Hi,
I am working with an ARM Cortex A9 processor. I have a document that describes the duration (in terms of cycle) of SIMD intrinsics (Cortex-A9 NEON Media Processing Engine Tech. Ref. Manual). There i found the result (execution time) and writeback cycle durations. It is written that some instructions have to wait until the writeback is finished (the value is written back to register file).
What I want to ask is,
1)Do these cycle durations refer to all NEON 128-bit intrinsics (float32x4, int32x4, etc.)? 2)which instructions have to wait till the end of any instruction's writeback cycle?
Thanks,
Murat