Kap, Re: "One problem, according to sources close to the matter, is that when the internal chip frequency hits 6GHz, as it does with the 3.06GHz because the adders are double clocked to get the work done in a single pipeline cycle, Intel will have to eventually re-engineer the design of the P4 to keep it viable. But such a task is of such magnitude that to achieve the frequencies the firm is projecting in 2005 or so, it is probably more realistic to start another design rather than re-engineer all the innards of the processor."
These "sources close to the matter" wouldn't be from the SI moderated AMD thread, would they? Last time I visited the "resident armchair experts" there, I saw The Watson Youth claiming that Intel couldn't get to 3GHz without selling "garbage" parts.
Message 17144921
Ali Chen, your "other resident genius", used to claim that higher frequency Northwood parts would dissipate 90-120W of leakage based power.
Message 17243632
Now, is Magee taking their advice? Who is his "reliable source" this time? I don't think he'll tell.
Meanwhile, like richanfamous said, Intel is still 1GHz ahead of AMD, and doing extremely well on all the benchmarks. Clearly, double clocked ALUs haven't been a limiter so far. As for future roadmaps, that depends on which ones Magee is reading - they do tend to change frequently, in case you haven't noticed. As far as my opinion, maybe Intel could have done better than a double clocked ALU. Maybe it wasn't the "brightest idea in the history of MPU design". But on the other hand, neither was it the "dumbest idea in the history of MPU design". What are you trying to prove with all the melodrama? Was the double clocked ALUs one of the things you disagreed with before leaving Intel?
wbmw |
