A good project if you are in to math. An even better project if you are on a dialup line, or have to do a lot of sneaker-netting. There are two things going on here. The first is the sieving, and the second is llrnet. llrnet is that part that is actually looking for the prime numbers. Sieving just screens though a block of numbers and eliminates definite no-go numbers. The AMD chips excell at sieving, and the P4's and celerons are the ones for running llrnet. llrnet uses SSE2, and the AMD-64 SSE2 implementation doesn't grind out as hot as the Intel implementation.

I grind under the name of Brucifer on RieselSieve, just as some of my accounts here were/are under Brucifer. I have 3 windows boxes going on sieving, and a small farm of AMD xp's. Sieving is a cpu frequency thing. A good item for sieving is the XP2400, runs at 2 gig, and for the cost/performance ratio on sieving you can't beat it. My farm runs on linux. Keeps the cost down. There is roughly a 3 to 4 Kp/s difference between windows and linux on sieving using xp2400's, in favor of windows.

Currently a re-write is in progress to bring up the speed of the sieving client. Once that is done, then the goal is to try and do an amd-64 client to take advantage of the chips benefits.

If you would like to give this a whirl, and want some help with it, send me a PM and I will help you get rolling. For those using windows there is a pretty picture walk though on the www.rieselsieve.com web site. Do be aware that there are different clients for amd's and intels.