Page MenuHomePhabricator

dcaballe (Diego Caballero)
User

Projects

User does not belong to any projects.

User Details

User Since
Sep 16 2016, 11:47 AM (213 w, 3 d)

Recent Activity

Fri, Oct 2

dcaballe added a comment to D88544: [MLIR] Implemented iterative topological sorting.

Hey! Thanks for the patch! I wondered if it would make sense to move this utility to a public utility header. I found myself needing it locally and I wouldn't like to just copy-paste it. There is another topological sort for ops under 'include/mlir/Analysis/SliceAnalysis.h'. Maybe we could move them both to 'include/mlir/IR/Utils.h'? WDYT?

Fri, Oct 2, 10:34 AM · Restricted Project
dcaballe committed rGa611f9a5c689: [mlir] Fix call op conversion in bare-ptr calling convention (authored by dcaballe).
[mlir] Fix call op conversion in bare-ptr calling convention
Fri, Oct 2, 8:59 AM
dcaballe closed D88709: [mlir] Fix call op conversion in bare-ptr calling convention.
Fri, Oct 2, 8:59 AM · Restricted Project

Thu, Oct 1

dcaballe requested review of D88709: [mlir] Fix call op conversion in bare-ptr calling convention.
Thu, Oct 1, 4:17 PM · Restricted Project
dcaballe added inline comments to D88362: Implement callee/caller type checking for llvm.call.
Thu, Oct 1, 4:09 PM · Restricted Project

Tue, Sep 29

dcaballe added inline comments to D88362: Implement callee/caller type checking for llvm.call.
Tue, Sep 29, 12:19 PM · Restricted Project
dcaballe committed rGa89fc12653c5: [mlir] Support return and call ops in bare-ptr calling convention (authored by dcaballe).
[mlir] Support return and call ops in bare-ptr calling convention
Tue, Sep 29, 12:10 PM
dcaballe closed D87724: [mlir] Support unranked memrefs, return and call ops in bare-ptr calling convention.
Tue, Sep 29, 12:09 PM · Restricted Project
dcaballe added a comment to D87724: [mlir] Support unranked memrefs, return and call ops in bare-ptr calling convention.

Thanks, Alex. Will do.

Tue, Sep 29, 11:32 AM · Restricted Project

Mon, Sep 28

dcaballe committed rG93936da90479: [mlir][Affine][VectorOps] Fix super vectorizer utility (D85869) (authored by dcaballe).
[mlir][Affine][VectorOps] Fix super vectorizer utility (D85869)
Mon, Sep 28, 4:33 PM
dcaballe closed D88346: [mlir][Affine][VectorOps] Fix super vectorizer utility (D85869).
Mon, Sep 28, 4:33 PM · Restricted Project

Sat, Sep 26

dcaballe accepted D88362: Implement callee/caller type checking for llvm.call.

Thanks, LGTM! Please, give some time for @ftynse to have a look.

Sat, Sep 26, 8:52 PM · Restricted Project
dcaballe added a comment to D88346: [mlir][Affine][VectorOps] Fix super vectorizer utility (D85869).

Thanks, Nicolas. The failure doesn't seem related to this:

Sat, Sep 26, 8:32 PM · Restricted Project

Fri, Sep 25

dcaballe requested review of D88346: [mlir][Affine][VectorOps] Fix super vectorizer utility (D85869).
Fri, Sep 25, 5:19 PM · Restricted Project

Thu, Sep 24

dcaballe abandoned D88042: [mlir] Fix LLVM lowering of memory space in unranked memrefs.

Sorry, I misunderstood what an unranked memref models. This indeed doesn't make sense as is. Thanks.

Thu, Sep 24, 7:43 PM · Restricted Project
dcaballe updated the diff for D87724: [mlir] Support unranked memrefs, return and call ops in bare-ptr calling convention.
  • Removed support for unranked memrefs
  • Addressed feedback
Thu, Sep 24, 7:40 PM · Restricted Project
dcaballe added a comment to D87724: [mlir] Support unranked memrefs, return and call ops in bare-ptr calling convention.

Thanks, Alex! Sorry, I misunderstood what an unranked memref is modeling. Let me remove that from the patch since it's incorrect and only leave the support for call and return ops. We can add support for unranked memrefs when needed.

