2016

ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Corrected include file order.

Feb 3 2016, 8:10 AM
ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Addressed Hal's comments.

Feb 3 2016, 8:02 AM
ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Thanks Hal for clarification, will make these changes and come back.

Feb 3 2016, 5:15 AM

Feb 2 2016

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Thanks Hal.

Feb 2 2016, 10:34 PM

Jan 26 2016

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Hi Hal,

Jan 26 2016, 10:26 PM

Jan 22 2016

ashutosh.nema added inline comments to D9151: Loop Versioning for LICM.
Jan 22 2016, 5:01 AM
ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Addressed Hal's & Adam's comments.

Jan 22 2016, 5:00 AM

Jan 18 2016

ashutosh.nema added inline comments to D9151: Loop Versioning for LICM.
Jan 18 2016, 10:34 PM

Jan 12 2016

ashutosh.nema added inline comments to D9151: Loop Versioning for LICM.
Jan 12 2016, 11:37 PM
ashutosh.nema added inline comments to D9151: Loop Versioning for LICM.
Jan 12 2016, 9:10 PM

Dec 22 2015

ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Addressed Hal's comment.

Dec 22 2015, 11:09 PM
ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Hi Hal,

Dec 22 2015, 12:49 AM

Dec 21 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Hi Tobias,

Dec 21 2015, 10:37 PM

Dec 20 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Hi Hal,

Dec 20 2015, 10:55 PM

Dec 14 2015

ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Thanks Hal, for again looking into this change.

Dec 14 2015, 1:48 AM

Nov 30 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Ping.

Nov 30 2015, 5:18 PM

Nov 12 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Hi Hal,

Nov 12 2015, 9:14 PM

Nov 5 2015

ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Incorporated Hal comments & updated LangRef.

Nov 5 2015, 1:18 AM
ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

This patch does not contain LangRef changes, I'll soon submitting it.

Nov 5 2015, 12:48 AM
ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Incorporated review comments from Hal.

Nov 5 2015, 12:39 AM

Oct 6 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Hal, is below comments on call handling looks OK to you ?

Oct 6 2015, 9:02 PM

Sep 28 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Thanks Hal for review.

Sep 28 2015, 10:06 AM

Sep 21 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Thanks Hal for looking into this again.

Sep 21 2015, 3:39 AM

Sep 6 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Ping.

Sep 6 2015, 9:31 PM

Sep 1 2015

ashutosh.nema added inline comments to D9151: Loop Versioning for LICM.
Sep 1 2015, 4:42 AM
ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

Incorporated review comments.

Sep 1 2015, 4:37 AM

Aug 31 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Thanks Charlie, I will correct these.

Aug 31 2015, 10:03 PM

Aug 26 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Ping.

Aug 26 2015, 8:41 PM

Aug 23 2015

ashutosh.nema added a comment to D9151: Loop Versioning for LICM.

Does this patch look OK to you now ?

Aug 23 2015, 9:35 PM

Aug 20 2015

ashutosh.nema closed D8653: Handling store to invariant address in LAA.
Aug 20 2015, 10:10 PM
ashutosh.nema closed D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided..

r244763

Aug 20 2015, 10:08 PM
ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.

LoopVersioning utility now calls ‘addPHINodes’ implicitly(r245579).
Updated LoopVersioningLICM to consider this change.

Aug 20 2015, 10:06 PM
ashutosh.nema closed D12071: Exposing findDefsUsedOutsideOfLoop as a Loop utility.

Revision: 245416

Aug 20 2015, 9:21 PM

Aug 19 2015

ashutosh.nema updated the diff for D9151: Loop Versioning for LICM.
  1. LoopVersioning utility is now available, so LoopVersioningLICM is using that utility.
Aug 19 2015, 5:21 AM

Aug 18 2015

ashutosh.nema added a comment to D12071: Exposing findDefsUsedOutsideOfLoop as a Loop utility.

Thanks Adam, will incorporate these comments and check-in.

Aug 18 2015, 8:41 PM

Aug 16 2015

ashutosh.nema retitled D12071: Exposing findDefsUsedOutsideOfLoop as a Loop utility from to Exposing findDefsUsedOutsideOfLoop as a Loop utility.
Aug 16 2015, 8:52 PM

Aug 11 2015

ashutosh.nema added a comment to D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided..

Thanks Adam.

Aug 11 2015, 10:08 PM
ashutosh.nema updated the diff for D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided..

Incorporated comments from Adam.

Aug 11 2015, 10:02 PM
ashutosh.nema updated the diff for D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided..

Incorporated comments from Adam.

Aug 11 2015, 9:03 PM
ashutosh.nema added inline comments to D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided..
Aug 11 2015, 7:56 PM

Aug 10 2015

ashutosh.nema updated the diff for D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided..

Incorporated comments from Adam.

Aug 10 2015, 11:14 PM
ashutosh.nema added a comment to D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided..

Sure Adam, will update it.

Aug 10 2015, 7:48 PM

Aug 9 2015

ashutosh.nema retitled D11892: LoopVersioning: Use default LAA runtimeCheck, When input check not provided. from to LoopVersioning: Use default LAA runtimeCheck, When input check not provided..
Aug 9 2015, 10:44 PM

Jul 27 2015

ashutosh.nema updated subscribers of D11043: Const fold vtable load from global variable.
Jul 27 2015, 3:29 AM

Jul 13 2015

ashutosh.nema abandoned D9521: Converting ‘sext of addrec’ to ‘addrec of sext’.
Jul 13 2015, 3:24 AM

Jul 12 2015

ashutosh.nema added a comment to D10835: getStrideFromPointer and its helper function from LoopVectorize moved to VectorUtils.

Thanks Hal.

Jul 12 2015, 9:09 PM

Jul 8 2015

ashutosh.nema updated the diff for D10835: getStrideFromPointer and its helper function from LoopVectorize moved to VectorUtils.

Incorporated comments from Hal.

Jul 8 2015, 10:10 PM
ashutosh.nema added a comment to D10835: getStrideFromPointer and its helper function from LoopVectorize moved to VectorUtils.

> Incorporated comments from Hal.

Hi Hal,

Does this looks OK to you ?

Yes, LGTM (with one small change in a comment requested below).

Jul 8 2015, 9:53 PM
ashutosh.nema added a comment to D10835: getStrideFromPointer and its helper function from LoopVectorize moved to VectorUtils.

Incorporated comments from Hal.

Jul 8 2015, 9:11 PM

Jul 7 2015

ashutosh.nema added a comment to D10835: getStrideFromPointer and its helper function from LoopVectorize moved to VectorUtils.

Ping.

Jul 7 2015, 7:34 AM