Page MenuHomePhabricator
People hanchung

hanchung (Han-Chung Wang)
User

Projects

User does not belong to any projects.

User Details

User Since
Jan 7 2020, 8:20 PM (124 w, 2 d)

Recent Activity
View All

Yesterday

hanchung committed rG5aefdafccf33: [mlir][Linalg] Relax vectorization condition to allow transposed output. (authored by hanchung).
[mlir][Linalg] Relax vectorization condition to allow transposed output.
Thu, May 26, 7:22 PM · Restricted Project, Restricted Project
hanchung closed D126454: [mlir][Linalg] Relax vectorization condition to allow transposed output..
Thu, May 26, 7:21 PM · Restricted Project, Restricted Project
hanchung updated the diff for D126454: [mlir][Linalg] Relax vectorization condition to allow transposed output..

fix test

Thu, May 26, 9:46 AM · Restricted Project, Restricted Project
hanchung added a reviewer for D126454: [mlir][Linalg] Relax vectorization condition to allow transposed output.: dcaballe.
Thu, May 26, 2:02 AM · Restricted Project, Restricted Project
hanchung added reviewers for D126454: [mlir][Linalg] Relax vectorization condition to allow transposed output.: ThomasRaoux, mravishankar, antiagainst.
Thu, May 26, 2:01 AM · Restricted Project, Restricted Project
hanchung requested review of D126454: [mlir][Linalg] Relax vectorization condition to allow transposed output..
Thu, May 26, 2:01 AM · Restricted Project, Restricted Project

Tue, May 24

hanchung accepted D126256: [mlir][spirv] Fix capability check for 64-bit element types.
Tue, May 24, 9:33 PM · Restricted Project, Restricted Project
hanchung accepted D126320: [mlir][scf] Retain existing attributes in scf.for transforms.
Tue, May 24, 9:03 PM · Restricted Project, Restricted Project

Mon, May 9

hanchung committed rGad7c49bef774: [mlir][linalg] Fix padding size calculation for Conv2d ops. (authored by pzread).
[mlir][linalg] Fix padding size calculation for Conv2d ops.
Mon, May 9, 8:46 AM · Restricted Project, Restricted Project
hanchung closed D124821: [mlir][linalg] Fix padding size calculation for Conv2d ops..
Mon, May 9, 8:46 AM · Restricted Project, Restricted Project

Wed, May 4

hanchung accepted D124821: [mlir][linalg] Fix padding size calculation for Conv2d ops..

Nice fix!

Wed, May 4, 1:08 AM · Restricted Project, Restricted Project

Tue, May 3

hanchung accepted D124849: [mlir][bufferize] Allow in-place bufferization for writes to init_tensors in loops.
`linalg.init_tensor` is an operation that materializes a tensor of
a given shape. The shape could be dynamic or static.
Tue, May 3, 11:30 PM · Restricted Project, Restricted Project
hanchung added a reviewer for D124849: [mlir][bufferize] Allow in-place bufferization for writes to init_tensors in loops: mravishankar.
Tue, May 3, 11:26 PM · Restricted Project, Restricted Project
hanchung committed rG919e459f1ba3: [Linalg] Remove Optional from getStaticLoopRanges interface method. (authored by hanchung).
[Linalg] Remove Optional from getStaticLoopRanges interface method.
Tue, May 3, 5:13 AM · Restricted Project, Restricted Project
hanchung closed D124596: [Linalg] Remove Optional from getStaticLoopRanges interface method..
Tue, May 3, 5:13 AM · Restricted Project, Restricted Project
hanchung added inline comments to D124596: [Linalg] Remove Optional from getStaticLoopRanges interface method..
Tue, May 3, 1:00 AM · Restricted Project, Restricted Project
hanchung updated the diff for D124596: [Linalg] Remove Optional from getStaticLoopRanges interface method..

add asserts

Tue, May 3, 12:59 AM · Restricted Project, Restricted Project

Thu, Apr 28

hanchung updated the diff for D124596: [Linalg] Remove Optional from getStaticLoopRanges interface method..

