Page MenuHomePhabricator

andydavis1 (Andy Davis)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 7 2020, 4:33 PM (37 w, 1 d)

Recent Activity

Aug 4 2020

andydavis1 accepted D85177: [MLIR][Affine] Fix createPrivateMemRef in affine fusion.

Thanks Diego!

Aug 4 2020, 10:19 AM · Restricted Project

Jun 23 2020

andydavis1 accepted D82409: [MLIR] Fix affine loop fusion private memref alloc.
Jun 23 2020, 3:38 PM · Restricted Project

Jun 22 2020

andydavis1 accepted D82158: [MLIR][Affine-loop-fusion] Fix a bug in affine-loop-fusion pass when there are non-affine operations.

Thanks for addressing this issue! Let's see if Uday has any more comments before committing, but your latest changes look good to me...

Jun 22 2020, 11:17 AM · Restricted Project
andydavis1 added a comment to D80860: Exact integer emptiness checks for FlatAffineConstraints.

Thanks for your contribution!

Jun 22 2020, 10:44 AM · Restricted Project, Restricted Project

Jun 15 2020

andydavis1 accepted D81792: [MLIR] Fix memref region compute for 0-d memref accesses.
Jun 15 2020, 7:33 AM · Restricted Project

Jun 12 2020

andydavis1 accepted D81753: [mlir] Affine symbols: do not expect AffineScope to always exist.
Jun 12 2020, 12:35 PM · Restricted Project

Jun 11 2020

andydavis1 added a comment to D81609: [mlir][Affine] Revisit fusion candidates after successful fusion.

Thanks for improving this. Looks good to me...

Jun 11 2020, 4:33 PM · Restricted Project

Jun 2 2020

andydavis1 accepted D80971: [mlir][Affine] Enable fusion of loops with vector loads/stores.
Jun 2 2020, 10:57 AM · Restricted Project

Jun 1 2020

andydavis1 added a comment to D80860: Exact integer emptiness checks for FlatAffineConstraints.

Can you please add bondhugula@ as a reviewer too?

Jun 1 2020, 8:33 AM · Restricted Project, Restricted Project

May 15 2020

andydavis1 accepted D80017: [mlir] Fix incorrect indexing of subview in DimOp folding..
May 15 2020, 10:18 AM · Restricted Project

May 12 2020

andydavis1 accepted D79815: [mlir] [VectorOps] Implement vector.constant_mask lowering to LLVM IR.
May 12 2020, 6:20 PM · Restricted Project
andydavis1 added a comment to D79829: [mlir][Affine] Introduce affine memory interfaces.

Thanks for doing this Diego! Will be very nice to have this. Especially for affine loop fusion and dma generation....

May 12 2020, 6:20 PM · Restricted Project
andydavis1 accepted D79662: [mlir] Revisit std.subview handling of static information..
May 12 2020, 4:42 PM · Restricted Project

May 6 2020

andydavis1 accepted D79547: [MLIR] Fix affine fusion bug/efficiency issue / enable more fusion.

Thanks Uday! Looks good...

May 6 2020, 9:35 PM · Restricted Project
andydavis1 accepted D79502: [mlir] Support partial folding of affine.min/max.

Thanks Alex!

May 6 2020, 1:00 PM · Restricted Project

May 5 2020

andydavis1 committed rG93d1108801dd: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp. (authored by andydavis1).
[MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp.
May 5 2020, 10:48 AM
andydavis1 closed D79184: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp..
May 5 2020, 10:47 AM · Restricted Project
andydavis1 updated the diff for D79184: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp..

Rebase

May 5 2020, 10:46 AM · Restricted Project
andydavis1 accepted D79408: [mlir] Harden verifiers for DMA ops.

Thanks for doing this Alex. The test coverage is great. Thanks!

May 5 2020, 10:46 AM · Restricted Project
andydavis1 updated the diff for D79184: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp..

addressing review comments

May 5 2020, 10:14 AM · Restricted Project
andydavis1 added a comment to D79184: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp..

Thanks Alex! Will rebase this with changes in a bit...

May 5 2020, 10:14 AM · Restricted Project

May 1 2020

andydavis1 requested review of D79184: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp..
May 1 2020, 8:23 AM · Restricted Project
andydavis1 added inline comments to D79184: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp..
May 1 2020, 8:22 AM · Restricted Project

Apr 30 2020

andydavis1 accepted D79198: [mlir] [VectorOps] Included i1 support for vector.print.
Apr 30 2020, 2:00 PM · Restricted Project
andydavis1 created D79184: [MLIR][LoopOps] Adds the loop unroll transformation for loop::ForOp..
Apr 30 2020, 10:41 AM · Restricted Project

Apr 27 2020

andydavis1 accepted D78863: [MLIR] Introduce op trait PolyhedralScope.

