Computer giant Apple made ripples through the industry in the last few weeks with (NO WINDOWS I WILL REBOOT LATER) the announcement they’re going to move from their IBM PowerPC hardware platform to Intel based chips. Apple has always had good hardware, even though it was more expensive. Although recently, it’s their OS that has been more of a pull for consumers. I haven’t read all the industry fallout from the announcement, but I’m sure when that finally happens and we can run OSX on a PC, we’ll be seeing more competition for Windows, and choice almost always helps the consumer. I tried OSX and didn’t really care for it, but I encourage people to try new things and use what they like.

Also significant in Apple news is on the RC5 front. I participate in distributed.net’s project to crack encryption with my computer’s idle cycles. I’ve been doing this since my freshman year of college (when 220k keys/sec was GOOD. Right now my computer is doing about 9Mk/s.) What’s significant is they store which OS each keyblock returned was cracked on, and MacOS overtook Linux for second place for RC5-72. (Windows is still in a strong, strong lead.) That’s significant because the G5 processor that new Macs use has incredible RC5 capabilities. (My 2.2GHz AMD gets just under 9Mk/s. A 2GHz G5 gets about 15Mk/s – a 2.7 over 20Mk/s. Of course all of that will change when Apple is running on the same hardware as Linux and Windows – then I suppose it will be sheer numbers, but for now the smarter architecture took the second place prize. After the World Series and the Final Four, I pay attention to the teams that come in second.

Also significant with respect to RC5 and Apple is the Turing cluster at UIUC (I’ll put links in later, I’m lazy.) Turing has a 640 node cluster with dual 2GHz G5s. That’s 1280 CPUs at 15Mk/s. That’s about 19.2Bk/s, or roughly 11% of yesterday’s total RC5 cracking speed. Daaaamn. I wonder who I have to convince to run dnetc on the cluster? :) Putting it in a different perspective, the #1 individual on rc5-72 has 22M stat blocks (2^28 keys in a stat block.) Turing would pump out 71 stat blocks A SECOND, putting it in the number one stats spot in just four days of runtime. We’ve been working on rc5-72 for 928 days and they could be #1 in four. That’s some serious cracking.

If you’re interested in RC5 and want to join my old team, search in stats for 1367 – we’re the Cracking Illini. Join up; we could use the numbers. If I ever get around to updating the static side of my site, I’ll put in a page for distributed.net, and you can wow your friends with knowledge of golumb rulers. (Almost as interesting as the origin of “Live from new york it’s saturday night,” but if I give you all of my stories I won’t be any fun at cocktail parties.)

Ha, one blog post with a single thought! And you thought I couldn’t do it. Also, I’m older than PacMan. (See, I couldn’t do it.)



One Comment to “An Apple a day”

  1. Flash | June 20th, 2005 at 9:41 am

    I’m still somewhat torn about distributed computing. Back in the day, I cracked with the Crackin’ Illini, but these days I’m folding with Team Egg Roll (Ars Technica). I’ve never liked SETI@Home, but I’m torn between the other two. I understand the implications of both, but don’t know which provides more actual benefit…or which needs my help more. Thoughts?