Can I prefetch specific data to a specific cache level in a CUDA kernel?

Question

I understand that Fermi GPUs support prefetching to L1 or L2 cache. However, in the CUDA reference manual I can not find any thing about it.

Dues CUDA allow my kernel code to prefetch specific data to a specific level of cache?

score 6 · Accepted Answer · edited Mar 19 '17 at 22:23

6

Well not at instruction level but detailed information about prefetching in GPUs in here:

You can find instruction reference in nVIDIA's PTX ISA reference document; the relevant instructions are prefetch and prefetchu.

edited Mar 19 '17 at 22:23

einpoklum

answered Feb 09 '11 at 21:10

kerem

1

I appreciate the information. It is a pity that CUDA does not provide prefetching instructions. – dalibocai Feb 14 '11 at 02:05
Updated the links... but is that paper still relevant these days (i.e. for the Maxwell and Pascal microarchitectures?) – einpoklum Mar 19 '17 at 22:26

1 Answers1