Date: Mon, 02 Apr 2012 21:43:16 +0200 From: magnum <john.magnum@...hmail.com> To: john-dev@...ts.openwall.com Subject: Re: JtR: GPU for slow hashes On 04/02/2012 06:33 PM, Solar Designer wrote: > On Mon, Apr 02, 2012 at 09:59:15PM +0530, Dhiru Kholia wrote: >> On Mon, Apr 2, 2012 at 9:47 PM, SAYANTAN DATTA <std2048@...il.com> wrote: >>> While running the benchmark I get the following error after a few seconds: >>> ../../../thread/semaphore.cpp:87: sem_wait() failed >>> aborted >> >> Are you using AMD APP SDK 2.6? This is a known issues with version >> 2.6. Downgrade to 2.5 and try again. > > FWIW, as discussed in here before, downgrading did not help in my case, > but setting OS_TIMER to 0 did. > > I think magnum has since set OS_TIMER to 0 in x86-64.h, but not in > x86-sse.h. We need to change that to only set OS_TIMER to 0 when > building with OpenCL support with AMD SDK specifically (since OS_TIMER 1 > is preferable when we can have it), but to do it from both header files. I was planning to do it for mingw but I haven't yet changed any arch.h. On the other hand, I've been running AMD 2.6 for a week now and only saw this sem_wait() fail once (but I only occasionally test with that platform - different OpenCL compilers give different clues to problems!) We should have a ticket system. Maybe I'll just add a crude to-do list to the wishlist wiki page for now, with things like this that need to be looked into. 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.