Sreelakshmi Haridas Maruthur

Fix for FMA verified. Looks good.

Can we remove focused review from this until we find someone to pick it up?

We are seeing failures in multiple tests with this patch and haven't fully resolved them to bugs in the test or in our implementation. List: Add -> Fails when one...

We would like to double check results at our end, can we get some more time.

We are opposed to this change merging in present form. The precision should be calculated in the same type as being used by the builtin (half for half, and float...

Could we remove focused review from this change since it is pending resolution of the spec discussion? We can add it back once we get there.

Need time, reviewing. Sreelakshmi Haridas Maruthur Qualcomm

It appears that the spec implicitly limits the ballot builtin to 128 so the bitset definition seems correct and implementations must be limiting max workgroup size to 128 when using...

> @lakshmih Thanks for review. Unfortunately I see some errors on our internal driver version. Until these issues will not be solved I do not plan any updates in this...

> > > @lakshmih Thanks for review. Unfortunately I see some errors on our internal driver version. Until these issues will not be solved I do not plan any updates...