[llvm] [AMDGPU] Add function attribute to disable TBUFFER combine (PR #156454)

[Piotr Sobczak via llvm-commits]

piotrAMD wrote:

> > Also note, the middle-end merging never happens for tbuffers, because they are intrinsincs all the way - at no point are they regular llvm loads.
> 
> We should fix that. We should _not_ have any of this MI level merging

Is the suggestion to add an extra ME pass that merges tbuffers? load-store-vectorizer works on the regular load/stores, so it is not easily possible to get that working with tbuffer representation which has extra information (format) on the instruction.

https://github.com/llvm/llvm-project/pull/156454