Charles - <What makes people think CuMine can keep up with Athlon once it moves to 0.18.
I ask this question very seriously. Any and all thoughtful answers are appreciated.>
1) Intel's P858 process is good, and ahead of AMD/MOT in ramping.
2) Coppermine design still has room for improvment w.r.t. Fmax, despite what your friend says. K7 probably does hold the advantage in ease of scaleability attributable to the design. However, this prevailing assumption that because Coppermine is bases on mature core, it won't scale any higher, is just flat out wrong. I will grant it can't go on forever, but it isn't done yet.
3) Don't believe .18 for AMD will provide magical performance pop. Believe the race will be won or lost, or kept close, by the front end of the respective processes. Intel admittedly is focusing on transistor performance, and I believe this is the correct tack for .18.
That's about all I can share on the subject. I did scan your response to another poster on this subject. I still don't believe your friend has very good visibility into what's really going on with Cumine / P858. I would have thought he could have clued you in to the cache and memory enhancements, but apparently he either wasn't aware of them, just did not share them to you, or underestimated their impact on performance. And I'm relatively sure that he is underestimating the P858 process roadmap, in terms of what it is capable of in device (transistor) performance.
OK, now I'm done. I believe I have shared in a general way all I can share. Time will tell. I have said for quite some time that I believe the MHz race will be close well into next year. Let's see how it goes. So far, so good.
PB |
