Robert,
This discussion is getting increasingly interesting-
>The largest x86 CPU on the market right now is running at the
>fastest clock speed.
So you are suggesting that clock speed is proportional to die size? Perhaps you should write a paper explaining the phenomenon. I think that it will be well received by the engineering community!
>RAMBUS is going to be very expensive for some time. The MXi will >have a 2GB/sec. memory bandwith using the relatively cheap and
>plentiful SDRAM. Oh yeah, and its die size is 90mmý.
Whew, good thing that SDRAM isn't available to Intel ;^) When can I expect to see "MXi" processors in the marketplace? What will the P55c die size be at that time?
>BTW, why don't you run the buyout idea by the shareholders on the
>Intel thread? It would be interesting to hear what they think of the >idea.
Good point. Intel hasn't bought another CPU design in almost a month now. I'm sure that Andy Grove consults with the SI INTC thread before he makes a move.
With the large amount of money I have invested in Cyrix the last few years, I was hoping that NSM could figure it out first. Are you aware that a large portion of NSM's business is supplying chips for the socket 7 marketplace? Maybe the management of National might want to build an integrated chipset for the Socket7/8/Slot1 marketplace, rather than ceding that business to Cirrus, Intel, Micron, etc., etc. The Cirrus integrated chip is already being sampled and time is slipping away. (I'm writing this note on a Cyrix P200+ computer with Cirrus graphics. Based on the earnings report, I'm guessing that the "robustly sized" CPU was sold by Cyrix below cost.)
I think that a small CPU combined with an integrated chipset will offer the best price/performance during the second half of 1998. Knee-jerk reaction to new ideas is not necessarily a good model for survival.
Scumbria |
