Page MenuHomePhabricator

flaub (Frank Laub)
User

Projects

User does not belong to any projects.

User Details

User Since
Dec 21 2019, 2:19 AM (9 w, 5 d)

Recent Activity

Mon, Feb 24

flaub committed rGfe210a1ff2e9: [MLIR] Add std.atomic_rmw op (authored by flaub).
[MLIR] Add std.atomic_rmw op
Mon, Feb 24, 4:55 PM
flaub closed D74401: [MLIR] Add std.atomic_rmw op.
Mon, Feb 24, 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
Mon, Feb 24, 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?

Mon, Feb 24, 4:36 PM · Restricted Project

Thu, Feb 20

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

Wed, Feb 19

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

Simplified implementation based on RFC feedback

Wed, Feb 19, 11:10 PM · Restricted Project

Thu, Feb 13

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?

Thu, Feb 13, 4:22 PM · Restricted Project

Wed, Feb 12

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.

Wed, Feb 12, 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
Wed, Feb 12, 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?

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

Rebase

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

Wed, Feb 12, 10:15 AM · Restricted Project

Tue, Feb 11

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

Review addressed in latest push.

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

Thanks, more changes incoming.

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

Mon, Feb 10

flaub added inline comments to D74288: [MLIR][Affine] Add affine.parallel op.
Mon, Feb 10, 2:24 PM · Restricted Project
flaub added a reviewer for D74288: [MLIR][Affine] Add affine.parallel op: jbruestle.
Mon, Feb 10, 12:00 PM · Restricted Project
flaub added a reviewer for D74288: [MLIR][Affine] Add affine.parallel op: earhart.
Mon, Feb 10, 11:43 AM · Restricted Project
flaub updated the diff for D74288: [MLIR][Affine] Add affine.parallel op.

Rebase

Mon, Feb 10, 2:31 AM · Restricted Project
flaub committed rGa248fa90a75f: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap (authored by flaub).
[MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap
Mon, Feb 10, 2:29 AM
flaub closed D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.
Mon, Feb 10, 2:28 AM · Restricted Project
flaub updated the diff for D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.

Remove inline

Mon, Feb 10, 1:53 AM · Restricted Project
flaub added inline comments to D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.
Mon, Feb 10, 12:32 AM · Restricted Project

Sat, Feb 8

flaub created D74288: [MLIR][Affine] Add affine.parallel op.
Sat, Feb 8, 12:51 PM · Restricted Project
flaub added a comment to D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.

The original diff included the AffineParallelOp, but this has been split out into its own diff based on this one: https://reviews.llvm.org/D74288

Sat, Feb 8, 12:51 PM · Restricted Project
flaub retitled D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap from [MLIR][Affine] Add affine.parallel op to [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap .
Sat, Feb 8, 12:42 PM · Restricted Project
flaub updated the diff for D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.
  • Split
Sat, Feb 8, 12:42 PM · Restricted Project
flaub updated the diff for D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.

arcanist

Sat, Feb 8, 1:54 AM · Restricted Project
flaub updated the diff for D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.
  • Fix example
Sat, Feb 8, 1:51 AM · Restricted Project
flaub created D74277: [MLIR][Affine] NFC: Move AffineValueMap and MutableAffineMap.
Sat, Feb 8, 1:33 AM · Restricted Project

Wed, Feb 5

flaub added a comment to D74056: [mlir][AVX512] Start an AVX512 dialect.

To be consistent with LLVM proper, I suggest using Target instead of Targets for the directory name.

Wed, Feb 5, 10:24 AM · Restricted Project

Mon, Feb 3

flaub added inline comments to D73912: [mlir] Turn flags in ConvertStandardToLLVM into pass flags.
Mon, Feb 3, 12:17 PM · Restricted Project

Sat, Feb 1

flaub added a comment to D73795: [mlir] Drop customization hooks from StandardToLLVM conversion.

Note: the issue I'm running into is being able to write lit tests that uses the LLVMLoweringPass along with some extra custom patterns. I was able to do that with this little bit of code:

Sat, Feb 1, 1:27 AM · Restricted Project
flaub added a comment to D73795: [mlir] Drop customization hooks from StandardToLLVM conversion.