update comments

Thu, Apr 28, 9:07 PM · Restricted Project, Restricted Project

Wed, Apr 27

hanchung added reviewers for D124596: [Linalg] Remove Optional from getStaticLoopRanges interface method.: springerm, antiagainst.
Wed, Apr 27, 11:54 PM · Restricted Project, Restricted Project
hanchung requested review of D124596: [Linalg] Remove Optional from getStaticLoopRanges interface method..
Wed, Apr 27, 11:54 PM · Restricted Project, Restricted Project
hanchung accepted D124543: [mlir][spirv] Remove layout decoration on unneeded storage classes.
Wed, Apr 27, 11:34 PM · Restricted Project, Restricted Project
hanchung accepted D124542: [mlir][spirv] Convert memref.alloca to spirv.Variable.
Wed, Apr 27, 11:33 PM · Restricted Project, Restricted Project

Apr 26 2022

hanchung accepted D124380: [mlir][spirv] Add conversion from vector.reduction.
Apr 26 2022, 11:07 PM · Restricted Project, Restricted Project
hanchung accepted D124451: [mlir][spirv] Allow converting from index type in unsigned ops.
Apr 26 2022, 11:02 PM · Restricted Project, Restricted Project

Apr 25 2022

hanchung accepted D124366: [mlir][vector] insert `alloca`s outside of loops.

Thanks!

Apr 25 2022, 1:47 AM · Restricted Project, Restricted Project

Apr 15 2022

hanchung accepted D123867: [mlir][vector] Fix bug in extractFromBroadcast folding.
Apr 15 2022, 12:03 PM · Restricted Project, Restricted Project

Apr 8 2022

hanchung accepted D123190: [mlir][vector] Swap ExtractSliceOp(TransferWriteOp)..
Apr 8 2022, 11:58 AM · Restricted Project, Restricted Project

Apr 5 2022

hanchung added a reverting change for rG64f659bee67b: [mlir] Rewrite canonicalization of collapse(expand) and expand(collapse).: rG96e9b6c9dc60: Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand….
Apr 5 2022, 3:06 PM · Restricted Project, Restricted Project
hanchung committed rG96e9b6c9dc60: Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand… (authored by hanchung).
Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand…
Apr 5 2022, 3:06 PM · Restricted Project, Restricted Project
hanchung added a reverting change for D122666: [mlir] Rewrite canonicalization of collapse(expand) and expand(collapse) .: rG96e9b6c9dc60: Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand….
Apr 5 2022, 3:06 PM · Restricted Project, Restricted Project
hanchung closed D123161: Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand(collapse).".
Apr 5 2022, 3:06 PM · Restricted Project, Restricted Project
hanchung added a reverting change for rG64f659bee67b: [mlir] Rewrite canonicalization of collapse(expand) and expand(collapse).: D123161: Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand(collapse).".
Apr 5 2022, 3:04 PM · Restricted Project, Restricted Project
hanchung added a reverting change for D122666: [mlir] Rewrite canonicalization of collapse(expand) and expand(collapse) .: D123161: Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand(collapse).".
Apr 5 2022, 3:04 PM · Restricted Project, Restricted Project
hanchung requested review of D123161: Revert "[mlir] Rewrite canonicalization of collapse(expand) and expand(collapse).".
Apr 5 2022, 3:04 PM · Restricted Project, Restricted Project
hanchung added a comment to D122666: [mlir] Rewrite canonicalization of collapse(expand) and expand(collapse) ..

I think there are bugs in the commit. The verifier failed with the commit. To repro:

Apr 5 2022, 2:48 PM · Restricted Project, Restricted Project

Apr 4 2022

hanchung accepted D123035: [mlir][spirv] Check nullptr before usage to fix crash.
Apr 4 2022, 12:13 PM · Restricted Project, Restricted Project

Mar 31 2022

hanchung accepted D122722: [mlir][tensor] Add pattern to fold ExtractSliceOp, PadOp chains..

