Page MenuHomePhabricator
Feed Advanced Search

Wed, May 5

akuegel added a comment to D101908: Add support for complex constants to MLIR core..

I probably need to add more tests, and maybe need to improve some things. Still, I wanted to already send this out to get feedback on the general approach here.
The RFC with more discussion is here:
https://llvm.discourse.group/t/rfc-add-a-builtin-complexattr/3316/28

Wed, May 5, 7:29 AM · Restricted Project
akuegel requested review of D101908: Add support for complex constants to MLIR core..
Wed, May 5, 7:27 AM · Restricted Project

Tue, May 4

akuegel committed rG93537fabcee8: [mlir] Add lowering from math.expm1 to LLVM. (authored by akuegel).
[mlir] Add lowering from math.expm1 to LLVM.
Tue, May 4, 5:22 AM
akuegel closed D96776: [mlir] Add lowering from math.expm1 to LLVM..
Tue, May 4, 5:22 AM · Restricted Project
akuegel added a comment to D96776: [mlir] Add lowering from math.expm1 to LLVM..

Like I said, I won't block this (and I haven't blocked https://reviews.llvm.org/D98662 either), but let's try and find time across teams to proceed with the refactoring -- it's not a priority for me either, but maintaining an ever-growing list of patterns may become hard.

Tue, May 4, 5:21 AM · Restricted Project
akuegel added a comment to D96776: [mlir] Add lowering from math.expm1 to LLVM..

Ping.
I just added log1p's approximation (https://reviews.llvm.org/D101765). I'd like this patch to go in so that I can also add expm1's approximation.

Tue, May 4, 3:49 AM · Restricted Project
akuegel updated the diff for D96776: [mlir] Add lowering from math.expm1 to LLVM..

Rebase

Tue, May 4, 3:31 AM · Restricted Project

Wed, Apr 28

akuegel committed rG2ea7fb7b1c04: [MLIR] Add ComplexToStandard conversion pass. (authored by akuegel).
[MLIR] Add ComplexToStandard conversion pass.
Wed, Apr 28, 5:18 AM
akuegel closed D101442: [MLIR] Add ComplexToStandard conversion pass..
Wed, Apr 28, 5:18 AM · Restricted Project
akuegel added inline comments to D101442: [MLIR] Add ComplexToStandard conversion pass..
Wed, Apr 28, 5:16 AM · Restricted Project
akuegel updated the diff for D101442: [MLIR] Add ComplexToStandard conversion pass..

Address review comments.

Wed, Apr 28, 5:15 AM · Restricted Project
akuegel requested review of D101442: [MLIR] Add ComplexToStandard conversion pass..
Wed, Apr 28, 4:16 AM · Restricted Project

Mar 22 2021

akuegel added a comment to D99059: [mlir] Add an option to still use bottom-up traversal.

No test? For a revision touching the core infrastructure this is a bit light and looked a bit rushed in right now.

Mar 22 2021, 8:25 AM · Restricted Project
akuegel committed rGc691b9686b77: [mlir] Add an option to still use bottom-up traversal (authored by akuegel).
[mlir] Add an option to still use bottom-up traversal
Mar 22 2021, 1:59 AM
akuegel closed D99059: [mlir] Add an option to still use bottom-up traversal.
Mar 22 2021, 1:59 AM · Restricted Project
akuegel requested review of D99059: [mlir] Add an option to still use bottom-up traversal.
Mar 22 2021, 1:49 AM · Restricted Project

Mar 17 2021

akuegel committed rG4a8c01a02b46: Move BaseOpWithOffsetSizesAndStrides to OpBase.td (authored by akuegel).
Move BaseOpWithOffsetSizesAndStrides to OpBase.td
Mar 17 2021, 5:54 AM
akuegel closed D98777: Move BaseOpWithOffsetSizesAndStrides to OpBase.td.
Mar 17 2021, 5:54 AM · Restricted Project
akuegel requested review of D98777: Move BaseOpWithOffsetSizesAndStrides to OpBase.td.
Mar 17 2021, 5:43 AM · Restricted Project

Mar 16 2021

akuegel committed rG2995e161b05f: [mlir]: Add canonicalization for dim of 1D alloc of size rank. (authored by akuegel).
[mlir]: Add canonicalization for dim of 1D alloc of size rank.
Mar 16 2021, 2:40 AM
akuegel closed D97542: [mlir]: Add folding for DimOp with AllocaOp operand.
Mar 16 2021, 2:39 AM · Restricted Project
akuegel added a comment to D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

Does it compile when you use AllocLikeOp? Yes, it only appears in the td file. A class doesn't have to inherit from another for cast/dyn_cast etc. to work. (For eg. Operation * cast to any derived OpTy.)

