Skip to content
This repository has been archived by the owner on Mar 21, 2024. It is now read-only.

Fix clang compile error #335

Merged
merged 1 commit into from
Jul 22, 2021
Merged

Fix clang compile error #335

merged 1 commit into from
Jul 22, 2021

Conversation

zingdle
Copy link
Contributor

@zingdle zingdle commented Jul 1, 2021

This will fix #334

@alliepiper
Copy link
Collaborator

LGTM, I'll run this through our CI in the next week or so. Thanks for the patch!

alliepiper added a commit to alliepiper/thrust that referenced this pull request Jul 8, 2021
@alliepiper
Copy link
Collaborator

Rebased to pull in recent CI fixes.

DVS CL: 30167018
gpuCI: NVIDIA/thrust#1481

@alliepiper alliepiper added testing: gpuCI in progress Started gpuCI testing. testing: internal ci in progress Currently testing on internal NVIDIA CI (DVS). labels Jul 8, 2021
alliepiper added a commit to alliepiper/thrust that referenced this pull request Jul 21, 2021
@alliepiper
Copy link
Collaborator

Rebased for more CI fixes...

DVS CL: 30211046
gpuCI: NVIDIA/thrust#1481

Copy link
Collaborator

@alliepiper alliepiper left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Tests pass, LGTM!

@alliepiper alliepiper merged commit eeb1a99 into NVIDIA:main Jul 22, 2021
rapids-bot bot pushed a commit to rapidsai/raft that referenced this pull request Feb 16, 2022
This makes RAFT sources compilable with clang.
It fixes some fragile code (using `static const` instead of `static constexpr` or `%laneid` in PTX relying on quirks in nvcc which make this happen).

RAFT is still not compilable with clang entirely though due to the dependencies:
1. cub has this issue before 1.14: NVIDIA/cub#335
2. libcudacxx has issues with atomic, which should be fixed in >= 1.7.0-ea (wasn't able to verify this yet)
3. libcudacxx has issues with variadic CUDA functions, which is apparently fixed by passing `-Xclang -fcuda-allow-variadic-functions` to clang (wasn't able to verify this yet)
3. cooperative_groups from CUDA does not work with clang 11.0 / 11.1 but only with >= 13

EDIT: this is necessary to close #84

Authors:
  - Matt Joux (https://github.com/MatthiasKohl)

Approvers:
  - Corey J. Nolet (https://github.com/cjnolet)
  - Artem M. Chirkin (https://github.com/achirkin)

URL: #424
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
testing: gpuCI in progress Started gpuCI testing. testing: internal ci in progress Currently testing on internal NVIDIA CI (DVS).
Projects
None yet
Development

Successfully merging this pull request may close these issues.

error: use of undeclared identifier 'ProcessFloatMinusZero' when compiling with clang
2 participants