Dear Tony:
Until it shows otherwise, it is over-hyped and under-performing. No benchmarks have been shown (by third parties). I am awaiting postings from you showing both x86 and native performance. I will put more creedence on benchs where the algorithm used is like linpack, Moldyn, QMC, or other similar workstation type software, compiled by others (although since it needs an optimized compiler, I will allow one from Intel here) from C (or Fortran) source. This is to prevent use of hand coded assembly which, is not done for others like PIII, RS/6000, PA-RISC, Alpha, SPARC, or Athlon. This allows at least a chance of an apples to apples comparison. The only rumors I can even lend a smidgen of notice to, have stated that the current samples are 500MHZ or 600MHz and performance is underwhelming.
Thus, talk is cheap and the proof is missing (and dear).
Pete |
