fancyix

How to controll opencl kernel configuration of assembly code generated by llvm clang AMDGPU backend?

Discussion created by fancyix on Aug 31, 2018

llvm clang can compile opencl file into assembly. A common format is hsa. There are certain configurations in hsa assembly file, such as enable_sgpr_dispatch_ptr and enable_sgpr_queue_ptr. When I compile my opencl file, the generated assembly file has some configure on and some off. e.g. enable_sgpr_dispatch_ptr is set to 1 but enable_sgpr_queue_ptr is set to 0. Can I control those configurations when llvm compile opencl? Something like passing a command line parameter or define some env? I am also wondering if this is done through libclc.

Outcomes