Mar 16 2021, 2:37 AM · Restricted Project
akuegel updated the diff for D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

Fix test comment.

Mar 16 2021, 2:30 AM · Restricted Project
akuegel updated the diff for D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

Rebase.

Mar 16 2021, 2:29 AM · Restricted Project

Mar 5 2021

akuegel added a comment to D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

Please take another look :)
I have removed the controversial part of the change.

Mar 5 2021, 1:54 AM · Restricted Project

Mar 3 2021

akuegel retitled D97542: [mlir]: Add folding for DimOp with AllocaOp operand from [mlir]: Add folding for dim of operand with rank 1 to [mlir]: Add folding for DimOp with AllocaOp operand.
Mar 3 2021, 12:50 AM · Restricted Project
akuegel added inline comments to D97542: [mlir]: Add folding for DimOp with AllocaOp operand.
Mar 3 2021, 12:48 AM · Restricted Project
akuegel updated the diff for D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

Address review comments.

Mar 3 2021, 12:47 AM · Restricted Project

Mar 1 2021

akuegel added a comment to D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

I think I did the requested changes. Can you please take another look? :)

Mar 1 2021, 11:28 PM · Restricted Project
akuegel retitled D97542: [mlir]: Add folding for DimOp with AllocaOp operand from [mlir]: Add canonicalization for dim of 1D AllocLikeOp to [mlir]: Add folding for dim of operand with rank 1.
Mar 1 2021, 1:31 AM · Restricted Project
akuegel added inline comments to D97542: [mlir]: Add folding for DimOp with AllocaOp operand.
Mar 1 2021, 1:29 AM · Restricted Project
akuegel updated the diff for D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

Move canonicalization pattern to folding.

Mar 1 2021, 1:25 AM · Restricted Project
akuegel retitled D97542: [mlir]: Add folding for DimOp with AllocaOp operand from [mlir]: Add canonicalization for dim of 1D alloc of size rank. to [mlir]: Add canonicalization for dim of 1D AllocLikeOp.
Mar 1 2021, 12:06 AM · Restricted Project
akuegel added inline comments to D97542: [mlir]: Add folding for DimOp with AllocaOp operand.
Mar 1 2021, 12:05 AM · Restricted Project
akuegel updated the diff for D97542: [mlir]: Add folding for DimOp with AllocaOp operand.

Make the pattern more general.

Mar 1 2021, 12:03 AM · Restricted Project

Feb 26 2021

akuegel added a comment to D96776: [mlir] Add lowering from math.expm1 to LLVM..

Sorry for not following up on this patch. I was very busy this week to work on a new op that we need for Kernel Generator broadcast specialization, and that is blocking and thus has priority over this patch.
I hope I can do the necessary adjustments to this patch soon.

Feb 26 2021, 5:46 AM · Restricted Project
akuegel added a reviewer for D97542: [mlir]: Add folding for DimOp with AllocaOp operand: nicolasvasilache.
Feb 26 2021, 5:05 AM · Restricted Project
akuegel requested review of D97542: [mlir]: Add folding for DimOp with AllocaOp operand.
Feb 26 2021, 4:42 AM · Restricted Project

Feb 16 2021

akuegel updated the diff for D96776: [mlir] Add lowering from math.expm1 to LLVM..

Fix comment that was still copy/pasted from rsqrt.

