Date: Thu, 04 Jun 2015 23:36:27 +0200 From: magnum <john.magnum@...hmail.com> To: john-dev@...ts.openwall.com Subject: Re: PHC: Parallel in OpenCL On 2015-06-04 16:49, Agnieszka Bielec wrote: > 2015-06-04 13:38 GMT+02:00 magnum <john.magnum@...hmail.com>: >> On 2015-06-04 12:07, magnum wrote: > >> Also, your use of the shared auto-tune function is totally busted. You'd be >> much better off not using auto-tune at all than configuring it wrong. >> Attached is a patch that mostly fixes it. > > I see that you changed BENCH_CLERROR to HANDLE_CLERROR. why? I saw > BENCH_CLERROR somewhere That was a mistake. I copied crypt_all() to crypt_all_benchmark to get all missing pieces. Normally it doesn't matter much though. Also, in crypt_all() all instances of multi_profilingEvent can be replaced with NULL. But that's cosmetic. More importantly, the for-loop around the loop kernel invocations can/should be commented away (so they only get called once) in crypt_all_benchmark() if you like to try auto-tuning with higher cost. magnum
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.