Fix for HIPCC breakage + update to enable some packet specializations in HIPCC

Copy-pasting the two commit messages here for convenience


Commit 1: Fix for HIP breakage detected on 191210

The following commit introduces compile errors when running eigen with hipcc

2918f85b

hipcc errors out because it requires the device attribute on the methods within the TensorBlockV2ResourceRequirements struct introduced by the commit above.

The fix is to add the device attribute to those methods


Commit 2 : Enable packet specializations when compiling host code with HIPCC

Currently some of the packet specializations within GPU/PacketMath.h are enabled only during device code compile phase, when compiling with HIPCC. This is incorrect, and leads to compile errors when building TF. Those packet specializations need to be enabled for both host and device compile phases, which is what this commit does.


@rmlarsen1 @chhtz

Merge request reports

Loading