Date: Tue, 28 Feb 2012 15:40:01 -0300 From: Claudio André <claudioandre.br@...il.com> To: john-dev@...ts.openwall.com Subject: Re: OpenCL patches Hi, The sem_wait stuff is related to some AMD/NVIDIA bad relationship. On a clean AMD machine, it works fine on 2.6. I can do some performace test, if you think is useful. By the way, the machine i reported the error some days ago (and did the same magnun did, OpenCL AMD CPU) has a AMD Radeon HD 6370 DDR3, 512 MB with 2 stream cores. New hardware for poor countries: http://www.submarino.com.br/produto/10/23836952/notebook+hp+g42-373br+c/amd%C2%AE+dual+core+4gb+500gb+led+14+windows+7 Date: Fri, 24 Feb 2012 18:05:54 +0100 From: magnum<john.magnum@...hmail.com> To: john-dev@...ts.openwall.com Subject: OpenCL patches Samuele, Lukas, I just committed a couple of trivial OpenCL patches to git (after manually syncing with Samueles tree). Among other things I added support for --platform. With this I could successfully chose whether to use CPU or GPU. I also plan to support --platform=LIST (I added a temporary separate tool to unused/ for now). I think we should rename --gpu to --device before next Jumbo release, because it doesn't really have to be a GPU. I also tried AMD's version of OpenCL for CPU. Worked fine. I can't really compare performance because this was on a machine in an whole different league. I had some problems (sem_wait() failed, seems to be a common regression problem) with AMD APP 2.6 so I reverted to 2.5. magnum Content of type "text/html" skipped
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.