.
I might soon replace my old desktop computer.
I want Fritz or Rybka to analyze more quickly.
In the past, when I have purchased multi-CPU computer at work, I have found that each of the multi-CPUs is individually slower than the one CPU in my other new computer.
Generally two slower CPUs is slower overall than is one faster CPU; for most practical uses.
** Does anyone have experience to answer whether the multi-CPU versions of Deep Fritz or Rybka are really overall faster, in the practical realm?
. .
Without commenting specifically on Fritz or Rybka, the reason the many programs are a bit slower on multicore hardware is that most programs do not use more than one of the cores at a time. Efficient multiprocessor programming is hard. For the usual mix of applications, multicore systems only usually shine when running multiple applications at once (and assuming that the multiple applications aren’t all trying to use the hard disk at the same time).
Programs like the multi-CPU versions of Fritz and Rybka are specifically written for multiprocessor/multicore use, and, unlike Excel or Word, they are reportedly significantly faster (i.e., more positions and nodes per second) than the single processor versions – however, Deep Whatever running on a four core system will only be about 3 to 3.2 times faster than running on a single processor of the same speed – again, efficient multiprocessor programming is difficult.
Bottom line: Yes, but the more cores you give rybka, you get less out of each core. Right now, there is a cluster Rybka going on, and thats pretty good play, but its also very expensive. But for the average joe, your probably not going to see much difference between say, a 4 core i7 and a 6 core i7 crunching on the problem.
The 6 core should statistically play better, but you’d have to play a lot of games to see the difference. You’d probably see the biggest difference in blitz games, and in problem positions in which the computer has to think for a long time to get the correct line of play.