Date: Fri, 7 Sep 2012 19:07:47 +0200 From: magnum <john.magnum@...hmail.com> To: john-dev@...ts.openwall.com Subject: Re: OpenCL auto-find GPU It turns out the below fix forced me to fix a kludge that I knew would bite us sooner or later: both CUDA and OpenCL used the same gpu_id global variable. I have now committed a patch that differentiates them. This fixes subtle problems with the below patch and also avoids several future problems one can imagine. The downside is anyone with non-committed work will have to rebase and possibly fix conflicts. Please "pull --rebase". magnum On 7 Sep, 2012, at 3:31 , magnum <john.magnum@...hmail.com> wrote: > I have committed an OpenCL change that make JtR pick first GPU found, unless a specific device is requested on command-line or in john.conf. If none is found, and none is specified, it reverts to the old default of platform 0, device 0. I got tired of keeping track of where the GPU ended up on various machines. > > 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.