Page MenuHomePhabricator

jlebar (Justin Lebar)
User

Projects

User does not belong to any projects.

User Details

User Since
Dec 8 2015, 10:33 AM (342 w, 2 d)

Google developer working on CUDA for Clang. On IRC as jlebar.

Recent Activity

Mar 9 2022

jlebar accepted D121281: [ADT] Make BitmaskEnum operations constant expressions.
Mar 9 2022, 7:28 AM · Restricted Project, Restricted Project

Mar 8 2022

jlebar accepted D121259: [clang] Fix CodeGenAction for LLVM IR MemBuffers.

Congrats on your first patch!

Mar 8 2022, 3:48 PM · Restricted Project, Restricted Project

Feb 24 2022

jlebar committed rGc2f501f39589: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments (authored by shangwuyao).
[CUDA][SPIRV] Assign global address space to CUDA kernel arguments
Feb 24 2022, 8:52 PM
jlebar closed D120366: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.
Feb 24 2022, 8:52 PM · Restricted Project
jlebar accepted D120366: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.

I'll land this for you.

Feb 24 2022, 8:49 PM · Restricted Project
jlebar added a comment to D120366: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.
  • What's different in this patch vs the previous one?
  • *Disabled a hip test on Windows that's breaking on head.* Can you clarify: Is this test broken at HEAD, or does it break with your patch?
Feb 24 2022, 3:09 PM · Restricted Project

Feb 17 2022

jlebar committed rG9de4fc0f2d3b: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments (authored by shangwuyao).
[CUDA][SPIRV] Assign global address space to CUDA kernel arguments
Feb 17 2022, 9:39 AM
jlebar added a comment to D119207: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.
commit 9de4fc0f2d3b60542956f7e5254951d049edeb1f (HEAD -> main, origin/main, origin/HEAD)
Author: Shangwu Yao <shangwuyao@waymo.com>
Date:   Thu Feb 17 09:38:06 2022 -0800
Feb 17 2022, 9:39 AM · Restricted Project
jlebar closed D119207: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.
Feb 17 2022, 9:39 AM · Restricted Project
jlebar added a comment to D119207: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.

Thanks for the review, if it looks good, can we get this to land now? Otherwise more comments are welcome!

Feb 17 2022, 9:32 AM · Restricted Project

Feb 8 2022

jlebar added inline comments to D119207: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.
Feb 8 2022, 12:09 AM · Restricted Project

Feb 7 2022

jlebar added a comment to D119207: [CUDA][SPIRV] Assign global address space to CUDA kernel arguments.

[CUDA][SPIRV] Convert CUDA kernels to SPIR-V kernels

Feb 7 2022, 11:21 PM · Restricted Project

Jan 28 2022

jlebar added a comment to D117137: [Driver] Add CUDA support for --offload param.

Pushed for Daniele:

Jan 28 2022, 2:51 PM · Restricted Project
jlebar committed rG6eb826567af0: [Driver] Add CUDA support for --offload param (authored by dcastagna).
[Driver] Add CUDA support for --offload param
Jan 28 2022, 2:51 PM
jlebar closed D117137: [Driver] Add CUDA support for --offload param.
Jan 28 2022, 2:51 PM · Restricted Project

Jan 12 2022

jlebar added a comment to D117137: [Driver] Add CUDA support for --offload param.

I defer to Art.

Jan 12 2022, 4:25 PM · Restricted Project
jlebar removed a reviewer for D117137: [Driver] Add CUDA support for --offload param: jlebar.
Jan 12 2022, 4:25 PM · Restricted Project

Dec 2 2021

jlebar accepted D114407: [CUDA][SPIRV] Use OpenCLKernel CC for CUDA -> SPIRV.
Dec 2 2021, 11:26 AM · Restricted Project
jlebar updated subscribers of D114407: [CUDA][SPIRV] Use OpenCLKernel CC for CUDA -> SPIRV.
Dec 2 2021, 11:26 AM · Restricted Project

Nov 24 2021

jlebar added a comment to D114454: [analyzer]Skip unstable CSA tests failing on several platforms.

Let me summon them.

Nov 24 2021, 10:05 AM · Restricted Project

Oct 25 2021

jlebar added a comment to D112466: [NVPTX] Drop memory references of LDG/LDU.

