Date: Mon, 7 Jan 2013 07:31:13 +0400 From: Solar Designer <solar@...nwall.com> To: john-dev@...ts.openwall.com Subject: Re: des-opencl On Mon, Jan 07, 2013 at 04:03:23AM +0400, Solar Designer wrote: > http://www.reddit.com/r/crypto/comments/162ufx/research_project_opencl_bitslice_des_bruteforce/ Another curious detail about mysterymath's implementation is that it obviously exceeds instruction cache, yet is reasonably fast on GCN. We should try that too as it avoids pointer indirection and reduces register pressure due to hard-coding key bit indices (no key schedule array needed anymore). Lukas reported very poor performance on 5850, though. While at it, I found this discussion thread: http://devgurus.amd.com/thread/158867 with some info on GCN ISA level optimizations - how scalar vs. vector instructions get scheduled, etc. With kernels exceeding instruction cache size, the recommendation is to schedule more wavefronts to help hide instruction fetch latency. Alexander
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.