Thanks Uday...

Apr 27 2020, 5:48 PM · Restricted Project

Apr 20 2020

andydavis1 accepted D78505: [MLIR] Fix affine loop tiling utility upper bound bug.
Apr 20 2020, 11:53 AM · Restricted Project

Apr 15 2020

andydavis1 accepted D77870: [MLIR] Introduce utility for affine loop unswitching / hoisting if/else.
Apr 15 2020, 10:56 AM · Restricted Project

Apr 13 2020

andydavis1 accepted D78055: [MLIR] remove dead affine.applys while generating pointwise copies.
Apr 13 2020, 2:13 PM · Restricted Project

Apr 10 2020

andydavis1 added inline comments to D77870: [MLIR] Introduce utility for affine loop unswitching / hoisting if/else.
Apr 10 2020, 8:27 AM · Restricted Project

Apr 8 2020

andydavis1 accepted D77738: [MLIR] Fix test case typos in unroll.mlir.
Apr 8 2020, 9:44 AM · Restricted Project
andydavis1 committed rG7006daa548c2: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value… (authored by andydavis1).
[MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value…
Apr 8 2020, 9:13 AM
andydavis1 closed D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..
Apr 8 2020, 9:13 AM · Restricted Project
andydavis1 updated the diff for D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..

rebase

Apr 8 2020, 9:12 AM · Restricted Project

Apr 7 2020

andydavis1 added inline comments to D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..
Apr 7 2020, 1:03 PM · Restricted Project
andydavis1 updated the diff for D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..

Adressing comments and rebasing.

Apr 7 2020, 1:03 PM · Restricted Project
andydavis1 updated the diff for D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..

addressing comments

Apr 7 2020, 11:55 AM · Restricted Project
andydavis1 added a comment to D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..

Thanks PTAL...

Apr 7 2020, 11:55 AM · Restricted Project

Apr 6 2020

andydavis1 accepted D77320: [MLIR] fix/update affine data copy utility for max/min bounds.
Apr 6 2020, 4:22 PM · Restricted Project

Apr 3 2020

andydavis1 added inline comments to D77320: [MLIR] fix/update affine data copy utility for max/min bounds.
Apr 3 2020, 9:40 AM · Restricted Project

Apr 2 2020

andydavis1 added inline comments to D77320: [MLIR] fix/update affine data copy utility for max/min bounds.
Apr 2 2020, 3:44 PM · Restricted Project

Apr 1 2020

andydavis1 retitled D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding. from BEGIN_PUBLIC [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding. to [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..
Apr 1 2020, 4:21 PM · Restricted Project
andydavis1 created D77253: [MLIR][Vector] Update ShapeCastOp folder to use producer-consumer value forwarding..
Apr 1 2020, 4:21 PM · Restricted Project
andydavis1 accepted D77237: [MLIR] Add method to drop duplicate result exprs from AffineMap.
Apr 1 2020, 2:13 PM · Restricted Project

Mar 31 2020

andydavis1 committed rG31a346cc35c8: [MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and… (authored by andydavis1).
[MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and…
Mar 31 2020, 8:53 AM
andydavis1 closed D76889: [MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and ExtractSlicesOp..
Mar 31 2020, 8:53 AM · Restricted Project
andydavis1 updated the diff for D76889: [MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and ExtractSlicesOp..

rebasing

Mar 31 2020, 8:49 AM · Restricted Project

Mar 30 2020

andydavis1 added a comment to D76889: [MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and ExtractSlicesOp..

PTAL. Let me know if there are any more comments...Thanks!

Mar 30 2020, 4:24 PM · Restricted Project
andydavis1 accepted D77004: [MLIR] Fix permuteLoops utility.

Thanks for fixing this bug Uday!

Mar 30 2020, 8:03 AM · Restricted Project

Mar 27 2020

andydavis1 accepted D76700: [MLIR] Introduce full/partial tile separation using if/else.
Mar 27 2020, 3:27 PM · Restricted Project
andydavis1 updated the diff for D76889: [MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and ExtractSlicesOp..

addressing feedback

Mar 27 2020, 12:01 PM · Restricted Project
andydavis1 added a comment to D76889: [MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and ExtractSlicesOp..

Thanks. I've addressed comments and will update with a new patch soon...

