We were trying to calculate how the latest model Cray X1 supercomputer compares to the model sitting on your desk. First I want to confirm some "facts" (more like "assumptions") gleaned from our search.
One "Flop" (F) equals one computational operation;
A typical AthlonXP performs 9 Flops per clock cycle;
A typical Pentium4 performs 6 Flops per clock cycle;
(Forgive my gross oversimplifications)
Ergo a top-of-the-line AthlonXP @ 3.0 GHz performs 27 Gigaflops;
A top-of-the-line P4 @ 3.2 GHz performs 19.2 Gigaflops;
A top-of-the-line Cray X1 performs 52.4 Teraflops (52,400 Gigaflops);
Making the Cray approximately 1,941 times faster than the 3.0 GHz AthlonXP, or approximately 2,729 times faster than a 3.2 GHz P4 (ignoring the function of system memory).
Feel free to punch as many holes as necessary in my assumptions/conclusions. For instance, I don't know if my information on Athlon or Pentium operations is correct (I read the White Papers for the Athlon and the Pentium, and only succeeded in confusing the heck out of myself!).
Thanks for your attention
Kevin
One "Flop" (F) equals one computational operation;
A typical AthlonXP performs 9 Flops per clock cycle;
A typical Pentium4 performs 6 Flops per clock cycle;
(Forgive my gross oversimplifications)
Ergo a top-of-the-line AthlonXP @ 3.0 GHz performs 27 Gigaflops;
A top-of-the-line P4 @ 3.2 GHz performs 19.2 Gigaflops;
A top-of-the-line Cray X1 performs 52.4 Teraflops (52,400 Gigaflops);
Making the Cray approximately 1,941 times faster than the 3.0 GHz AthlonXP, or approximately 2,729 times faster than a 3.2 GHz P4 (ignoring the function of system memory).
Feel free to punch as many holes as necessary in my assumptions/conclusions. For instance, I don't know if my information on Athlon or Pentium operations is correct (I read the White Papers for the Athlon and the Pentium, and only succeeded in confusing the heck out of myself!).
Thanks for your attention
Kevin
Comment