* remove llvm requirement from amd
* tests pass
* test
* sink kernarg_size
* move stuff
* amd_asm_matmul to new style
* default type
* fix tests, simpler
* cu mode is faster and simpler
* darken
* Revert "hotfix: skip test/amd in macpytest"
This reverts commit b7dade2adf.
* no llvm subprocess
* simpler
* sys.exec
* cleanup
* process safe
* diag
* arm ftz support
* 5 sec
* this one