Mar 27 2020, 9:05 AM · Restricted Project

Mar 26 2020

andydavis1 created D76889: [MLIR][Vector] Add support for TupleGetOp folding through InsertSlicesOp and ExtractSlicesOp..
Mar 26 2020, 3:14 PM · Restricted Project

Mar 25 2020

andydavis1 added inline comments to D76700: [MLIR] Introduce full/partial tile separation using if/else.
Mar 25 2020, 5:21 PM · Restricted Project
andydavis1 accepted D76701: [MLIR] Add flat affine constraints method to round trip integer set.
Mar 25 2020, 3:11 PM · Restricted Project

Mar 23 2020

andydavis1 accepted D76667: [MLIR][NFC] flat affine constraints - refactor to share, renames.
Mar 23 2020, 10:17 PM · Restricted Project
andydavis1 accepted D76633: [MLIR] move loopUnrollJamBy*Factor to loop transforms utils.
Mar 23 2020, 1:06 PM · Restricted Project

Mar 20 2020

andydavis1 added a comment to D76460: [mlir] [VectorOps] Introduce vector.transpose.

Thanks Aart. Yes, if its not too much work, I'd love to see this op take a permutation of dimensions of an N-D vector.

Mar 20 2020, 10:17 AM · Restricted Project

Mar 19 2020

andydavis1 added a comment to D76460: [mlir] [VectorOps] Introduce vector.transpose.

Thanks Aart. How hard would this be to generalize this op to make it N-D transpose op, instead of limit to 2D cases? For example, I may want to transpose the inner two dimensions of a N-D vector

Mar 19 2020, 5:00 PM · Restricted Project

Mar 13 2020

andydavis1 accepted D76117: NFC: modernize / clean up some loop transform utils, affine analysis utils.

Nice clean up!

Mar 13 2020, 7:29 AM · Restricted Project

Mar 12 2020

andydavis1 accepted D75956: [mlir] [VectorOps] Progressively lower vector.outerproduct to LLVM.

Aart. This looks good to me, but please wait for Nicolas' feedback as well. I think the two of you have chatted more about the outer-product stuff...

Mar 12 2020, 10:18 AM · Restricted Project

Mar 9 2020

andydavis1 accepted D75792: [mlir] NFC: remove IntegerValueSet / MutableIntegerSet.
Mar 9 2020, 10:46 AM · Restricted Project

Mar 5 2020

andydavis1 accepted D75694: NFC: remove stray decl of toAffineExpr, rename for readability.
Mar 5 2020, 10:57 AM · Restricted Project

Feb 24 2020

andydavis1 accepted D75058: [mlir] simplify affine maps and operands in affine.min/max.

Thanks Alex!

Feb 24 2020, 8:58 AM · Restricted Project

Feb 20 2020

andydavis1 accepted D74880: [mlir] [VectorOps] Multi-dim reductions for lowering vector.contract.

Looks good, thanks Aart! One question, is the unrolling here deterministic in that the same op will always produce the same sequence of reductions after unrolling? Context is numerical stability across test runs....

Feb 20 2020, 1:25 PM · Restricted Project

Feb 19 2020

andydavis1 accepted D74797: [mlir] [VectorOps] Framework for progressive lowering of vector.contract.

Thanks Aart!

Feb 19 2020, 11:09 AM · Restricted Project

Feb 18 2020

andydavis1 added inline comments to D74797: [mlir] [VectorOps] Framework for progressive lowering of vector.contract.
Feb 18 2020, 2:53 PM · Restricted Project
andydavis1 accepted D74658: [mlir][NFC] Fix 'gatherLoops' utility.
Feb 18 2020, 12:44 PM · Restricted Project

Feb 11 2020

andydavis1 committed rG40b2eb353043: [mlir][AffineOps] Adds affine loop fusion transformation function to… (authored by andydavis1).
[mlir][AffineOps] Adds affine loop fusion transformation function to…
Feb 11 2020, 2:04 PM
andydavis1 closed D74330: [mlir][AffineOps] Adds affine loop fusion transformation function to LoopFusionUtils..
Feb 11 2020, 2:04 PM · Restricted Project
andydavis1 updated the diff for D74330: [mlir][AffineOps] Adds affine loop fusion transformation function to LoopFusionUtils..

addressing review comments

Feb 11 2020, 1:36 PM · Restricted Project
andydavis1 added a comment to D74330: [mlir][AffineOps] Adds affine loop fusion transformation function to LoopFusionUtils..

addressing comments.

Feb 11 2020, 1:36 PM · Restricted Project
andydavis1 added a comment to D74330: [mlir][AffineOps] Adds affine loop fusion transformation function to LoopFusionUtils..

addressing comments.

