Wanna-bmw Re.. I'm sure you realize that if Intel wanted to show one processor generation as having an advantage over another one, that they can easily create any environment to give them the favorable benchmark results<<<<<<<<
I doubt if even Elmer will argue with that one. All companies try to show their products off in a good light.
With a haphazard mix of SPEC95 + other synthetic tests (like what is used in iCOMP), they can probably show the Pentium 4 to be as proficient over the Pentium III today, as the Pentium II was over the Pentium MMX several years ago. The results aren't "inaccurate", but they are certainly skewed to prove a point... it's a high score amongst a graph with many data points.<<<<<<<
Agreed again. Everyone here knows Intel fixes the benches to show their processors off in the light Intc wishes to showcase. But Milo didn't fix the benches. He used an existing bench which was common to all three processors and was optimized for all three, to show running comparisons between the three.
Milo's point I believe isn't that you can make up benches, but that when you apply the existing benches to all three processors, indeed, compare all of the past Intel upgrades, that up until P4, Intel improved the IPC for each generation, except the P4. Milo, I believe was talking about all benches, and used ICOMP as an example. , such that a 100% mhz increase also increased the performance,100%; instead of the normal 70% performance increase one would expect when you increase MHZ only. I have no doubt you can make up special benches to disprove that; however, Milo didn't make up the benches to make a point. Instead of trying to sidetrack Milo's point, why don't you argue the point, and prove that the IPC of the P4 improved; or at least, stayed the same; as previous Intel upgrades had done.. |
