Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
My first choice would be a dual 2.5GHz G5. If the program can use Altivec and dual procs, it will be fast and accurate. And quiet. Load that sucker full of RAM, you'll be good to go. Or an xServe, but that's louder and not as fast. You can easily cluster either.

An Opteron would work too. But use Linux, not Windows. A P4 or Xeon would probably work, but I don't know if the FP error still exists in them. It was the Pentium that had the problem BTW, not the PowerPC.
 
earthtoandy said:
hmm sounds fishy. sounds like issues with the algorithim rather than the chip. because a chip wouldnt get very far if it couldnt do math since thats what it does

It was a chip problem in the Pentium 90 involving the FDIV instruction (Floating Point Divide) which came to light in late 1993. Under obsure circumstances significant calculation errors would occur. There were software workarounds developed to compensate, for example Windows NT would check the hardware during install and offer to implement a software workaround. I still have gray hairs from dealing with customers who were very concerned over this problem.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.