[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 23 Aug 2013 01:23:44 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: *pcount type should be uint64_t
Sayantan,
On Fri, Aug 23, 2013 at 02:46:00AM +0530, Sayantan Datta wrote:
> For raw-md5 kernel, kernel run time is around 700ms for 7970 with 2.8G keys
> per crypt. For raw-sha1 kernel, run time is 2.5s with 5.5G keys per crypt,
> but this config seems to give most optimal performance. For md5 kernel c/s
> report seems to be correct but for raw-sha1 it is incorrect. Maybe no
> information is lost in case of 2.8G (< 2^32) keys but some bits are lost
> in case of 5.5G(> 2^32) keys.
>
> Since raw-sha1 kernel kernel run time seems bit high , I will reduce kpc
> for raw-sha1, which should give correct c/s output.
>
> For other hashes especially raw-md4 I think we might be able to limit
> run-time to less than 1s even with 4G+ keys. So we are pretty much on the
> edge of 32bit limit for current generation GPUs. Maybe the change would be
> required for next or next to next generation of GPUs.
All of the kernel running times you mention above are way too high. Can
you bring them to under 100 ms, without much performance impact? Please
try. Per our previous experience, about the absolute maximum is 200 ms.
Above that are ASIC hangs. Maybe things have changed with newer
Catalyst, and also this did appear to vary by kernel (for some kernels
longer running times did not trigger ASIC hangs), but to be on the safe
side we should avoid kernel running times of above 200 ms, and
preferably keep them under 100 ms.
The *pcount data type does not worry me much (although indeed we need to
avoid overflowing it, either by keeping the values smaller or by making
it 64-bit). Those excessive kernel running times are the real problem,
and one we don't address by merely avoiding *pcount overflow.
Thanks,
Alexander
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.
