Re: True, but prior to P6, x86 were really only the low end of the servers, and it was P6 that allowed x86 to take over the mainstream of the server market.
Not in the beginning. It took years to promote PPro as a viable alternative in the enterprise. Intel couldn't do it overnight. And versus RISC, they had the benefit of huge economies with inexpensive Linux operating systems, vastly less expensive hardware, and huge 3rd party infrastructure. These benefits don't exactly translate into something Intel can use to compete against AMD, because AMD uses the same infrastructure and is supported by the same OS's. The best Intel can do competitively is raise the performance bar and continue to offer features that increase RASUM. I feel that if all else is equal, Intel has the marketing muscle, the OEM relationships, and the manufacturing and supply chain to do better.
Re: On balance, I think Pii and Piii were ahead in absolute performance vs. K6. And it went head to head vs. K7 (Coppermive vs. Athlon classic) for some time in the very important Integer centered apps while clock speeds were roughly on par (700 to 800 MHz). Tbird gained clear leadership when Piii could not keep up with clock speeds, and when Tbird features started piling on - on die L2 (to reach parity in that department), higher bandwidth, PC-133, DDR-2100. But at that point, Piii was eol.
That's about how I remember it, but performance was close and AMD occasionally swung in front, at least for brief moments between product releases. Even so, the mind set was different back then. AMD could have equal performance and still be perceived as being behind. I don't think the same will be true these days. Intel will need to swing ahead, at least for some time, in order to change the perception.
Re: I am not expecting miracles either, but sometimes, something that seems within grasp is not really there yet. Case in point Mentocito. When was the first time Intel flashed a waver with Montecito die on it? It seems like more than a year ago. It may be another year before full featured Montecito (2 GHz, 24MB L3) arrives.
I don't know what Montecito's problem is, or even if the Inq has this one correct. But if they are, then they are also saying that the underlying cause is due to Intel moving engineers off of IPF projects and back on to x86. If this is their future strategy, then it doesn't surprise me that IPF projects will continue to fail. But this is all speculation, and I am skeptical that the picture painted by the Inq is completely accurate. In terms of Yonah and Merom, I have much higher confidence.
Market Data 
Markets 
AI