Thu, Sep 24, 7:40 PM · Restricted Project
dcaballe committed rG0a925a813a50: [mlir][NFC] Promote memory space to BaseMemRefType (authored by dcaballe).
[mlir][NFC] Promote memory space to BaseMemRefType
Thu, Sep 24, 2:08 PM
dcaballe closed D87649: [mlir][NFC] Promote memory space to BaseMemRefType.
Thu, Sep 24, 2:07 PM · Restricted Project

Wed, Sep 23

dcaballe added a comment to D87649: [mlir][NFC] Promote memory space to BaseMemRefType.

Thanks!

Wed, Sep 23, 7:04 PM · Restricted Project

Mon, Sep 21

dcaballe committed rG14d0735d3453: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy (authored by dcaballe).
[MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy
Mon, Sep 21, 4:38 PM
dcaballe closed D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.
Mon, Sep 21, 4:37 PM · Restricted Project
dcaballe added a comment to D87649: [mlir][NFC] Promote memory space to BaseMemRefType.

Ping :)

Mon, Sep 21, 3:34 PM · Restricted Project
dcaballe added a comment to D87724: [mlir] Support unranked memrefs, return and call ops in bare-ptr calling convention.

As a general observation, do you folks actually want the bare-pointer convention, or is it only there to attach noalias argument attribute? Maybe we should consider investing into expressing aliasing information in MLIR rather than keep the maintenance burden.

Mon, Sep 21, 3:33 PM · Restricted Project
dcaballe updated the diff for D87724: [mlir] Support unranked memrefs, return and call ops in bare-ptr calling convention.
  • Refactor code to a utility function.
  • Replace i64 type with index type.
  • Address Alex's feedback.
Mon, Sep 21, 3:32 PM · Restricted Project
dcaballe requested review of D88042: [mlir] Fix LLVM lowering of memory space in unranked memrefs.
Mon, Sep 21, 1:35 PM · Restricted Project

Sep 15 2020

dcaballe updated the diff for D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.

Sorry for the delay and sorry, Nicolas, I had written the example that
you were asking for in the .cpp doc but I forgot to copy it to the header
file. This patch is doing that. Please, let me know if there are any other
comments. Thanks!

Sep 15 2020, 2:32 PM · Restricted Project
dcaballe committed rG609f5e050cea: [mlir] Rename 'setInsertionPointAfter' to avoid ambiguity (authored by dcaballe).
[mlir] Rename 'setInsertionPointAfter' to avoid ambiguity
Sep 15 2020, 2:18 PM
dcaballe closed D87155: [mlir] Rename 'setInsertionPointAfter' to avoid ambiguity.
Sep 15 2020, 2:17 PM · Restricted Project
dcaballe requested review of D87724: [mlir] Support unranked memrefs, return and call ops in bare-ptr calling convention.
Sep 15 2020, 2:01 PM · Restricted Project

Sep 14 2020

dcaballe requested review of D87649: [mlir][NFC] Promote memory space to BaseMemRefType.
Sep 14 2020, 4:01 PM · Restricted Project
dcaballe added a comment to D86979: [docs][NewPM] Add docs for writing NPM passes.

This commit seems to be breaking the build with -DBUILD_SHARED_LIBS=TRUE. Could you please have a quick look? Thanks!

Sep 14 2020, 3:30 PM · Restricted Project

Sep 10 2020

dcaballe added a comment to D87155: [mlir] Rename 'setInsertionPointAfter' to avoid ambiguity.

Ping :)

Sep 10 2020, 3:12 PM · Restricted Project
dcaballe added a comment to D87437: [MLIR] Support for return values in Affine.For yield.

LGTM! Just some nit comments! Thanks for contributing this feature!