It appears that mayLoad was omitted for a reason, but I cannot figure out why. If a load is done from a read-only memory, it should still be treated as a load, right?

Oct 25 2021, 9:59 AM · Restricted Project, Restricted Project

Oct 21 2021

jlebar accepted D112265: Remove unused parallel-libs project.
Oct 21 2021, 2:04 PM · Restricted Project

Sep 27 2021

jlebar accepted D110596: [CUDA] Move CUDA SDK include path further down the include search path..
Sep 27 2021, 4:03 PM · Restricted Project

Sep 24 2021

jlebar added a comment to D110089: [CUDA] Implement experimental support for texture lookups..

Presumably as a separate commit we should add tests to the test_suite repository to ensure that this at least still compiles with different versions of CUDA?

Sep 24 2021, 1:27 PM · Restricted Project
jlebar accepted D110089: [CUDA] Implement experimental support for texture lookups..

Okay, I give up on the phab interface. It's unreadable with all the existing
comments and lint errors.

Sep 24 2021, 1:26 PM · Restricted Project

Sep 22 2021

jlebar added a comment to D110089: [CUDA] Implement experimental support for texture lookups..

Depending on which particular operation is used, the arguments vary, too.

Sep 22 2021, 12:06 PM · Restricted Project

Sep 21 2021

jlebar added a comment to D110089: [CUDA] Implement experimental support for texture lookups..

One alternative would be to use run-time dispatch, but, given that texture lookup is a single instruction, the overhead would be substantial-to-prohibitive.

Sep 21 2021, 11:53 PM · Restricted Project

Sep 8 2021

jlebar accepted D109458: [NVPTX] Simplify and generalize constant printer..
Sep 8 2021, 12:39 PM · Restricted Project
jlebar added a comment to D109458: [NVPTX] Simplify and generalize constant printer..

Is it a pain to add a test? I know the reproducer from the bug still crashes llvm, not sure if it's hard to reproduce the i128 issue independently.

Sep 8 2021, 12:07 PM · Restricted Project

Aug 26 2021

jlebar accepted D108787: [CUDA] Pass ExecConfig through BuildCallToMemberFunction.

Thanks, Art.

Aug 26 2021, 2:18 PM · Restricted Project

Jun 30 2021

jlebar accepted D105226: [Clang] allow overriding -fbasic-block-sections.
Jun 30 2021, 2:17 PM · Restricted Project

Jun 15 2021

jlebar added a comment to D103088: [clang] pre-0388 array parm list initialization.

... let's try adding some other reviewers who've had their fingers in SemaOverload ...

Jun 15 2021, 2:08 PM · Restricted Project
jlebar removed a reviewer for D103088: [clang] pre-0388 array parm list initialization: jlebar.
Jun 15 2021, 2:07 PM · Restricted Project

Jun 9 2021

jlebar added a comment to D103996: Save/restore OuterTemplateParams in AbstractManglingParser::parseEncoding..

Thank you for the review, Richard!

Jun 9 2021, 5:50 PM · Restricted Project, Restricted Project
jlebar added a comment to D103996: Save/restore OuterTemplateParams in AbstractManglingParser::parseEncoding..

@rsmith I have no idea if the other test change is correct, and gnu cxxfilt doesn't work with that mangled name. Hope this doesn't actually break something...

Jun 9 2021, 5:11 PM · Restricted Project, Restricted Project
jlebar requested review of D103996: Save/restore OuterTemplateParams in AbstractManglingParser::parseEncoding..
Jun 9 2021, 5:10 PM · Restricted Project, Restricted Project

May 21 2021

jlebar accepted D102936: [CUDA] Work around compatibility issue with libstdc++ 11.1.0.
May 21 2021, 11:13 AM · Restricted Project

Apr 20 2021

jlebar added a comment to D100794: [HIP] Support overloaded math functions for hipRTC.

Other than the declval issue discussed above, this looks reasonable to me. I don't notice anything wrong with the standard library reimplementations here.

Apr 20 2021, 12:40 PM · Restricted Project

Apr 12 2021

jlebar committed rG7dd606889925: [clang-rename] Handle designated initializers. (authored by dcastagna).
[clang-rename] Handle designated initializers.
Apr 12 2021, 1:19 PM
jlebar closed D100310: Add field designated initializers logic in Tooling/Rename.
Apr 12 2021, 1:19 PM · Restricted Project
jlebar updated subscribers of D100310: Add field designated initializers logic in Tooling/Rename.

I guess you need me or Michael to push this. Happy to do so once you're
happy with it.

Apr 12 2021, 11:43 AM · Restricted Project
jlebar accepted D100310: Add field designated initializers logic in Tooling/Rename.

I can't argue with this.

Apr 12 2021, 8:48 AM · Restricted Project

Apr 6 2021

jlebar accepted D99979: [NVPTX] Handle bitcast and ASC(101) when trying to avoid argument copy..

Looks great to me. Thanks, Art.

Apr 6 2021, 12:12 PM · Restricted Project

Mar 12 2021

jlebar accepted D98469: [NVPTX] Avoid temp copy of byval kernel parameters..

...in which we all remember why we like writing this recursively. :D

Mar 12 2021, 1:31 PM · Restricted Project

Mar 11 2021

jlebar added a comment to D98469: [NVPTX] Avoid temp copy of byval kernel parameters..

This is going to be so great. Thanks, Art.

Mar 11 2021, 4:42 PM · Restricted Project

Mar 1 2021

jlebar accepted D97708: [CUDA] Remove `noreturn` attribute from __assertfail()..

Maybe add a comment on assertfail that we don't want it to be noreturn because blah blah? So someone doesn't come back and add it when we're not looking. :)

Mar 1 2021, 11:01 AM · Restricted Project

Feb 25 2021

jlebar committed rGe890fffcab8b: Fix signed-compare warning. (authored by jlebar).
Fix signed-compare warning.
Feb 25 2021, 6:16 PM
jlebar committed rGc90dac27e94e: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best. (authored by jlebar).
[clang] Print 32 candidates on the first failure, with -fshow-overloads=best.
Feb 25 2021, 5:47 PM
jlebar closed D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..
Feb 25 2021, 5:47 PM · Restricted Project

Feb 24 2021

jlebar accepted D97404: [CUDA, test-suite] Disable testing of non-FP complex types..
Feb 24 2021, 11:40 AM

Feb 20 2021

jlebar added a comment to D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..

Thank you for the review!

Feb 20 2021, 8:37 AM · Restricted Project

Feb 19 2021

jlebar added a comment to D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..

I guess I was confused by the function name

Feb 19 2021, 10:21 PM · Restricted Project
jlebar updated the diff for D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..

Rename noteNumOverloadCandidatesShown -> overloadCandidatesShown.

Feb 19 2021, 10:20 PM · Restricted Project

Feb 15 2021

jlebar added inline comments to D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..
Feb 15 2021, 10:57 PM · Restricted Project
jlebar added a comment to D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..

Thank you for your comments, Aaron!

Feb 15 2021, 8:51 AM · Restricted Project

Feb 13 2021

jlebar added a comment to D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..

Not sure who can review this, but looking through blame it seems like maybe @aaronpuchert?

Feb 13 2021, 11:28 AM · Restricted Project
jlebar added a reviewer for D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best.: aaronpuchert.
Feb 13 2021, 11:28 AM · Restricted Project
jlebar added a comment to D92210: Don't sink ptrtoint/inttoptr sequences into non-noop addrspacecasts..

Bump, anything else needed here?

Feb 13 2021, 11:18 AM · Restricted Project, Restricted Project

Jan 30 2021

jlebar added a reviewer for D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best.: rsmith.
Jan 30 2021, 7:06 PM · Restricted Project
jlebar requested review of D95754: [clang] Print 32 candidates on the first failure, with -fshow-overloads=best..
Jan 30 2021, 7:05 PM · Restricted Project

Jan 22 2021

jlebar accepted D95255: [RGT][ADT] Remove test assertion that will not be executed..
Jan 22 2021, 1:18 PM · Restricted Project

Jan 10 2021

jlebar added a comment to D94337: Add cuda header type for cuh files.

Hi, welcome! Thank you for the careful and well-motivated first commit. (I also see https://github.com/ccache/ccache/issues/772, hooray for noticing that...)

Jan 10 2021, 5:39 PM · Restricted Project

Dec 4 2020

jlebar accepted D92684: [CUDA, tet-suite] enable testing with C++17 and C++20.
Dec 4 2020, 1:39 PM
jlebar accepted D91807: [CUDA] Unbreak CUDA compilation with -std=c++20.
Dec 4 2020, 11:43 AM · Restricted Project

Nov 29 2020

jlebar added a comment to D92210: Don't sink ptrtoint/inttoptr sequences into non-noop addrspacecasts..

This looks reasonable to me (and I appreciate all the debugging in the Julia bug!), but I have never touched this code, so I don't 100% feel comfortable approving the change.

Nov 29 2020, 11:19 AM · Restricted Project, Restricted Project

Nov 27 2020

jlebar added a reviewer for D92210: Don't sink ptrtoint/inttoptr sequences into non-noop addrspacecasts.: tra.
Nov 27 2020, 10:34 AM · Restricted Project, Restricted Project

Nov 22 2020

jlebar added a comment to D91928: [nvptx] Skip alloca for read-only byval arguments..

In case it's not used in PHI or SELECT and cannot ensure the result is also a pointer to the parameter space, we could skip alloca insertion.

Nov 22 2020, 2:38 PM · Restricted Project
jlebar added a comment to D91928: [nvptx] Skip alloca for read-only byval arguments..

I don't believe there's any exception to prove deduction [of the readonly attribute] wrong.

Nov 22 2020, 9:40 AM · Restricted Project
jlebar added a comment to D91928: [nvptx] Skip alloca for read-only byval arguments..

This looks really simple, which is awesome. I am enthusiastic. But I am worried it may not be correct.

Nov 22 2020, 12:01 AM · Restricted Project

Nov 19 2020

jlebar accepted D91807: [CUDA] Unbreak CUDA compilation with -std=c++20.

How fun. :)

Nov 19 2020, 10:22 AM · Restricted Project

Nov 17 2020

jlebar added a comment to D91590: [NVPTX] Efficently support dynamic index on CUDA kernel aggregate parameters..

I am legit excited about this if we could figure out how to make it work, but I don't have anything to add beyond what tra said.

Nov 17 2020, 3:24 PM · Restricted Project, Restricted Project

Nov 9 2020

jlebar added a comment to D85603: IR: Add convergence control operand bundle and intrinsics.

The bottomline seems to be that the proposed formalism achieves the following: <snip>

Nov 9 2020, 4:49 PM · Restricted Project

Nov 3 2020

jlebar added a comment to D90409: [HIP] Math Headers to use type promotion.

LGTM. I think the change would make sense for CUDA, too. @jlebar - WDYT?

Nov 3 2020, 10:45 AM · Restricted Project, Restricted Project

Nov 1 2020

jlebar added a comment to D85603: IR: Add convergence control operand bundle and intrinsics.

Man, phab doesn't make this easy, does it?

Nov 1 2020, 9:49 AM · Restricted Project

Oct 28 2020

jlebar updated subscribers of D85603: IR: Add convergence control operand bundle and intrinsics.

Hi. :) A few people pinged me asking for my feedback here, since I touched the convergent attr way back in the day, for CUDA.

Oct 28 2020, 11:34 PM · Restricted Project

Oct 22 2020

jlebar accepted D89832: [CUDA] Extract CUDA version from cuda.h if version.txt is not found.

LGTM modulo emankov's comment.

Oct 22 2020, 10:04 AM · Restricted Project

Oct 14 2020

jlebar committed rGe9ac1869a82a: Preserve param alignment in NVPTXLowerArgs pass. (authored by jlebar).
Preserve param alignment in NVPTXLowerArgs pass.
Oct 14 2020, 11:16 AM
jlebar closed D89404: Preserve param alignment in NVPTXLowerArgs pass..
Oct 14 2020, 11:15 AM · Restricted Project
jlebar updated the summary of D89404: Preserve param alignment in NVPTXLowerArgs pass..
Oct 14 2020, 11:10 AM · Restricted Project
jlebar updated the diff for D89404: Preserve param alignment in NVPTXLowerArgs pass..

Fix bug in test command line, and fix commit message.

Oct 14 2020, 11:09 AM · Restricted Project
jlebar added a comment to D89404: Preserve param alignment in NVPTXLowerArgs pass..

