Wanna: Look at the strides the Pentium 4 has already made. It is already pushing faster frequencies than any other CPU that exists, and it's performance already beats the best thing AMD has to offer (although this may be challenged with AMD's next round of products).
It's a pleasure to see the world your Intel-rosey-glasses again - almost as much fun as Niceguys same ;-)
Seriously…
We agree that P4 performance surpasses Athlon performance (2GHz P4 vs 1.4GHz Tbird), but if the market price was determined based solely on performance, the P4 would be dead in the water. Remember, P4 die size is almost twice that of Tbird.
Of course, I presume at least some of that is due to "dark transistors" from HyperThreading.
As the platform matures further, this advantage will only increase. With extra cache, higher frequencies on .13u, 533MHz front side bus, and more software optimizations (all coming next year), I can't see how you can call it a crock.
I think you are vastly overestimated the "software optimizations". You need to remember that virtually no one uses Intel's compilers - they use Microsoft's (and often not even the latest ones!). Yes, benchmarks will be better optimized next year, but for all intents and purposes, software people use… won't. I'll grant you that there will be a few exceptions - I fully expect Intel will optimize a few filters here and there (which may or may not come with the software when you buy it)… and the like. But little more.
With regards to Hammer, there's little doubt that AMD is in deep doodoo if it's late (or doesn't perform). How great the damage will be depends on how late it is, of course.
The same can be said of SOI. Since this seems to be where AMD has put the majority of its research focus, the .13µ bulk Si process likely isn't going to be good enough in the long run. It's probably only enough to keep AMD from falling helplessly behind Intel until SOI, but only time will actually tell.
-fyo |