Overall looks good to me, just few nits. :)

Mar 31 2022, 3:27 PM · Restricted Project, Restricted Project
hanchung added inline comments to D122847: [mlir][linalg] Move linalg.fill folding into linalg.generic pattern to elementwise op fusion.
Mar 31 2022, 2:53 PM · Restricted Project, Restricted Project
hanchung added inline comments to D122847: [mlir][linalg] Move linalg.fill folding into linalg.generic pattern to elementwise op fusion.
Mar 31 2022, 2:53 PM · Restricted Project, Restricted Project
hanchung added inline comments to D122847: [mlir][linalg] Move linalg.fill folding into linalg.generic pattern to elementwise op fusion.
Mar 31 2022, 12:59 PM · Restricted Project, Restricted Project

Mar 29 2022

hanchung abandoned D119211: Enhance pad transform to handle scf.for producers..
Mar 29 2022, 5:24 PM · Restricted Project, Restricted Project

Mar 24 2022

hanchung accepted D122414: [mlir][spirv] Add Apple into the vendor list.
Mar 24 2022, 10:46 AM · Restricted Project, Restricted Project

Mar 23 2022

hanchung accepted D120981: [mlir][linalg] Do not fuse shape-only producers..
Mar 23 2022, 8:02 PM · Restricted Project, Restricted Project
hanchung abandoned D119390: [mlir][linalg] Propagate static padding informations through Linalg ops..
Mar 23 2022, 8:01 PM · Restricted Project, Restricted Project
hanchung accepted D122116: [mlir][linalg] Support padding LinalgOps in use-def chain..
Mar 23 2022, 8:01 PM · Restricted Project, Restricted Project

Mar 14 2022

hanchung accepted D121648: [mlir][Linalg] Fix error in `LinalgOp` -> `tensor::CastOp` folding..
Mar 14 2022, 5:39 PM · Restricted Project, Restricted Project

Mar 8 2022

hanchung accepted D121247: [mlir][spirv] Define spv.VectorTimesScalar op.
Mar 8 2022, 12:49 PM · Restricted Project, Restricted Project
hanchung accepted D121238: [mlir][spirv] Fix spv.GLSL.{S|U}Clamp type checking.
Mar 8 2022, 12:48 PM · Restricted Project, Restricted Project
hanchung accepted D121227: [mlir][spirv] Handle another form of folding comparsion into clamp.
Mar 8 2022, 12:46 PM · Restricted Project, Restricted Project

Mar 7 2022

hanchung committed rG1538bd518cd2: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose… (authored by hanchung).
[mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose…
Mar 7 2022, 12:52 PM · Restricted Project
hanchung closed D120760: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose ops..
Mar 7 2022, 12:52 PM · Restricted Project, Restricted Project
hanchung added inline comments to D120760: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose ops..
Mar 7 2022, 11:57 AM · Restricted Project, Restricted Project
hanchung updated the diff for D120760: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose ops..

Update comments

Mar 7 2022, 11:57 AM · Restricted Project, Restricted Project
hanchung updated the diff for D120760: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose ops..

Add more comments and remove template types

Mar 7 2022, 10:44 AM · Restricted Project, Restricted Project

Mar 3 2022

hanchung added inline comments to D120760: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose ops..
Mar 3 2022, 4:17 PM · Restricted Project, Restricted Project
hanchung updated the diff for D120760: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose ops..

Address comments

Mar 3 2022, 4:17 PM · Restricted Project, Restricted Project

Mar 1 2022

hanchung requested review of D120760: [mlir][Vector] Add patterns to reorder elementwise ops and broadcast/transpose ops..
Mar 1 2022, 12:04 PM · Restricted Project, Restricted Project

Feb 25 2022

hanchung committed rG748bf4bb289b: [mlir][Linalg] Add support for tileFuseAndDistribute on tensors. (authored by hanchung).
[mlir][Linalg] Add support for tileFuseAndDistribute on tensors.
Feb 25 2022, 11:52 AM
hanchung closed D120441: [mlir][Linalg] Add support for tileFuseAndDistribute on tensors..
Feb 25 2022, 11:51 AM · Restricted Project
hanchung updated the diff for D120441: [mlir][Linalg] Add support for tileFuseAndDistribute on tensors..

rebase

Feb 25 2022, 10:52 AM · Restricted Project

Feb 24 2022

hanchung updated the diff for D120441: [mlir][Linalg] Add support for tileFuseAndDistribute on tensors..