Feb 16 2021, 5:55 AM · Restricted Project
akuegel requested review of D96776: [mlir] Add lowering from math.expm1 to LLVM..
Feb 16 2021, 5:40 AM · Restricted Project
akuegel committed rG07cc77187a2f: Lower math.expm1 to intrinsics in the GPUToNVVM and GPUToROCDL conversions. (authored by akuegel).
Lower math.expm1 to intrinsics in the GPUToNVVM and GPUToROCDL conversions.
Feb 16 2021, 1:24 AM
akuegel closed D96756: Lower math.expm1 to intrinsics in the GPUToNVVM and GPUToROCDL conversions..
Feb 16 2021, 1:24 AM · Restricted Project
akuegel added a comment to D96756: Lower math.expm1 to intrinsics in the GPUToNVVM and GPUToROCDL conversions..

Already landed as 9f581815ae4d5df7453f0b3bea845c0ba508c91e (Mehdi approved the review in https://reviews.llvm.org/D96704)

Feb 16 2021, 12:49 AM · Restricted Project
akuegel requested review of D96756: Lower math.expm1 to intrinsics in the GPUToNVVM and GPUToROCDL conversions..
Feb 16 2021, 12:29 AM · Restricted Project

Feb 15 2021

akuegel committed rG9f581815ae4d: Add Expm1 op to the math dialect. (authored by akuegel).
Add Expm1 op to the math dialect.
Feb 15 2021, 11:34 PM
akuegel closed D96704: Add Expm1 op to the math dialect..
Feb 15 2021, 11:34 PM · Restricted Project
akuegel requested review of D96704: Add Expm1 op to the math dialect..
Feb 15 2021, 4:57 AM · Restricted Project

Feb 12 2021

akuegel accepted D96579: [mlir] Allow users of `PromoteBuffersToStackPass` to customize `small buffer` func..
Feb 12 2021, 1:08 AM · Restricted Project

Feb 5 2021

akuegel added a reverting change for rG3e5ce49e5371: [LV] Unconditionally branch from middle to scalar preheader if the scalar loop…: rG7fe41ac3dff2: Revert "[LV] Unconditionally branch from middle to scalar preheader if the….
Feb 5 2021, 3:53 AM
akuegel committed rG7fe41ac3dff2: Revert "[LV] Unconditionally branch from middle to scalar preheader if the… (authored by akuegel).
Revert "[LV] Unconditionally branch from middle to scalar preheader if the…
Feb 5 2021, 3:53 AM
akuegel added a reverting change for D94892: [LV] Unconditionally branch from middle to scalar preheader if the scalar loop must execute: rG7fe41ac3dff2: Revert "[LV] Unconditionally branch from middle to scalar preheader if the….
Feb 5 2021, 3:53 AM · Restricted Project

Feb 2 2021

akuegel added a reverting change for rG9a03058d6322: [CSSPGO] Factor out common part for CSSPGO inline and AFDO inline: rG48ca6da9d2f5: Revert "[CSSPGO] Factor out common part for CSSPGO inline and AFDO inline".
Feb 2 2021, 2:52 AM
akuegel committed rG48ca6da9d2f5: Revert "[CSSPGO] Factor out common part for CSSPGO inline and AFDO inline" (authored by akuegel).
Revert "[CSSPGO] Factor out common part for CSSPGO inline and AFDO inline"
Feb 2 2021, 2:52 AM
akuegel added a reverting change for rG09cd849fdef2: Fix build break from D95024: rG3a65ec4bf96b: Revert "Fix build break from D95024".
Feb 2 2021, 2:51 AM
akuegel added a reverting change for D95024: [CSSPGO] Factor out common part for CSSPGO inline and AFDO inline: rG48ca6da9d2f5: Revert "[CSSPGO] Factor out common part for CSSPGO inline and AFDO inline".
Feb 2 2021, 2:51 AM · Restricted Project
akuegel committed rG3a65ec4bf96b: Revert "Fix build break from D95024" (authored by akuegel).
Revert "Fix build break from D95024"
Feb 2 2021, 2:51 AM

Jan 10 2021

akuegel committed rGaf339f89a142: Remove redundant casts. (authored by akuegel).
Remove redundant casts.
Jan 10 2021, 11:52 PM
akuegel closed D94305: Remove redundant casts..
Jan 10 2021, 11:52 PM · Restricted Project

Jan 8 2021

akuegel requested review of D94305: Remove redundant casts..
Jan 8 2021, 6:38 AM · Restricted Project

