Bill: <In effect he [Vin Dham] has the entire chip in his minds eye, and can they replace him at all?>
Just for the historical recall, the picture might be not so rose as you wish.
First, he was fired from Intel for mistakes that lead to FDIV bug in Pentiums. At least, he was officially responsible for that. He was never listed as Pentium architect or designer, just some manager.
When he "joined" NexGen in 95, the Nex586/686 chip was in design for about 10 years. Therefore, he probably had little idea what the design is about and for.
I can admit that he probably did a serious contribution to K6 design and to bringing the K6 to market. As you might know, the original Nx686 was totally incompatible with any chipset/bus, had no FPU, consumed too much power. Everything had to be redesigned to be Socket7 compatible, for more or less successive market penetration. It is really amusing how AMD could make it in about a year.
Now, in 2Q AMD had shipped 150k K6, and much more in 3Q, with decent frequency yield. Many parts, as it was reported here and on Tom's page, were working stably at 260Mhz. Remember, the FAB was the same, and the process was the same. Therefore, some serious changes must be done to design in summer that resulted in yield drop at high frequencies. What has been changed? Bug fixes? Or something else? Who was responsible for all that? Probably Vin. This means he effectively killed the good working chip and the whole AMD business program since the windows of opportunity is very time-sensitive. In addition, Intel's overreaction and fire sales added to the problem. IMHO, he seriously miscalculated something and got fired for that.
<Where will they go for their counterpart to the slot 1 designer?>
According to last Microprocessor Forum, the K7 is going to be a Slot-1-like with Digital EV6 bus interface. Therefore, this work was long time in progress and seems to be independent from K6-V.Dham project, and I do not see any reason to worry here.
Ali |