I'm wondering how it will be possible to extend the list of patterns for non-standard ops and still make use of the existing LLVMLoweringPass? I'm currently using the patternListFiller to add my own patterns to the pass. If the LLVMLoweringPass was exposed in a header, I could at least reuse that.

Sat, Feb 1, 1:25 AM · Restricted Project

Fri, Jan 31

flaub added a comment to D71961: Free ReturnOp from being restricted to a FuncOp.

FWIW I'm +1 with using a single op for all cases of returning control to the parent op. I personally don't have a strong preference on yield vs return.

Fri, Jan 31, 8:29 PM

Thu, Jan 30

flaub added a comment to D73468: [mlir] LLVM dialect: Generate conversions between EnumAttrCase and LLVM API.

Great, once this lands I can go add the missing volatile attribute to the new atomic ops.

Thu, Jan 30, 8:23 AM · Restricted Project
flaub committed rGfcabccd3d965: [MLIR] Add the sqrt operation to mlir. (authored by llitchev).
[MLIR] Add the sqrt operation to mlir.
Thu, Jan 30, 8:14 AM
flaub closed D73571: [MLIR] Add the sqrt operation to mlir..
Thu, Jan 30, 8:13 AM · Restricted Project
flaub added a comment to D73571: [MLIR] Add the sqrt operation to mlir..

Thanks @ftynse, I'll land this shortly.

Thu, Jan 30, 7:45 AM · Restricted Project

Tue, Jan 28

flaub added inline comments to D71961: Free ReturnOp from being restricted to a FuncOp.
Tue, Jan 28, 9:21 PM
flaub retitled D73571: [MLIR] Add the sqrt operation to mlir. from Add the sqrt operation to mlir. to [MLIR] Add the sqrt operation to mlir..
Tue, Jan 28, 12:23 PM · Restricted Project

Jan 24 2020

flaub added a comment to rGa919919fcc1b: Temporary patch for pxa.

Phew, yeah we made a mistake on our end by making a PR against the base repo instead of our fork. Sorry about that.

Jan 24 2020, 5:48 PM
flaub added a comment to D73362: [mlir] EnumsGen: dissociate string form of integer enum from C++ symbol name.

This is great, I can update the recent LLVM dialect enums that I added to use this. Thanks!

Jan 24 2020, 3:50 PM · Restricted Project
flaub added a comment to rGa919919fcc1b: Temporary patch for pxa.

I have no idea how this happened, this was never meant to reach the LLVM repo.

Jan 24 2020, 3:41 PM
flaub committed rGa919919fcc1b: Temporary patch for pxa (authored by flaub).
Temporary patch for pxa
Jan 24 2020, 1:59 PM

Jan 21 2020

flaub committed rGfffea2842d28: [MLIR] LLVM Dialect: add llvm.cmpxchg and improve llvm.atomicrmw custom parser (authored by flaub).
[MLIR] LLVM Dialect: add llvm.cmpxchg and improve llvm.atomicrmw custom parser
Jan 21 2020, 1:17 AM
flaub closed D72995: [MLIR] LLVM Dialect: add llvm.cmpxchg and improve llvm.atomicrmw custom parser.
Jan 21 2020, 1:17 AM · Restricted Project
flaub updated the diff for D72995: [MLIR] LLVM Dialect: add llvm.cmpxchg and improve llvm.atomicrmw custom parser.
  • CR changes
  • Merge remote-tracking branch 'origin/master' into flaub-cmpxchg
Jan 21 2020, 12:49 AM · Restricted Project

Jan 20 2020

flaub added inline comments to D72995: [MLIR] LLVM Dialect: add llvm.cmpxchg and improve llvm.atomicrmw custom parser.
Jan 20 2020, 4:00 PM · Restricted Project

Jan 19 2020

flaub created D72995: [MLIR] LLVM Dialect: add llvm.cmpxchg and improve llvm.atomicrmw custom parser.
Jan 19 2020, 3:01 AM · Restricted Project

Jan 18 2020

flaub added a comment to D72957: [mlir] Unbreak MSVC build with implicit constructor.

I think I found a solution to this, try:

Jan 18 2020, 12:34 PM · Restricted Project
flaub added a comment to D72957: [mlir] Unbreak MSVC build with implicit constructor.

Haha, aganea beat me to it :)

Jan 18 2020, 12:34 PM · Restricted Project

Jan 17 2020

flaub committed rGee2de95507a1: [MLIR] LLVM dialect: modernize and cleanups (authored by flaub).
[MLIR] LLVM dialect: modernize and cleanups
Jan 17 2020, 5:15 PM
flaub closed D72967: [MLIR] LLVM dialect: modernize and cleanups.
Jan 17 2020, 5:15 PM · Restricted Project
flaub updated the diff for D72967: [MLIR] LLVM dialect: modernize and cleanups.
  • clang-format
Jan 17 2020, 4:46 PM · Restricted Project
flaub added a comment to D72967: [MLIR] LLVM dialect: modernize and cleanups.

I'm not sure this is a net-win because of the less effective error messages, but I thought I'd submit this and see what happens.

Jan 17 2020, 4:27 PM · Restricted Project
flaub created D72967: [MLIR] LLVM dialect: modernize and cleanups.
Jan 17 2020, 4:18 PM · Restricted Project
flaub added a comment to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.

Thanks everyone!

Jan 17 2020, 12:28 PM · Restricted Project
flaub added a comment to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.

I'm happy to do whatever the community wants to do. I was under the impression from @ftynse that we shouldn't be changing the MLIR IR, that it should match exactly whatever the LLVM IR is. It's unclear to me what the ramifications of making such a change in LLVM IR are, so I don't feel qualified at this time to make such a change. But, I could certainly take some guidance on this. (this is where my pushback is coming from, sorry if this came off the wrong way).

Jan 17 2020, 11:21 AM · Restricted Project
flaub added a comment to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.

@jfb What process do I need to follow to ensure that any changes I make to LLVM IR doesn't break the world? This would include updates to the documentation. I'm new here and I'm not sure which tests I should focus on and where the docs live.

Jan 17 2020, 11:11 AM · Restricted Project
flaub added a comment to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.

I don't think this commit should be about making changes to LLVM IR. I can do that in a followup, but this commit should be in sync with whatever LLVM IR is today. There seems to be consensus with this approach.

Jan 17 2020, 11:02 AM · Restricted Project
flaub added a comment to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.

@ftynse I believe this is ready to go, anything else you'd like?

Jan 17 2020, 10:42 AM · Restricted Project

Jan 16 2020

flaub added a comment to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.

More CR changes incoming

Jan 16 2020, 1:45 PM · Restricted Project
flaub updated the diff for D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.
  • CR changes
Jan 16 2020, 1:45 PM · Restricted Project
Herald updated subscribers of D72533: [mlir] Add a signedness semantics bit to IntegerType.
Jan 16 2020, 1:49 AM · Restricted Project

Jan 15 2020

flaub updated the diff for D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.
  • CR changes
Jan 15 2020, 4:34 PM · Restricted Project
flaub added inline comments to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.
Jan 15 2020, 4:32 PM · Restricted Project
flaub added a comment to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.

Thanks for the review :)

Jan 15 2020, 2:10 AM · Restricted Project

Jan 14 2020

flaub added inline comments to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.
Jan 14 2020, 4:43 PM · Restricted Project
flaub added inline comments to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.
Jan 14 2020, 4:43 PM · Restricted Project
flaub added inline comments to D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.
Jan 14 2020, 4:33 PM · Restricted Project
flaub created D72741: [MLIR] LLVM dialect: Add llvm.atomicrmw.
Jan 14 2020, 3:52 PM · Restricted Project

Dec 23 2019

A. Unique TensorFlower <gardener@tensorflow.org> committed rG7811ad3c2b31: Allow dialect to create friendly names for region arguments (authored by flaub).
Allow dialect to create friendly names for region arguments
Dec 23 2019, 9:27 PM