Dec 23 2020

akuegel added a reverting change for rGd3bf0bb18952: PR24076, PR33655, C++ CWG 1558: Consider the instantiation-dependence of: rG25a02c3d1a68: Revert "PR24076, PR33655, C++ CWG 1558: Consider the instantiation-dependence….
Dec 23 2020, 3:42 AM
akuegel committed rG25a02c3d1a68: Revert "PR24076, PR33655, C++ CWG 1558: Consider the instantiation-dependence… (authored by akuegel).
Revert "PR24076, PR33655, C++ CWG 1558: Consider the instantiation-dependence…
Dec 23 2020, 3:42 AM

Dec 14 2020

akuegel accepted D93124: [MLIR] Add atan and atan2 lowerings to CUDA intrinsics.

Thanks :)

Dec 14 2020, 12:02 AM · Restricted Project

Dec 13 2020

akuegel accepted D93123: [MLIR] Add lowerings for atan and atan2 to ROCDL intrinsics.

Thanks :)

Dec 13 2020, 11:44 PM · Restricted Project

Dec 11 2020

akuegel committed rG91220705632e: [mlir] Expose target configuration for lowering to ROCDL. (authored by akuegel).
[mlir] Expose target configuration for lowering to ROCDL.
Dec 11 2020, 4:27 AM
akuegel closed D93028: [mlir] Expose target configuration for lowering to ROCDL..
Dec 11 2020, 4:27 AM · Restricted Project
akuegel committed rGada4c7a35171: Add rsqrt lowering from standard to ROCDL. (authored by akuegel).
Add rsqrt lowering from standard to ROCDL.
Dec 11 2020, 4:20 AM
akuegel closed D93011: Add rsqrt lowering from standard to ROCDL..
Dec 11 2020, 4:20 AM · Restricted Project
akuegel added a comment to D93011: Add rsqrt lowering from standard to ROCDL..

Friendly ping?

Dec 11 2020, 3:52 AM · Restricted Project
akuegel added a comment to D93028: [mlir] Expose target configuration for lowering to ROCDL..

Friendly ping?

Dec 11 2020, 3:52 AM · Restricted Project

Dec 10 2020

akuegel updated the summary of D93028: [mlir] Expose target configuration for lowering to ROCDL..
Dec 10 2020, 5:04 AM · Restricted Project
akuegel requested review of D93028: [mlir] Expose target configuration for lowering to ROCDL..
Dec 10 2020, 5:03 AM · Restricted Project
akuegel requested review of D93011: Add rsqrt lowering from standard to ROCDL..
Dec 10 2020, 1:48 AM · Restricted Project
akuegel committed rG09f717b929ae: Add sqrt lowering from standard to ROCDL (authored by akuegel).
Add sqrt lowering from standard to ROCDL
Dec 10 2020, 12:48 AM
akuegel closed D92921: Add sqrt lowering from standard to ROCDL.
Dec 10 2020, 12:48 AM · Restricted Project
akuegel added inline comments to D92921: Add sqrt lowering from standard to ROCDL.
Dec 10 2020, 12:43 AM · Restricted Project

Dec 9 2020

akuegel added a comment to D92921: Add sqrt lowering from standard to ROCDL.

Ok, I just checked the recent presubmit results, and all seem to fail on Windows:
https://buildkite.com/llvm-project/premerge-checks/builds

Dec 9 2020, 6:09 AM · Restricted Project
akuegel added a comment to D92921: Add sqrt lowering from standard to ROCDL.

Frederik, it looks like there was an error on Windows during the pre-merge checks. Does that look familiar to the one you saw? Do you know if this is for real?

Dec 9 2020, 6:05 AM · Restricted Project
akuegel added a reviewer for D92921: Add sqrt lowering from standard to ROCDL: frgossen.
Dec 9 2020, 6:04 AM · Restricted Project
akuegel added a comment to D92921: Add sqrt lowering from standard to ROCDL.

I have copied the Sqrt test for the GPUToNVVM lowering and adapted it.

Dec 9 2020, 2:16 AM · Restricted Project
akuegel updated the diff for D92921: Add sqrt lowering from standard to ROCDL.

Add test.

Dec 9 2020, 2:15 AM · Restricted Project
akuegel added a comment to D92921: Add sqrt lowering from standard to ROCDL.

Working on a test now.

