Add a new owner for libdevice to CODEOWNERS

[romanovvlad]

No description provided.

[pvchupin]

@cperkinsintel, any ideas why Sampler/unnormalized-none-linear.cpp is failing?

[cperkinsintel]

@cperkinsintel, any ideas why Sampler/unnormalized-none-linear.cpp is failing?

@pvchupin ,
No, I'm not sure why it would fail. I checked that test, and while it is XFAIL for CPU and CUDA devices, it should pass on LevelZero and Windows. I just logged into my Win dev machine to double check it, and it passes there fine, further it had to pass on the CI system before it could be merged into llvm-test-suite. I'll see what else I can find.

It's failing on the very last output of that test. I cannot reproduce the failure. However, I had already made a comment in the test itself about that very line, that interpolation at that position seems possibly incorrect. And there is a discussion with the OpenCL folk on some of the outstanding tickets about the pixel data requirements when using linear interpolation. So, for the moment, I'm going to push a PR to the test that removes that last output check. And I may soon be updating all the linear interpolation lit tests.

[pvchupin]

Thanks @cperkinsintel.