Skip to content

now works with nvcc compiler#184

Merged
JohanMabille merged 2 commits intoxtensor-stack:masterfrom
gridley:nvcc_compat
Jan 13, 2020
Merged

now works with nvcc compiler#184
JohanMabille merged 2 commits intoxtensor-stack:masterfrom
gridley:nvcc_compat

Conversation

@gridley
Copy link
Copy Markdown
Contributor

@gridley gridley commented Dec 31, 2019

This solves issue #183 and allows XTL to be used with host-side CUDA code. Previously, the nvcc compiler failed to correctly parse two LOC in XTL, and these expressions have been modified to something more friendly for nvcc. As a result of this, xtensor now also works with host-side CUDA!

I think adding a separate test for the nvcc compiler may not make sense at this point though.

Also... I hope I understood the rule on modifying the changelog correctly. Does each PR modify the changelog?

@JohanMabille
Copy link
Copy Markdown
Member

JohanMabille commented Dec 31, 2019 •
edited
Loading

Thanks for this PR!
Regarding the changelog, I will remove the requirement since this will lead to conflicts if many PRs are opened simultaneously.

JohanMabille
JohanMabille requested changes Dec 31, 2019
View reviewed changes
@gridley gridley mentioned this pull request Dec 31, 2019
Merged
@gridley
Copy link
Copy Markdown
Contributor Author

gridley commented Jan 13, 2020

Hey, just checking up on this. I think all of your concerns have been addressed now @JohanMabille.

@JohanMabille
Copy link
Copy Markdown
Member

JohanMabille commented Jan 13, 2020

Indeed, sorry for the late reply, I've been busy for the past few days. Let's merge this PR, next step is to upgrade to last implementation of variant.

@JohanMabille JohanMabille merged commit 69c0d12 into xtensor-stack:master Jan 13, 2020
@gridley
Copy link
Copy Markdown
Contributor Author

gridley commented Jan 14, 2020

@JohanMabille are you a git submodule type of guy, or no (I've heard some dislike using them)? May make sense for mpark/variant. If so I'd totally make a PR for that.

@JohanMabille
Copy link
Copy Markdown
Member

JohanMabille commented Jan 14, 2020

We usually consider git submodule as a terrible practice. The ideal approach would be to make mpark/variant a conda package and have xtl depending on it. However, since it's a single file, it is totally acceptable (and simpler) to have a copy of it in xtl, that we synchronize whith upstream when it is needed.

@gridley gridley deleted the nvcc_compat branch January 17, 2020 18:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@JohanMabille JohanMabille JohanMabille approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@gridley @JohanMabille