Sep 10 2020, 9:46 AM · Restricted Project

Sep 4 2020

dcaballe requested review of D87155: [mlir] Rename 'setInsertionPointAfter' to avoid ambiguity.
Sep 4 2020, 12:01 PM · Restricted Project
dcaballe added inline comments to D86756: [MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer.
Sep 4 2020, 11:22 AM · Restricted Project

Sep 2 2020

dcaballe committed rG46781630a34e: [MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer (authored by dcaballe).
[MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer
Sep 2 2020, 3:19 PM
dcaballe closed D86756: [MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer.
Sep 2 2020, 3:19 PM · Restricted Project
dcaballe committed rG553bfc8fa17f: [mlir][Affine] Support affine vector loads/stores in LICM (authored by dcaballe).
[mlir][Affine] Support affine vector loads/stores in LICM
Sep 2 2020, 2:45 PM
dcaballe closed D86986: [mlir][Affine] Support affine vector loads/stores in LICM.
Sep 2 2020, 2:45 PM · Restricted Project
dcaballe committed rG65f20ea1133b: [mlir][Affine] Fix AffineLoopInvariantCodeMotion (authored by dcaballe).
[mlir][Affine] Fix AffineLoopInvariantCodeMotion
Sep 2 2020, 2:15 PM
dcaballe closed D86982: [mlir][Affine] Fix AffineLoopInvariantCodeMotion.
Sep 2 2020, 2:15 PM · Restricted Project
dcaballe added a comment to D86756: [MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer.

Thanks! Any other comments, @aartbik?

Sep 2 2020, 11:38 AM · Restricted Project
dcaballe added a comment to D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.

Thanks! Any other comments? @aartbik?

Sep 2 2020, 11:36 AM · Restricted Project
dcaballe updated the diff for D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.

Addressed feedback. Thanks!

Sep 2 2020, 11:36 AM · Restricted Project
dcaballe updated the diff for D86982: [mlir][Affine] Fix AffineLoopInvariantCodeMotion.

Addressed feedback. Thanks!

Sep 2 2020, 9:47 AM · Restricted Project
dcaballe added a comment to D86986: [mlir][Affine] Support affine vector loads/stores in LICM.

I'd reword the commit summary a bit to replace "Introduce" - instead, something like "Make use of ..."

Sep 2 2020, 9:41 AM · Restricted Project

Sep 1 2020

dcaballe added a comment to D86839: [CMake] Remove -Wl,-allow-shlib-undefined which was added in rL221530.

Thanks for the patch! I can confirm that it is fixing our build locally (using gcc 7.5.0).

Sep 1 2020, 5:29 PM · Restricted Project
dcaballe requested review of D86986: [mlir][Affine] Support affine vector loads/stores in LICM.
Sep 1 2020, 3:58 PM · Restricted Project
dcaballe requested review of D86982: [mlir][Affine] Fix AffineLoopInvariantCodeMotion.
Sep 1 2020, 3:01 PM · Restricted Project

Aug 31 2020

dcaballe added inline comments to D86756: [MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer.
Aug 31 2020, 4:26 PM · Restricted Project
dcaballe updated the diff for D86756: [MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer.

Thanks! Addressing feedback.

Aug 31 2020, 4:25 PM · Restricted Project
dcaballe added inline comments to D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.
Aug 31 2020, 3:09 PM · Restricted Project
dcaballe updated the diff for D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.

Addressed all the comments and replaced the DFS loop container in the API with
an 'std::vector<SmallVector<AffineForOp, 2>>' (we use something similar in other
loop utilities). This container is a bit more friendly for a public API than
providing a plain list of loops in DFS order and also explicitly represents the
nesting level relationship of the loops that we want to vectorized, which is
useful for the vectorization algorithm itself. The vectorizer is therefore updated
to work on this container.

Aug 31 2020, 3:09 PM · Restricted Project

Aug 27 2020

dcaballe planned changes to D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.

Thanks, Aart! Sorry for the delay. I've been OOO. I'll take a look at your comments and I'll also try to use 'std::vector<SmallVector<AffineForOp, 2>>' for the API so that we can preserve some kind of loop nest structure.

Aug 27 2020, 7:49 PM · Restricted Project
dcaballe requested review of D86756: [MLIR][Affine][VectorOps] Vectorize uniform values in SuperVectorizer.
Aug 27 2020, 7:38 PM · Restricted Project

Aug 24 2020

dcaballe added a comment to D86380: [MLIR][Affine] Add custom builders for AffineVectorLoadOp/AffineVectorStoreOp.

Thanks Frank! Sorry, I was OOO last week. Quick comments for now, I'll come back later.

Aug 24 2020, 10:07 PM · Restricted Project

Aug 16 2020

dcaballe added a comment to D85899: [mlir] do not emit bitcasts between structs in StandardToLLVM.

Thanks, Alex!

Aug 16 2020, 6:22 PM · Restricted Project
dcaballe resigned from D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Unblocking the review from my side. OOO next week.

Aug 16 2020, 6:19 PM · Restricted Project
dcaballe added a comment to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Thanks for addressing the comments, Frank! LGTM. I'll leave the final approval to Uday and River since I'm OOO next week.

Aug 16 2020, 6:17 PM · Restricted Project

Aug 12 2020

dcaballe requested review of D85869: [MLIR][Affine][VectorOps] Utility to vectorize loop nest using strategy.
Aug 12 2020, 5:30 PM · Restricted Project

Aug 10 2020

dcaballe added inline comments to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 10 2020, 2:07 PM · Restricted Project

Aug 6 2020

dcaballe updated subscribers of D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 6 2020, 11:56 AM · Restricted Project

Aug 5 2020

dcaballe requested changes to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Thanks Frank! It looks good to me. Just a few minor comments

Aug 5 2020, 4:12 PM · Restricted Project

Aug 4 2020

dcaballe committed rG3bfbc5df87cb: [MLIR][Affine] Fix createPrivateMemRef in affine fusion (authored by dcaballe).
[MLIR][Affine] Fix createPrivateMemRef in affine fusion
Aug 4 2020, 12:34 PM
dcaballe closed D85177: [MLIR][Affine] Fix createPrivateMemRef in affine fusion.
Aug 4 2020, 12:34 PM · Restricted Project
dcaballe added inline comments to D85177: [MLIR][Affine] Fix createPrivateMemRef in affine fusion.
Aug 4 2020, 11:39 AM · Restricted Project
dcaballe added inline comments to D85177: [MLIR][Affine] Fix createPrivateMemRef in affine fusion.
Aug 4 2020, 7:50 AM · Restricted Project

Aug 3 2020

dcaballe requested review of D85177: [MLIR][Affine] Fix createPrivateMemRef in affine fusion.
Aug 3 2020, 7:27 PM · Restricted Project
dcaballe added a comment to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Thanks for the patch! Just a high-level comment for now. It looks good overall.

Aug 3 2020, 11:11 AM · Restricted Project

Jul 28 2020

dcaballe committed rG834133c950fc: [MLIR] Vector store to load forwarding (authored by anandkodnani).
[MLIR] Vector store to load forwarding
Jul 28 2020, 11:41 AM
dcaballe closed D84302: [MLIR] Vector store to load forwarding.
Jul 28 2020, 11:41 AM · Restricted Project

Jul 27 2020

dcaballe accepted D84302: [MLIR] Vector store to load forwarding.
Jul 27 2020, 10:32 PM · Restricted Project

Jul 23 2020

dcaballe added inline comments to D84302: [MLIR] Vector store to load forwarding.
Jul 23 2020, 11:08 AM · Restricted Project

Jul 22 2020

dcaballe committed rG3fff5acd8f5f: [mlir][VectorOps] Expose SuperVectorizer as a utility (authored by dcaballe).
[mlir][VectorOps] Expose SuperVectorizer as a utility
Jul 22 2020, 2:36 PM
dcaballe closed D84289: [mlir][VectorOps] Expose SuperVectorizer as a utility.
Jul 22 2020, 2:36 PM · Restricted Project
dcaballe added reviewers for D84302: [MLIR] Vector store to load forwarding: bondhugula, ftynse.
Jul 22 2020, 9:24 AM · Restricted Project

Jul 21 2020

Herald added a project to D84289: [mlir][VectorOps] Expose SuperVectorizer as a utility: Restricted Project.
Jul 21 2020, 6:46 PM · Restricted Project
dcaballe committed rGf8b72fba868b: [MLIR][EDSC] Add fptrunc and fpext to EDSC (authored by dcaballe).
[MLIR][EDSC] Add fptrunc and fpext to EDSC
Jul 21 2020, 9:06 AM
dcaballe closed D84216: [MLIR][EDSC] Add fptrunc and fpext to EDSC.
Jul 21 2020, 9:05 AM · Restricted Project

Jul 20 2020

Herald added a project to D84216: [MLIR][EDSC] Add fptrunc and fpext to EDSC: Restricted Project.
Jul 20 2020, 9:16 PM · Restricted Project

Jun 26 2020

dcaballe added a comment to D82600: IR changes to add yield semantics for affine.if and affine.parallel, additional affine.parallel canonicalization.