Feb 11 2020, 1:27 PM · Restricted Project
andydavis1 committed rG813bfffec34b: [mlir][VectorOps] Adds canonicalization rewrite patterns for vector ShapeCastOp. (authored by andydavis1).
[mlir][VectorOps] Adds canonicalization rewrite patterns for vector ShapeCastOp.
Feb 11 2020, 1:19 PM
andydavis1 closed D74327: [mlir][VectorOps] Adds canonicalization rewrite patterns for vector ShapeCastOp..
Feb 11 2020, 1:19 PM · Restricted Project
andydavis1 updated the diff for D74327: [mlir][VectorOps] Adds canonicalization rewrite patterns for vector ShapeCastOp..

rebasing with head

Feb 11 2020, 1:00 PM · Restricted Project
andydavis1 added a comment to D74327: [mlir][VectorOps] Adds canonicalization rewrite patterns for vector ShapeCastOp..

addressing comments

Feb 11 2020, 1:00 PM · Restricted Project

Feb 10 2020

andydavis1 created D74330: [mlir][AffineOps] Adds affine loop fusion transformation function to LoopFusionUtils..
Feb 10 2020, 8:43 AM · Restricted Project
andydavis1 created D74327: [mlir][VectorOps] Adds canonicalization rewrite patterns for vector ShapeCastOp..
Feb 10 2020, 8:11 AM · Restricted Project

Feb 6 2020

andydavis1 added a comment to D73190: [mlir] Adds affine loop fusion transformation function to LoopFusionUtils..

Thanks. A fix is on the way...

Feb 6 2020, 9:53 AM · Restricted Project

Feb 5 2020

andydavis1 committed rGf9efce1dd5ce: [mlir][VectorOps] Support vector transfer_read/write unrolling for memrefs with… (authored by andydavis1).
[mlir][VectorOps] Support vector transfer_read/write unrolling for memrefs with…
Feb 5 2020, 4:29 PM
andydavis1 accepted D74014: [mlir] [VectorOps] generalized vector.contract semantics.

Looks good from my perspective.

Feb 5 2020, 4:29 PM · Restricted Project
andydavis1 closed D72965: [mlir][VectorOps] Support vector transfer_read/write unrolling for memrefs with vector element type..
Feb 5 2020, 4:29 PM · Restricted Project
andydavis1 updated the diff for D72965: [mlir][VectorOps] Support vector transfer_read/write unrolling for memrefs with vector element type..

rebasing

Feb 5 2020, 4:10 PM · Restricted Project
andydavis1 committed rG64871f778d52: [mlir] Adds affine loop fusion transformation function to LoopFusionUtils. (authored by andydavis1).
[mlir] Adds affine loop fusion transformation function to LoopFusionUtils.
Feb 5 2020, 4:01 PM
andydavis1 closed D73190: [mlir] Adds affine loop fusion transformation function to LoopFusionUtils..
Feb 5 2020, 4:01 PM · Restricted Project
andydavis1 updated the diff for D73190: [mlir] Adds affine loop fusion transformation function to LoopFusionUtils..

rebasing with head

Feb 5 2020, 3:52 PM · Restricted Project
andydavis1 committed rG3ce8095c295e: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect. (authored by andydavis1).
[mlir][VectorOps] Add ShapeCastOp to the vector ops dialect.
Feb 5 2020, 3:52 PM
andydavis1 closed D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..
Feb 5 2020, 3:51 PM · Restricted Project
andydavis1 updated the diff for D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

rebasing with head.

Feb 5 2020, 3:39 PM · Restricted Project
andydavis1 updated the diff for D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

fixing formating take 2

Feb 5 2020, 3:20 PM · Restricted Project
andydavis1 updated the diff for D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

fixed format.

Feb 5 2020, 1:58 PM · Restricted Project
andydavis1 updated the diff for D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

addressing comments

Feb 5 2020, 1:40 PM · Restricted Project
andydavis1 added a comment to D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

addressing comments

Feb 5 2020, 1:30 PM · Restricted Project
andydavis1 updated the diff for D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

rebasing with head

Feb 5 2020, 12:44 PM · Restricted Project

Feb 4 2020

andydavis1 accepted D73923: [mlir] [VectorOps] refined description of vector.contract.

Thanks Aart

Feb 4 2020, 10:58 AM · Restricted Project

Jan 31 2020

andydavis1 updated the diff for D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

Sync with github master

Jan 31 2020, 3:21 PM · Restricted Project
andydavis1 retitled D72965: [mlir][VectorOps] Support vector transfer_read/write unrolling for memrefs with vector element type. from Support vector transfer_read/write unrolling for memrefs with vector element type. to [mlir][VectorOps] Support vector transfer_read/write unrolling for memrefs with vector element type..
Jan 31 2020, 3:12 PM · Restricted Project
andydavis1 added a comment to D72965: [mlir][VectorOps] Support vector transfer_read/write unrolling for memrefs with vector element type..

updated tags

Jan 31 2020, 3:12 PM · Restricted Project
andydavis1 added a comment to D73635: [mlir][VectorOps] Add ShapeCastOp to the vector ops dialect..

Addressed River's comment.

Jan 31 2020, 2:16 PM · Restricted Project