It does not quite match what we do here.

Oct 14 2020, 11:08 AM · Restricted Project
jlebar requested review of D89404: Preserve param alignment in NVPTXLowerArgs pass..
Oct 14 2020, 9:59 AM · Restricted Project

Oct 13 2020

jlebar accepted D88345: [CUDA] Allow local `static const {__constant__, __device__}` variables..
Oct 13 2020, 3:31 PM · Restricted Project

Oct 2 2020

jlebar added a comment to D88345: [CUDA] Allow local `static const {__constant__, __device__}` variables..

Hey, I'm leaving on a vacation tomorrow and didn't have a chance to get to
this review today.

Oct 2 2020, 10:14 PM · Restricted Project

Oct 1 2020

jlebar added a comment to D88668: [CUDA] Add support for 11.1.

It looks like 11.1 doesn't have a version.txt file

Oct 1 2020, 9:18 AM · Restricted Project

Sep 28 2020

jlebar added a comment to D88345: [CUDA] Allow local `static const {__constant__, __device__}` variables..

It should. I did mention in a previous comment that > Looks like the const-ness check should not be there, either. I need to revise the patch.

Sep 28 2020, 4:38 PM · Restricted Project
jlebar added a comment to D88345: [CUDA] Allow local `static const {__constant__, __device__}` variables..

OK, now I'm starting to I understand this change..

Sep 28 2020, 12:03 PM · Restricted Project
jlebar added a comment to D88345: [CUDA] Allow local `static const {__constant__, __device__}` variables..

OK, backing up, what are the semantics of static on __constant__, __device__, and __shared__?

Sep 28 2020, 10:46 AM · Restricted Project

Sep 25 2020

jlebar added a comment to D88345: [CUDA] Allow local `static const {__constant__, __device__}` variables..

wha... As you know, const doesn't mean anything, that can be const-casted away. And then you'll be able to observe that this nominally-static variable is just a normal variable.

Sep 25 2020, 4:47 PM · Restricted Project

Sep 24 2020

jlebar accepted D88255: [test-suite, CUDA] Compile-time test for builtin variables..
Sep 24 2020, 1:03 PM
jlebar accepted D88250: [CUDA] Added dim3/uint3 conversion functions to builtin vars..

I know it comes in a separate change, but can we add a check to the test-suite?

Sep 24 2020, 11:14 AM · Restricted Project

Aug 5 2020

jlebar accepted D85352: [CUDA, test-suite] More test cases for rint() and nearint().
Aug 5 2020, 12:59 PM
jlebar accepted D85349: [CUDA, test-suite] Prevent constant folding of the test inputs..
Aug 5 2020, 12:58 PM

Aug 4 2020

jlebar accepted D85236: [CUDA] Work around a bug in rint() caused by a broken implementation provided by CUDA..

LGTM, and can we write a test in the test-suite?

Aug 4 2020, 12:25 PM · Restricted Project

Jul 15 2020

jlebar added a comment to D83893: [CUDA][HIP] Always defer diagnostics for wrong-sided reference.

tra and I talked offline and I...think this makes sense.

Jul 15 2020, 4:33 PM · Restricted Project

Mar 1 2020

jlebar added a comment to D74165: [x86] [DAGCombine] Prefer shifts of constant widths..

@RKSimon and others, thank you for the review and comments. Sorry for my delay here; this has changed from being my day job to my weekend hobby, and that made a bigger difference in my responsiveness than I'd like or than I expected.

Mar 1 2020, 12:43 PM · Restricted Project
jlebar updated the diff for D74165: [x86] [DAGCombine] Prefer shifts of constant widths..

Update per comments.

Mar 1 2020, 12:40 PM · Restricted Project

Feb 23 2020

jlebar committed rG10a0ec62f671: Replace std::foo with std::foo_t in LLVM. (authored by jlebar).
Replace std::foo with std::foo_t in LLVM.
Feb 23 2020, 11:27 AM

Feb 18 2020

jlebar added inline comments to D74165: [x86] [DAGCombine] Prefer shifts of constant widths..
Feb 18 2020, 1:30 PM · Restricted Project
jlebar added inline comments to D74165: [x86] [DAGCombine] Prefer shifts of constant widths..
Feb 18 2020, 1:12 PM · Restricted Project