Dec 9 2020, 2:02 AM · Restricted Project
akuegel added inline comments to D92921: Add sqrt lowering from standard to ROCDL.
Dec 9 2020, 2:01 AM · Restricted Project
akuegel updated the diff for D92921: Add sqrt lowering from standard to ROCDL.

Fixed copy/paste bug.

Dec 9 2020, 2:00 AM · Restricted Project
akuegel requested review of D92921: Add sqrt lowering from standard to ROCDL.
Dec 9 2020, 1:47 AM · Restricted Project

Nov 11 2020

akuegel committed rGa719eef73ec4: MLIR: Remove TanhOp from ops list. It caused a build failure. (authored by akuegel).
MLIR: Remove TanhOp from ops list. It caused a build failure.
Nov 11 2020, 6:00 AM
akuegel committed rG5248047c9396: MLIR: add SinOp Lowering to __ocml_sin_f32 and __ocml_sin_f64 (authored by akuegel).
MLIR: add SinOp Lowering to __ocml_sin_f32 and __ocml_sin_f64
Nov 11 2020, 5:40 AM
akuegel closed D91252: MLIR: add SinOp Lowering to __ocml_sin_f32 and __ocml_sin_f64.
Nov 11 2020, 5:39 AM · Restricted Project
akuegel added a comment to D91252: MLIR: add SinOp Lowering to __ocml_sin_f32 and __ocml_sin_f64.

Thanks for the review comments :)

Nov 11 2020, 5:30 AM · Restricted Project
akuegel updated the diff for D91252: MLIR: add SinOp Lowering to __ocml_sin_f32 and __ocml_sin_f64.

Address review comments.

Nov 11 2020, 5:29 AM · Restricted Project
akuegel added a reviewer for D91252: MLIR: add SinOp Lowering to __ocml_sin_f32 and __ocml_sin_f64: ftynse.
Nov 11 2020, 5:22 AM · Restricted Project
akuegel requested review of D91252: MLIR: add SinOp Lowering to __ocml_sin_f32 and __ocml_sin_f64.
Nov 11 2020, 5:18 AM · Restricted Project

Oct 15 2020

akuegel committed rGead2aa7098cf: Fix unused variable warning when compiling with asserts disabled. (authored by akuegel).
Fix unused variable warning when compiling with asserts disabled.
Oct 15 2020, 3:56 AM
akuegel closed D89454: Fix unused variable warning when compiling with asserts disabled..
Oct 15 2020, 3:56 AM · Restricted Project
akuegel added a reviewer for D89454: Fix unused variable warning when compiling with asserts disabled.: bkramer.
Oct 15 2020, 3:52 AM · Restricted Project
akuegel requested review of D89454: Fix unused variable warning when compiling with asserts disabled..
Oct 15 2020, 3:50 AM · Restricted Project

Jul 17 2020

akuegel committed rGde0c6bd56b41: Add -o /dev/null to make it explicit that we don't care about the compiler… (authored by akuegel).
Add -o /dev/null to make it explicit that we don't care about the compiler…
Jul 17 2020, 5:23 AM

Jul 16 2020

akuegel committed rG1067d3e176ea: Revert "[NFCI] createCFGSimplificationPass(): migrate to also take… (authored by akuegel).
Revert "[NFCI] createCFGSimplificationPass(): migrate to also take…
Jul 16 2020, 1:58 AM
akuegel added a reverting change for rGb2018198c32a: [NFCI] createCFGSimplificationPass(): migrate to also take SimplifyCFGOptions: rG1067d3e176ea: Revert "[NFCI] createCFGSimplificationPass(): migrate to also take….
Jul 16 2020, 1:58 AM
akuegel committed rG5831e8619096: Revert "[NFC] SimplifyCFGOptions: drop multi-parameter ctor, use default member… (authored by akuegel).
Revert "[NFC] SimplifyCFGOptions: drop multi-parameter ctor, use default member…
Jul 16 2020, 1:51 AM
akuegel added a reverting change for rG90c1b0442a03: [NFC] SimplifyCFGOptions: drop multi-parameter ctor, use default member-init: rG5831e8619096: Revert "[NFC] SimplifyCFGOptions: drop multi-parameter ctor, use default member….
Jul 16 2020, 1:51 AM