Skip to content

Conversation

@rachtsingh
Copy link
Contributor

@rachtsingh rachtsingh commented Jan 25, 2018

I've taken a stab at reducing code duplication between the CUDA/CPU implementations using one strategy, but happy to try something else if there's an alternative.

@pytorchbot
Copy link
Collaborator

@rachtsingh, thanks for your PR! We identified @zdevito to be a potential reviewer.

@apaszke
Copy link
Contributor

apaszke commented Jan 25, 2018

@rachtsingh I think you have accidentally updated the subtrees

@rachtsingh
Copy link
Contributor Author

Yes, sorry, just fixed (I think my understanding of subtrees is still a little incomplete).

@rachtsingh
Copy link
Contributor Author

Ok, I think it's ready for review - don't love the code dedup solution but very open to alternatives. Maybe @apaszke ?

@unittest.skipIf(not TEST_NUMPY, "Numpy not found")
def test_poisson_gpu_sample(self):
set_rng_seed(0)
set_rng_seed(1)

This comment was marked as off-topic.

@gchanan
Copy link
Contributor

gchanan commented Jan 26, 2018

can you manually check that test_distributions.py still works when you compile with WITH_SCALARS=1?

@rachtsingh
Copy link
Contributor Author

Sure, running the build now.

@rachtsingh
Copy link
Contributor Author

Ok, it works just fine after I cleaned and rebuilt with WITH_SCALARS=1.

@rachtsingh
Copy link
Contributor Author

this is no longer ready for review, assuming the changes in this PR: #4892 affect it. I'll revisit this later today and see what changes need to be made.

@rachtsingh
Copy link
Contributor Author

Closing for now to avoid retriggering spurious builds (and because it doesn't work as-is)

@rachtsingh rachtsingh closed this Feb 26, 2018
@t-vi t-vi mentioned this pull request Apr 23, 2018
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.

5 participants