Address comments

Feb 24 2022, 4:14 PM · Restricted Project

Feb 23 2022

hanchung added inline comments to D120441: [mlir][Linalg] Add support for tileFuseAndDistribute on tensors..
Feb 23 2022, 4:08 PM · Restricted Project
hanchung requested review of D120441: [mlir][Linalg] Add support for tileFuseAndDistribute on tensors..
Feb 23 2022, 4:06 PM · Restricted Project

Feb 14 2022

hanchung updated the diff for D119390: [mlir][linalg] Propagate static padding informations through Linalg ops..

address comments

Feb 14 2022, 9:09 PM · Restricted Project, Restricted Project

Feb 9 2022

hanchung planned changes to D119211: Enhance pad transform to handle scf.for producers..

We might want https://reviews.llvm.org/D119390 instead because scf.for iter types might change during runtime.

Feb 9 2022, 5:43 PM · Restricted Project, Restricted Project
hanchung added a reviewer for D119390: [mlir][linalg] Propagate static padding informations through Linalg ops.: gysit.
Feb 9 2022, 5:42 PM · Restricted Project, Restricted Project
hanchung published D119390: [mlir][linalg] Propagate static padding informations through Linalg ops. for review.
Feb 9 2022, 5:41 PM · Restricted Project, Restricted Project

Feb 8 2022

hanchung updated the diff for D119211: Enhance pad transform to handle scf.for producers..

Address comments

Feb 8 2022, 2:37 PM · Restricted Project, Restricted Project

Feb 7 2022

hanchung published D119211: Enhance pad transform to handle scf.for producers. for review.
Feb 7 2022, 8:22 PM · Restricted Project, Restricted Project

Jan 24 2022

hanchung accepted D117908: [mlir][Linalg] Fix incorrect fusion with reshape ops by linearization..
Jan 24 2022, 5:25 PM · Restricted Project

Dec 16 2021

hanchung accepted D115196: [mlir][tosa] Support dynamic batch dimension for ops where the batch dim is explicit.
Dec 16 2021, 5:44 PM · Restricted Project

Dec 15 2021

hanchung committed rG501674dc3b14: [mlir][Vector] Further fix to avoid infinite loop in… (authored by hanchung).
[mlir][Vector] Further fix to avoid infinite loop in…
Dec 15 2021, 1:55 PM
hanchung closed D115820: [mlir][Vector] Further fix to avoid infinite loop in InnerOuterDimReductionConversion.
Dec 15 2021, 1:55 PM · Restricted Project
hanchung retitled D115820: [mlir][Vector] Further fix to avoid infinite loop in InnerOuterDimReductionConversion from [mlir][Vector] Further fix to Avoid infinite loop in InnerOuterDimReductionConversion to [mlir][Vector] Further fix to avoid infinite loop in InnerOuterDimReductionConversion.
Dec 15 2021, 1:02 PM · Restricted Project
hanchung requested review of D115820: [mlir][Vector] Further fix to avoid infinite loop in InnerOuterDimReductionConversion.
Dec 15 2021, 12:03 PM · Restricted Project

Dec 8 2021

hanchung accepted D115321: [mlir][MemRef] Add folder for no-op `memref.subview`..
Dec 8 2021, 10:27 AM · Restricted Project

Dec 7 2021

hanchung requested changes to D115196: [mlir][tosa] Support dynamic batch dimension for ops where the batch dim is explicit.
Dec 7 2021, 10:49 AM · Restricted Project

Nov 3 2021

hanchung accepted D110262: [mlir][linalg] Add support for transitive fusion..
Nov 3 2021, 10:08 AM · Restricted Project

Nov 1 2021

hanchung added inline comments to D110262: [mlir][linalg] Add support for transitive fusion..
Nov 1 2021, 5:16 PM · Restricted Project
hanchung added a reviewer for D110262: [mlir][linalg] Add support for transitive fusion.: hanchung.
Nov 1 2021, 3:25 PM · Restricted Project
hanchung added a comment to D110262: [mlir][linalg] Add support for transitive fusion..

