Skip to content

[SYCL] Fix regression with program building #865

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged

Conversation

s-kanaev
Copy link
Contributor

@s-kanaev s-kanaev commented Nov 22, 2019

This patch fixes regression introduced by #847: programs and kernels are cached when program is built with sequence of compile_with_kernel_type()/compile_with_source() and then linked with link().

By design, caching of kernels and programs is permitted only when program is built with build_with_kernel_type() using the default options.

@s-kanaev s-kanaev force-pushed the private/s-kanaev/fix-thread-safety-regression branch from c018a08 to a4bb7f4 Compare November 22, 2019 13:46
@s-kanaev
Copy link
Contributor Author

Split to two pull requests

@s-kanaev s-kanaev changed the title [SYCL] Introduce thread-safety for kernels and programs cache. Improve tests. Improve implementation of programs and kernels cache [SYCL] Improve tests. Improve implementation of programs and kernels cache Nov 22, 2019
@s-kanaev s-kanaev requested a review from asavonic November 22, 2019 13:49
@s-kanaev s-kanaev force-pushed the private/s-kanaev/fix-thread-safety-regression branch from a4bb7f4 to 02b83dc Compare November 25, 2019 07:20
@bader
Copy link
Contributor

bader commented Nov 25, 2019

This patch fixes regression

Could you elaborate on the problem this patch fixes, please?
"Improve tests. Improve implementation ..." is too vague. Please, be more specific.
Useful guide lines:

@s-kanaev s-kanaev changed the title [SYCL] Improve tests. Improve implementation of programs and kernels cache [SYCL] Fix regression with program building Nov 25, 2019
@s-kanaev s-kanaev requested a review from bader November 25, 2019 12:17
@bader
Copy link
Contributor

bader commented Nov 25, 2019

What problem is introduced by #847?

@s-kanaev
Copy link
Contributor Author

What problem is introduced by #847?

Programs and kernels tend to be cached in some illegal (for caching) use-cases.

@bader
Copy link
Contributor

bader commented Nov 25, 2019

What problem is introduced by #847?

Programs and kernels tend to be cached in some illegal (for caching) use-cases.

What illegal cases are allowed with #847?

@s-kanaev
Copy link
Contributor Author

What illegal cases are allowed with #847?

Programs and kernels are cached when program is built with sequence of compile_with_kernel_type/compile_with_source and then linked.

By design, caching of kernels and programs is permitted only when program is built with build_with_kernel_type with default options.

@bader
Copy link
Contributor

bader commented Nov 25, 2019

What illegal cases are allowed with #847?

Programs and kernels are cached when program is built with sequence of compile_with_kernel_type/compile_with_source and then linked.

By design, caching of kernels and programs is permitted only when program is built with build_with_kernel_type with default options.

This is not clear from the patch, so I suggest adding this information to the commit message.

@s-kanaev
Copy link
Contributor Author

This is not clear from the patch, so I suggest adding this information to the commit message.

Done

@s-kanaev s-kanaev force-pushed the private/s-kanaev/fix-thread-safety-regression branch from 8d70a84 to aa858c7 Compare November 25, 2019 13:48
@s-kanaev s-kanaev requested a review from bader November 25, 2019 13:49
bader
bader previously approved these changes Nov 25, 2019
@s-kanaev s-kanaev force-pushed the private/s-kanaev/fix-thread-safety-regression branch from 2aeba24 to 930bed7 Compare November 25, 2019 14:31
@s-kanaev
Copy link
Contributor Author

Fixed sign-off

bader
bader previously approved these changes Nov 25, 2019
Signed-off-by: Sergey Kanaev <[email protected]>
@bader
Copy link
Contributor

bader commented Nov 25, 2019

This is not clear from the patch, so I suggest adding this information to the commit message.

Done

I can't find the fixed commit message. Could you share the link, please?
I'd like to use it for the squashed commit message.

@s-kanaev
Copy link
Contributor Author

I can't find the fixed commit message. Could you share the link, please?
I'd like to use it for the squashed commit message.

Somehow, the empty commit just vanished away. I've edited description of the patch.

@bader bader merged commit d442364 into intel:sycl Nov 25, 2019
@s-kanaev s-kanaev deleted the private/s-kanaev/fix-thread-safety-regression branch November 26, 2019 07:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants