Yeounoh Chung
Yeounoh Chung
Re-assigning the issue to @vanbasten23 .
Going to breakdown this PR into smaller ones and land them separately. Some of the unit tests need to land in the end, after landing all the changes cc @JackCaoG...
cc @jonb377 FYI
@JackCaoG could you help review this PR, this would be the last one of the three PoC PRs. We need this to unblock @steventk-g for his virtual device work. I...
> This pr is still too big and hard to review. I would argue that we don't have enough test to confidently land this. It is OK to make an...
> you need to update this pr a bit to account for new tf update, should not be too much code changes through Do we have a reference PR in...
@JackCaoG I addressed the comments, thanks again for the review. I was able to build locally and test.
> Seems like test strict out crashed. @yeounoh were you able to run a simple pytorch/xla program locally? Thanks @JackCaoG , the tests were green at some point (or not??)....
CPU test passes, but the GPU fails with the following somewhat unrelated (at least on the outset) error: ``` *** Begin stack trace *** tsl::CurrentStackTrace[abi:cxx11]() gsignal abort xla::XrtLocalService::XrtLocalService(std::__cxx11::basic_string const&, std::__cxx11::basic_string...
> Seesm irrelevant, let me just restart the gpu ci Yea, this [one](https://app.circleci.com/pipelines/github/pytorch/xla/13795/workflows/7206b828-64a7-4a6c-8307-cee3a8b54e06) succeeded. Thanks @JackCaoG