Thanks Jeremy! Some minor comments.

Jun 26 2020, 3:39 PM · Restricted Project
dcaballe added a reviewer for D82600: IR changes to add yield semantics for affine.if and affine.parallel, additional affine.parallel canonicalization: ftynse.
Jun 26 2020, 2:41 PM · Restricted Project

Jun 25 2020

dcaballe committed rGa72887831a22: [mlir][EDSC] Add divis and diviu and vector.extractelement (authored by dcaballe).
[mlir][EDSC] Add divis and diviu and vector.extractelement
Jun 25 2020, 8:35 AM
dcaballe closed D82515: [mlir][EDSL] Add divis and diviu and vector.extractelement.
Jun 25 2020, 8:35 AM · Restricted Project

Jun 24 2020

dcaballe created D82515: [mlir][EDSL] Add divis and diviu and vector.extractelement.
Jun 24 2020, 5:56 PM · Restricted Project

Jun 23 2020

dcaballe accepted D82409: [MLIR] Fix affine loop fusion private memref alloc.

Thanks for fixing this! LGTM

Jun 23 2020, 4:11 PM · Restricted Project

Jun 15 2020

dcaballe accepted D81792: [MLIR] Fix memref region compute for 0-d memref accesses.
Jun 15 2020, 9:13 AM · Restricted Project

Jun 11 2020

dcaballe committed rG2e7a0845913e: [mlir][Affine] Revisit fusion candidates after successful fusion (authored by dcaballe).
[mlir][Affine] Revisit fusion candidates after successful fusion
Jun 11 2020, 3:28 PM
dcaballe closed D81609: [mlir][Affine] Revisit fusion candidates after successful fusion.
Jun 11 2020, 3:27 PM · Restricted Project

Jun 10 2020

dcaballe created D81609: [mlir][Affine] Revisit fusion candidates after successful fusion.
Jun 10 2020, 12:47 PM · Restricted Project

Jun 8 2020

dcaballe added a comment to D80971: [mlir][Affine] Enable fusion of loops with vector loads/stores.

Thanks Uday! Good point. I haven't looked too much into the slice computation but if you can send me some pointers and some examples on how the regions should be computed I could have a look. Thanks!

Jun 8 2020, 11:36 AM · Restricted Project

Jun 5 2020

dcaballe committed rG7d59f49bdadd: [mlir] Fix representation of BF16 constants (authored by dcaballe).
[mlir] Fix representation of BF16 constants
Jun 5 2020, 5:55 PM
dcaballe committed rGa2588948febc: Fix convertBFloatAPFloatToAPInt for NaN/Inf values (authored by dcaballe).
Fix convertBFloatAPFloatToAPInt for NaN/Inf values
Jun 5 2020, 5:55 PM
dcaballe closed D81218: [mlir] Fix representation of BF16 constants.
Jun 5 2020, 5:55 PM · Restricted Project
dcaballe closed D81302: Fix convertBFloatAPFloatToAPInt for NaN/Inf values.
Jun 5 2020, 5:54 PM · Restricted Project
dcaballe added a comment to D81218: [mlir] Fix representation of BF16 constants.

If no more comments, I'll land this after https://reviews.llvm.org/D81302

Jun 5 2020, 2:02 PM · Restricted Project
dcaballe created D81302: Fix convertBFloatAPFloatToAPInt for NaN/Inf values.
Jun 5 2020, 1:30 PM · Restricted Project

Jun 4 2020

dcaballe added inline comments to D81218: [mlir] Fix representation of BF16 constants.
Jun 4 2020, 8:56 PM · Restricted Project
dcaballe created D81218: [mlir] Fix representation of BF16 constants.
Jun 4 2020, 8:23 PM · Restricted Project