Date: Thu, 19 Mar 2015 00:20:44 +0100 From: magnum <john.magnum@...hmail.com> To: john-dev@...ts.openwall.com Subject: Re: oldoffice-opencl fails on 7990 in super On 2014-12-18 17:12, magnum wrote: > On 2014-12-18 16:40, Solar Designer wrote: >> On super: >> >> Device 0: Tahiti >> Local worksize (LWS) 64, global worksize (GWS) 4194304 >> Benchmarking: oldoffice-opencl, MS Office <= 2003 [MD5/SHA1 RC4 OpenCL]... FAILED (cmp_all(1)) >> >> It's too late to fix this for -jumbo-1 now, but maybe for next revision. > > I spend so much time getting things to work with recent drivers, I just > can't spend much time on this old ones. Having said that, I did try a > little. A workaround for a newer driver may very well ruin it for an > older one and vice versa so regression testing is an endless job. This is finally fixed. My rant above was spot on, it was actually an existing bug workaround for Catalyst 14.9 that triggered some *other* bug in the older versions. That's hopeless o.O > One problem is we have no way to tell at build-time what Cat version > we're dealing with. Theoretically we could add logic to common-opencl.c > and pass something like "-DDRIVER_VER=11242", maybe we should try that. We now have such version macros: diff --git a/src/opencl_rc4.h b/src/opencl_rc4.h index fb32a73..f750731 100644 --- a/src/opencl_rc4.h +++ b/src/opencl_rc4.h @@ -22,7 +22,8 @@ // None 2885 626 13860 8097 #define RC4_IV32 // 3633 696 14278 8118 -#if !gpu_amd(DEVICE_INFO) /* bug in Catalyst 14.9, besides it is slower */ +#if !gpu_amd(DEVICE_INFO) || DEV_VER_MAJOR < 1445 +/* bug in Catalyst 14.9, besides it is slower */ #define RC4_UNROLLED_KEY // 3893 817 14340 7245 #define RC4_UNROLLED // 3932 848 14348 7847 #endif oclHashcat supports one specific Catalyst version, that is annoying but I can totally relate to that... 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.