projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 2815782) | patch
R600: Treat CONSTANT_ADDRESS loads like GLOBAL_ADDRESS loads when necessary
authorTom Stellard <thomas.stellard@amd.com>
Tue, 23 Jul 2013 23:54:56 +0000 (23:54 +0000)
committerTom Stellard <thomas.stellard@amd.com>
Tue, 23 Jul 2013 23:54:56 +0000 (23:54 +0000)
commit8cb0e47c9ee2061cedb06bbb648e1d5c2efea82e
tree3656bbef9c009cd3caed330b86d2e12327664963tree | snapshot
parent28157828313582762803cd751acce078e8ebd49ecommit | diff
R600: Treat CONSTANT_ADDRESS loads like GLOBAL_ADDRESS loads when necessary

These are really the same address space in hardware.  The only
difference is that CONSTANT_ADDRESS uses a special cache for faster
access.  When we are unable to use the constant kcache for some reason
(e.g. smaller types or lack of indirect addressing) then the instruction
selector must use GLOBAL_ADDRESS loads instead.

llvm-svn: 187006
llvm/lib/Target/R600/AMDGPUISelDAGToDAG.cpp diff | blob | history
llvm/lib/Target/R600/R600Instructions.td diff | blob | history
llvm/test/CodeGen/R600/load.ll diff | blob | history
Domain: System / Toolchain;