Page MenuHomePhabricator

flaub (Frank Laub)
User

Projects

User does not belong to any projects.

User Details

User Since
Dec 21 2019, 2:19 AM (38 w, 6 d)

Recent Activity

Sat, Aug 22

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

Can you please separate out the fix to normalizeAffineParallel? It has really nothing to do with the builder additions.

Sat, Aug 22, 7:48 PM · Restricted Project
flaub added inline comments to D86380: [MLIR][Affine] Add custom builders for AffineVectorLoadOp/AffineVectorStoreOp.
Sat, Aug 22, 7:45 PM · Restricted Project
flaub added inline comments to D86380: [MLIR][Affine] Add custom builders for AffineVectorLoadOp/AffineVectorStoreOp.
Sat, Aug 22, 7:44 PM · Restricted Project

Fri, Aug 21

flaub requested review of D86380: [MLIR][Affine] Add custom builders for AffineVectorLoadOp/AffineVectorStoreOp.
Fri, Aug 21, 3:31 PM · Restricted Project

Thu, Aug 20

flaub added a comment to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Thanks for the review everyone, I've made tiny tweaks after further feedback. This includes extracting the core transformation of normalization into a separate utility function (in case other users want to integrate this transformation in a different pass), and also adding a helper method AffineParallelOp::getSteps().

Thu, Aug 20, 3:26 PM · Restricted Project
flaub committed rGcca3f3dd2681: [MLIR] Add affine.parallel folder and normalizer (authored by flaub).
[MLIR] Add affine.parallel folder and normalizer
Thu, Aug 20, 3:24 PM
flaub closed D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Thu, Aug 20, 3:24 PM · Restricted Project
flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Thu, Aug 20, 3:15 PM · Restricted Project

Wed, Aug 19

flaub added a comment to D85557: Implement FPToUI and UIToFP ops in standard dialect.

Thanks @ftynse :)

Wed, Aug 19, 2:22 PM · Restricted Project

Aug 18 2020

flaub added a comment to D85557: Implement FPToUI and UIToFP ops in standard dialect.

Friendly poke, wondering what the status of this is. Looks like we have an RFC here https://llvm.discourse.group/t/rfc-standard-add-std-fptoui-and-std-uitofp-operations/1567. AFAICT, there aren't any outstanding issues. This is a required op for certain situations. Because the standard dialect only deals with 'signless' integers, we must be able to reintroduce signedness in a few key areas, one of them being casts, otherwise we get incorrect behavior (we've run into this in real-world situations).

Aug 18 2020, 8:57 PM · Restricted Project
flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 18 2020, 8:52 PM · Restricted Project
flaub added a comment to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Hey Frank, didn't mean for this to force you to scale back the patch. If it helps I can implement the pattern rewriter hooks that you would need. Just let me know.

  • River
Aug 18 2020, 8:40 PM · Restricted Project

Aug 14 2020

flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

After some more discussions and thought, I've removed the complex canonicalizations that:

  • break the PatternRewriter rules
  • have hacky special exemptions
Aug 14 2020, 12:29 PM · Restricted Project
flaub accepted D85900: [mlir] do not use llvm.cmpxchg with floats.

Looks like xchg can handle floats but not cmpxchg, so I'm in favor of proceeding with this. Might want to also adjust the IR to reflect the tighter constraints.

Aug 14 2020, 10:32 AM · Restricted Project

Aug 13 2020

flaub added a comment to D85900: [mlir] do not use llvm.cmpxchg with floats.

Additionally, there are some operations that can't feasibly be implemented without floating-point cmpxchg working. I can see wanting to unblock these tests, but without changes to the verification of the GenericAtomicRMWOp, it will be possible for users to create floating-point cmpxchg.

Aug 13 2020, 12:18 PM · Restricted Project

Aug 9 2020

flaub added a comment to D85587: [MLIR][NFC] Fix misleading diagnostic error + clang-tidy fix.

LGTM, thanks!

Aug 9 2020, 6:03 PM · Restricted Project
flaub added a comment to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

@dcaballe & @bondhugula Is there anything else you'd like for me to address?

Aug 9 2020, 5:23 PM · Restricted Project

Aug 6 2020

flaub updated the summary of D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 6 2020, 2:13 PM · Restricted Project
flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Address comments.

