Skip to content
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

[WIP] Add full test case for variadic delayed Matrix allocation #420

Open
wants to merge 6 commits into
base: development
Choose a base branch
from

Conversation

devreal
Copy link
Member

@devreal devreal commented Jul 10, 2017

@fuchsto: Please have a look at this. Your test MatrixTest.DelayedAlloc succeeds while the new test MatrixTest.VariadicDelayedAlloc fails while checking the validity of the pattern arguments, which are the same as in the first test. So either the first test case is invalid or the check in PatternArguments is overly strict. I cannot decide that :)

@devreal devreal added the bug label Jul 10, 2017
@devreal devreal requested a review from fuchsto July 10, 2017 20:33
@devreal
Copy link
Member Author

devreal commented Jul 11, 2017

Hmpf, the error is not triggered in CI. Here is what I see on my laptop:

[= -1 LOG =]               TestBase.h : 251 | ===> Running test case MatrixTest.VariadicDelayedAlloc ...
[=  1 LOG =]               TestBase.h : 254 | -==- DASH initialized with 4 units
[    1 ERROR ] [  9679 ] PatternArguments.h       :280  | dash::exception::InvalidArgument             | [ Unit 1 ] Pattern arguments invalid: Mixed distribution types
[  ERROR   ] [UNIT 1] in [=  1 LOG =]               TestBase.h : 264 | -==- Test case finished at unit 1
[=  1 LOG =]               TestBase.h : 267 | -==- Finalize DASH at unit 1
[= -1 LOG =]               TestBase.h : 272 | <=== Finished test case MatrixTest.VariadicDelayedAlloc with 4 units

@ddiefenthaler
Copy link
Contributor

I can report the same error, for me it even triggers in the first test MatrixTest.DelayedAlloc.
It occurs in a 32bit VM, only when run with 4 or 6 units.

I tried to debug it a little bit and it seems to fail in check_tile_constraints() of PatternArguments.h#L265.
The distribution passed down seems to be [NONE, TILE(5), TILE(3)].

@devreal
Copy link
Member Author

devreal commented Oct 1, 2017

The main issue has been fixed in #444 (thanks to @fmoessbauer). What is left in this PR is splitting the test for variadic matrix allocation into a separate test case.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants