Dear Elmer:
The proof is in the benchmarks and what Intel has done so far. They can't make it run with current software fast. They can't make it without massive bandwidth. They can't make it with current chipsets. If the Athlon did not exist, it would have slipped until they could make it work by trying different things until it did work. But competition forced them to make do and release what they had. Waiting is a luxury they can no longer afford.
Scumbria does design CPUs. You may not like what he says politically but, he does know something about it. He may simply state the prevailing wisdom, but, it has proven more times to be correct. I do know digital electronics, logic, and have much experience in both CPU architecture, from both a hardware and a software view, and the applications and systems that work with them. I can cut through the FUD that tries to claim something that is not true (such as "performance leaps forward when, you run P4 with RDRAM past 2GHz" being pure baloney as FSB bottlenecks become larger (more problematic) with higher multipliers, not less).
IMO, as a designer of systems, it shows bottlenecks in areas it needs to correct. Thus the original design fails, and instead of trying to tweak it, normally one would address the problems by changing the design, instead of simply patching a bad design with quick fixes as it would lead to a larger foul up later. The P3 is at the end of the patch up fixes and needs a process advantage to become competitive again (against Athlon as it is close to the performance of P4 especially on current software).
Pete |
