[rc5] Re: rc5-digest V1 #103

andrew meggs insect at antennahead.com
Fri Aug 29 15:15:36 EDT 1997

>Hey, i did have one question, i would have to imagine that by now we have
>at least the
>equivalent computing power of the DESchall (sp?) effort...  I was never
>involved with that
>and have never tried the client out to see for myself.  Does anyone have a
>rough idea
>what our keyrate would be if we tried to crack DES?

It's really tough to compare since every architecture had different
performance characteristics.  On a 604e Mac my final deschall client ran
about 90% faster than the current bovine client. The x86 deschall client
did about 10% behind the ppc one. But Darrell Kindred's feat of getting
64-bit code to work under a 32-bit version of SunOS in the final days of
the project was the main reason WHY those were the final days of the
project -- at the end, DES throughput on Sparcs blew away anything
distributed.net is doing on any platform. See
<http://www.frii.com/~rcv/desbench.htm> for all the details.

Also, the platform mix was decidedly different. As it happened, deschall
had a large number of sparcs and massive support from Sun. Meanwhile, the
MacOS client is the fastest at RC5, and if we had working os/cpu stats
distributed.net we'd probably see that distributed.net has a huge number of
Macs and massive support from Apple -- the different platform distributions
are probably NOT a coincidence, and mine's still bigger.

Bottom line: DESChall clients at 12916 distinct IP's checked 0.776% of the
keyspace in the final 24 hours. A completely unknown number of RC5 clients
checked 0.40% of the keyspace yesterday.

Andrew Meggs, content provider                  Antennahead Industries, Inc.
<mailto:insect at antennahead.com>                 <http://www.antennahead.com>

To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.

More information about the rc5 mailing list