Aug 6 2020, 2:08 PM · Restricted Project
flaub added inline comments to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 6 2020, 1:39 PM · Restricted Project
flaub added inline comments to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 6 2020, 1:30 PM · Restricted Project
flaub added inline comments to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 6 2020, 1:27 PM · Restricted Project
flaub added inline comments to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 6 2020, 12:55 PM · Restricted Project
flaub added inline comments to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 6 2020, 11:04 AM · Restricted Project

Aug 5 2020

flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 5 2020, 2:13 PM · Restricted Project
flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

@rriddle I've attempted to follow your advice, could you take another look? Is this what you were thinking of?

Aug 5 2020, 2:08 PM · Restricted Project

Aug 3 2020

flaub added inline comments to D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Aug 3 2020, 3:59 PM · Restricted Project

Jul 31 2020

flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Address CR comments

Jul 31 2020, 11:58 AM · Restricted Project

Jul 30 2020

flaub updated the diff for D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.

Fix pre-merge lint check

Jul 30 2020, 8:47 PM · Restricted Project
flaub requested review of D84998: [MLIR] Add affine.parallel folder and AffineParallelNormalizePass.
Jul 30 2020, 7:39 PM · Restricted Project

Jun 26 2020

flaub added inline comments to D82600: IR changes to add yield semantics for affine.if and affine.parallel, additional affine.parallel canonicalization.
Jun 26 2020, 5:05 PM · Restricted Project
flaub added inline comments to D82600: IR changes to add yield semantics for affine.if and affine.parallel, additional affine.parallel canonicalization.
Jun 26 2020, 5:05 PM · Restricted Project
flaub added reviewers for D82600: IR changes to add yield semantics for affine.if and affine.parallel, additional affine.parallel canonicalization: dcaballe, mravishankar.
Jun 26 2020, 2:08 PM · Restricted Project
flaub accepted D82600: IR changes to add yield semantics for affine.if and affine.parallel, additional affine.parallel canonicalization.
Jun 26 2020, 1:08 PM · Restricted Project

Jun 25 2020

flaub added inline comments to D82600: IR changes to add yield semantics for affine.if and affine.parallel, additional affine.parallel canonicalization.
Jun 25 2020, 6:02 PM · Restricted Project

Apr 24 2020

flaub accepted D78855: [MLIR]][DOC] fix dimension validity constraint in affine dialect doc.
Apr 24 2020, 10:09 PM · Restricted Project

Apr 22 2020

flaub accepted D78647: [MLIR] Lower GenericAtomicRMWOp to llvm.cmpxchg..
Apr 22 2020, 11:58 AM · Restricted Project
flaub added a comment to D78647: [MLIR] Lower GenericAtomicRMWOp to llvm.cmpxchg..

OK, I could see always lowering enums that would lower to cmpxchg via the generic op. But at the moment we have a higher layer that uses the enumerated version for min/max, so I'd prefer to don't remove that. It's much easier to pattern match (even declaratively) the enumerated version than the one with a region.

Apr 22 2020, 11:26 AM · Restricted Project

Apr 20 2020

flaub accepted D78352: [MLIR] Add GenericAtomicRMWOp..

LGTM

Apr 20 2020, 11:21 AM · Restricted Project

Apr 15 2020

flaub added inline comments to D78226: [MLIR] Improve support for 0-dimensional Affine Maps..
Apr 15 2020, 12:03 PM · Restricted Project

Apr 14 2020

flaub accepted D78146: [MLIR] Bug fix for simplifyAffineMap.

LGTM

Apr 14 2020, 1:30 PM · Restricted Project

Mar 13 2020

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

LGTM

Mar 13 2020, 11:19 AM · Restricted Project

Mar 11 2020

flaub added a comment to D76031: [mlir][CRunnerUtils] Enable compilation with C++11 toolchain on microcontroller platforms..

I don't have MSVC at this minute but this seems OK. Is it that the older compiler doesn't understand constexpr functions? I guess there's no fundamental reason the accessor operators need to be constexpr.

Mar 11 2020, 4:06 PM · Restricted Project

Mar 4 2020

flaub committed rGcdc5cba7211a: [MLIR][Affine][NFC] Expose expandAffineMap (authored by flaub).
[MLIR][Affine][NFC] Expose expandAffineMap
Mar 4 2020, 2:41 PM
flaub closed D75589: [MLIR][Affine][NFC] Expose expandAffineMap.
Mar 4 2020, 2:40 PM · Restricted Project
flaub updated the diff for D75589: [MLIR][Affine][NFC] Expose expandAffineMap.

arc?

