iree
iree copied to clipboard
[LLVMGPU] Add shared memory swizzle transformation
Enable shared memory swizzle transformation as well as picking a good unrolling order for tensorcore