Date: Thu, 27 Aug 2015 17:53:20 +0200 From: magnum <john.magnum@...hmail.com> To: john-dev@...ts.openwall.com Subject: Re: LWS and GWS auto-tuning On 2015-08-27 01:11, magnum wrote: > On 2015-08-27 01:07, Solar Designer wrote: >> On Thu, Aug 27, 2015 at 01:01:27AM +0200, magnum wrote: >>> Max Work Group Size: 8192 <---- here! >>> Parallel compute cores: 8 >>> >>> I'm do not think the de-facto limit of 1024 we've been used to is an >>> actual maximum per any specifications. Also, when I tried this it ran >>> just fine through the tests up to 8192 but picked a lower number as >>> best. If it wasn't actually supported, we should get an >>> CL_INVALID_WORK_GROUP_SIZE error and it would have been caught and >>> handled properly. >>> >>> I presume your segfault was unrelated to the work size. >> >> OK. There was also "OpenCL error (CL_INVALID_VALUE) in file >> (opencl_cryptmd5_fmt_plug.c) at line (381) - (Copy data back)". >> >> Are you going to look into these, or/and create GitHub Issues so they >> are not forgotten? > > That is already fixed (handled) by today's changes. If you get a > segfault running edfc502630 or later, I will regard it a problem. After some (well a lot) regression testing with nvidia and Intel CPU, some issues where addressed. cf6f3457f seems to be a decent version. Now another round of regression testing on AMD and others o.O Current version has this in john.conf (might be a temporary feature) # Initial local work-size for auto-tune (CPU devices excepted). # 0 means let the OpenCL implementation pick a suitable value. # 1 means query for "best multiple" (usually corresponds to "warp size"). # Any other value (eg. 64) will be taken verbatim. AutotuneLWS = 1 We could add a couple other queries as 2, 3 and so on if we come up with any good ones. 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.