Mar 4 2020, 2:06 PM · Restricted Project
flaub committed rGc4119a5b900a: [MLIR][Affine][NFC] Remove obsolete and ambiguous definitions (authored by flaub).
[MLIR][Affine][NFC] Remove obsolete and ambiguous definitions
Mar 4 2020, 1:32 PM
flaub closed D75586: [MLIR][Affine][NFC] Remove obsolete and ambiguous definitions.
Mar 4 2020, 1:32 PM · Restricted Project
flaub updated the diff for D75589: [MLIR][Affine][NFC] Expose expandAffineMap.

Fix formatting

Mar 4 2020, 1:30 PM · Restricted Project
flaub updated the diff for D75586: [MLIR][Affine][NFC] Remove obsolete and ambiguous definitions.

Use git-clang-format

Mar 4 2020, 1:29 PM · Restricted Project
flaub abandoned D75641: git-clang-format.
Mar 4 2020, 1:29 PM · Restricted Project
flaub created D75641: git-clang-format.
Mar 4 2020, 1:29 PM · Restricted Project
flaub added a comment to D75586: [MLIR][Affine][NFC] Remove obsolete and ambiguous definitions.

I tried to use this from outside of MLIR core and it ended up not possible to even call the getFlattendAffineExprs that doesn't take a 3rd argument due to that being ambiguous with the one defined in AffineStructures.

Mar 4 2020, 12:21 PM · Restricted Project

Mar 3 2020

flaub created D75589: [MLIR][Affine][NFC] Expose expandAffineMap.
Mar 3 2020, 5:52 PM · Restricted Project
flaub committed rG216ef5b9abb8: [MLIR][NFC] Fix for VS2017 bug. (authored by flaub).
[MLIR][NFC] Fix for VS2017 bug.
Mar 3 2020, 5:18 PM
flaub closed D75584: [MLIR] Fix for MSVC.
Mar 3 2020, 5:17 PM · Restricted Project
flaub created D75586: [MLIR][Affine][NFC] Remove obsolete and ambiguous definitions.
Mar 3 2020, 5:16 PM · Restricted Project
flaub retitled D75584: [MLIR] Fix for MSVC from Fix for MSVC to [MLIR] Fix for MSVC.
Mar 3 2020, 4:42 PM · Restricted Project
flaub added a comment to D75459: [mlir] Add padding to 1-D Vector in CRunnerUtils.h.

See https://reviews.llvm.org/D75584 for a potential fix.

Mar 3 2020, 4:40 PM · Restricted Project
flaub created D75584: [MLIR] Fix for MSVC.
Mar 3 2020, 4:40 PM · Restricted Project
flaub added inline comments to D75459: [mlir] Add padding to 1-D Vector in CRunnerUtils.h.
Mar 3 2020, 4:39 PM · Restricted Project
flaub added inline comments to D75459: [mlir] Add padding to 1-D Vector in CRunnerUtils.h.
Mar 3 2020, 4:39 PM · Restricted Project

Feb 24 2020

flaub committed rGfe210a1ff2e9: [MLIR] Add std.atomic_rmw op (authored by flaub).
[MLIR] Add std.atomic_rmw op
Feb 24 2020, 4:55 PM
flaub closed D74401: [MLIR] Add std.atomic_rmw op.
Feb 24 2020, 4:54 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Merge branch 'master' into arcpatch-D74401
  • Merge branch 'master' into arcpatch-D74401
  • Update with master
Feb 24 2020, 4:36 PM · Restricted Project
flaub added a comment to D74401: [MLIR] Add std.atomic_rmw op.

@rriddle Thanks for the review, does this look better now?

Feb 24 2020, 4:36 PM · Restricted Project

Feb 20 2020

flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Fix example
Feb 20 2020, 1:06 AM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Review feedback
Feb 20 2020, 12:57 AM · Restricted Project
flaub added inline comments to D74401: [MLIR] Add std.atomic_rmw op.
Feb 20 2020, 12:21 AM · Restricted Project

Feb 19 2020

flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Fix example
Feb 19 2020, 11:19 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Comments
Feb 19 2020, 11:19 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.

Simplified design based on RFC feedback

Feb 19 2020, 11:10 PM · Restricted Project

Feb 13 2020

flaub added a comment to D74525: [mlir][DeclarativeParser] Add support for formatting enum attributes in the string form..

This is great! Would it be possible to parse keywords into enum attrs as well?

Feb 13 2020, 4:22 PM · Restricted Project

Feb 12 2020

flaub added a comment to D73826: [mlir][Linalg] Provide a Tablegen backend to specify named Linalg ops declaratively.