This is the PR I'm looking for, thanks!

Nov 1 2021, 3:24 PM · Restricted Project

Oct 26 2021

hanchung accepted D112560: [mlir][python] Segment MLIR Python test dialect to avoid testonly dependency..

Thanks for the fix!

Oct 26 2021, 10:20 AM · Restricted Project

Jul 30 2021

hanchung accepted D107122: [mlir][spirv] Add support for i8 serialization.
Jul 30 2021, 1:46 PM · Restricted Project
hanchung accepted D107119: [mlir][spirv] Fix loading bool with proper storage capabilities.
Jul 30 2021, 1:45 PM · Restricted Project
hanchung accepted D107114: [mlir][spirv] Fix storing bool with proper storage capabilities.
Jul 30 2021, 1:42 PM · Restricted Project

Jul 29 2021

hanchung accepted D107094: [mlir] NFC: split MemRef to SPIR-V conversion into their own files.
Jul 29 2021, 12:05 PM · Restricted Project
hanchung accepted D107093: [mlir] NFC: split Math to SPIR-V conversion into their own files.

Nice cleanup!

Jul 29 2021, 12:02 PM · Restricted Project
hanchung accepted D107092: [mlir] Fix CMake option for enabling SPIR-V CPU runner.
Jul 29 2021, 12:01 PM · Restricted Project

Jul 22 2021

hanchung added inline comments to D105940: [mlir][tosa] Tosa shape propagation for tosa.cond_if.
Jul 22 2021, 3:13 PM · Restricted Project
hanchung accepted D106407: [mlir][tosa] Quantized Conv2DOp lowering to linalg added..

LGTM if adding a roundtrip test to mlir/test/Dialect/Linalg/named-ops.mlir.

Jul 22 2021, 3:02 PM · Restricted Project

Jul 21 2021

hanchung requested changes to D106407: [mlir][tosa] Quantized Conv2DOp lowering to linalg added..

We should add a roundtrip test when adding a new op. You can add it to mlir/test/Dialect/Linalg/named-ops.mlir

Jul 21 2021, 12:37 PM · Restricted Project

Jul 20 2021

hanchung committed rG3de06260f746: [mlir][Linalg] Update signatures of the callback functions. (authored by hanchung).
[mlir][Linalg] Update signatures of the callback functions.
Jul 20 2021, 5:04 PM
hanchung closed D106322: [mlir][Linalg] Update signatures of the callback functions..
Jul 20 2021, 5:03 PM · Restricted Project
hanchung added inline comments to D106322: [mlir][Linalg] Update signatures of the callback functions..
Jul 20 2021, 4:31 PM · Restricted Project
hanchung updated the diff for D106322: [mlir][Linalg] Update signatures of the callback functions..

adjust comments a bit

Jul 20 2021, 4:29 PM · Restricted Project

Jul 19 2021

hanchung requested review of D106322: [mlir][Linalg] Update signatures of the callback functions..
Jul 19 2021, 4:15 PM · Restricted Project
hanchung committed rG9c49195330d7: [mlir][Linalg] Migrate 2D pooling ops from tc definition to yaml definition. (authored by hanchung).
[mlir][Linalg] Migrate 2D pooling ops from tc definition to yaml definition.
Jul 19 2021, 9:24 AM
hanchung closed D106181: [mlir][Linalg] Migrate 2D pooling ops from tc definition to yaml definition..
Jul 19 2021, 9:24 AM · Restricted Project

Jul 16 2021

hanchung updated the diff for D106181: [mlir][Linalg] Migrate 2D pooling ops from tc definition to yaml definition..

update test

Jul 16 2021, 5:10 PM · Restricted Project
hanchung accepted D106132: [mlir][Linalg] NFC: Rename FusionOfTensors pass to FusionOfElementwiseOps pass..
Jul 16 2021, 12:55 PM · Restricted Project