Merge r319025 into the 5.0 branch : [AMDGPU] Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics

[llvmbot]

Bugzilla Link	35442
Version	5.0
OS	All
Blocks	#33840
Reporter	LLVM Bugzilla Contributor
CC	@jvesely,@tstellar
Fixed by commit(s)	r319025

Extended Description

Is it OK to merge the following revision(s) to the 5.0 branch?

[llvmbot]

https://reviews.llvm.org/rL319025

[tstellar]

Some test cases fail when I apply this to the release_50 branch would you be able to supply a patch that applies cleanly and doesn't regress tests?

Also, it's pretty late in the process, so I'm not going to hold -rc2 back waiting for this to be resolved.

[tstellar]

This was submitted too late to make it into 5.0.1-rc2, so it won't be in the release unless it is determined to be a critical fix.

[arsenm]

Old release

[llvmbot]

@llvm/issue-subscribers-backend-amdgpu