Just throwing an idea out there, but what if instead of 'named' ops (which seems to be the source of the need to extend TableGen), we could use something like the 'contraction' in PlaidML's tile dialect. It's not as flexible as arbitrary generic ops, but it seems to cover most of the ML workloads we've thrown at it. Here's what the ContractionOp looks like: https://github.com/plaidml/plaidml/blob/plaidml-v1/pmlc/dialect/tile/ir/ops.td#L232 (you can ignore the other ops in there, they are meant for the construction phase, for turning values into the type system via attributes/AffineMaps, etc). I'm thinking we really only need 2 kinds of ops: Contraction op and a generic Elementwise op. Bonus: both of these forms are auto-differentiable. We could add the other properties of structured ops.

Feb 12 2020, 7:58 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Remove stdx.
  • Review updates
  • Remove iv
  • Review updates
  • Drop trivial braces
  • Trivial braces
  • Address feedback
Feb 12 2020, 6:37 PM · Restricted Project
flaub added a comment to D74401: [MLIR] Add std.atomic_rmw op.

@rriddle Are there anymore blockers that need to be addressed?

Feb 12 2020, 6:37 PM · Restricted Project
flaub committed rGfdc7a16a8275: [MLIR][Affine] Add affine.parallel op (authored by flaub).
[MLIR][Affine] Add affine.parallel op
Feb 12 2020, 6:02 PM
flaub closed D74288: [MLIR][Affine] Add affine.parallel op.
Feb 12 2020, 6:02 PM · Restricted Project
flaub updated the diff for D74288: [MLIR][Affine] Add affine.parallel op.

Rebase

Feb 12 2020, 6:02 PM · Restricted Project
flaub updated the diff for D74288: [MLIR][Affine] Add affine.parallel op.
  • stream
Feb 12 2020, 6:02 PM · Restricted Project
flaub added inline comments to D74288: [MLIR][Affine] Add affine.parallel op.
Feb 12 2020, 6:02 PM · Restricted Project
flaub added inline comments to D74288: [MLIR][Affine] Add affine.parallel op.
Feb 12 2020, 5:52 PM · Restricted Project
flaub updated the diff for D74288: [MLIR][Affine] Add affine.parallel op.
  • - Review updates
Feb 12 2020, 5:52 PM · Restricted Project
flaub added a comment to D74288: [MLIR][Affine] Add affine.parallel op.

@rriddle Are you OK with the latest changes? Thanks :)

Feb 12 2020, 10:15 AM · Restricted Project

Feb 11 2020

flaub added inline comments to D74401: [MLIR] Add std.atomic_rmw op.
Feb 11 2020, 5:14 PM · Restricted Project
flaub added inline comments to D74401: [MLIR] Add std.atomic_rmw op.
Feb 11 2020, 4:11 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Drop trivial braces
  • Trivial braces
Feb 11 2020, 4:02 PM · Restricted Project
flaub added inline comments to D74401: [MLIR] Add std.atomic_rmw op.
Feb 11 2020, 3:44 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Review updates
Feb 11 2020, 3:44 PM · Restricted Project
flaub added inline comments to D74288: [MLIR][Affine] Add affine.parallel op.
Feb 11 2020, 2:59 PM · Restricted Project
flaub added a comment to D74401: [MLIR] Add std.atomic_rmw op.

Review addressed in latest push.

Feb 11 2020, 1:01 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Remove iv
Feb 11 2020, 12:51 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Remove iv
Feb 11 2020, 12:51 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Review updates
Feb 11 2020, 12:51 PM · Restricted Project
flaub added a comment to D74288: [MLIR][Affine] Add affine.parallel op.

Thanks, more changes incoming.

Feb 11 2020, 12:33 PM · Restricted Project
flaub updated the diff for D74288: [MLIR][Affine] Add affine.parallel op.
  • Review updates
Feb 11 2020, 12:24 PM · Restricted Project
flaub updated the diff for D74401: [MLIR] Add std.atomic_rmw op.
  • Remove stdx.
Feb 11 2020, 6:10 AM · Restricted Project
flaub created D74401: [MLIR] Add std.atomic_rmw op.
Feb 11 2020, 6:10 AM · Restricted Project
flaub updated the diff for D74288: [MLIR][Affine] Add affine.parallel op.
  • Improve error msg
Feb 11 2020, 4:21 AM · Restricted Project
flaub added inline comments to D74288: [MLIR][Affine] Add affine.parallel op.
Feb 11 2020, 4:12 AM · Restricted Project