Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Maxwell who wrote (41660)	11/17/1998 12:34:00 PM
From: Scumbria	Read Replies (2) \| Respond to of 1576894

Maxwell, Although it would be nice to have a K6-2 400MHz for Christmas, my real concern is the yield of 400 MHz parts. I believe that the behavior of AMD stock is intimately linked to the MHz ramp of K6-2 parts. Scumbria

To: Maxwell who wrote (41660)	11/17/1998 4:45:00 PM
From: Tenchusatsu	Respond to of 1576894

Maxwell, I just read the most recent Microprocessor Report (well, not really recent anymore; it only popped up here recently) on the K7. Now I understand your enormous enthusiam over the K7, though I've still got to wonder how much more instruction-level parallelism (ILP) can be extracted from the x86 architecture using P6-like techniques. A couple of points I'd like to bring out: 1) You mentioned one time that the K7's branch predictor makes the P6's look ancient. But the K7's predictor is simpler than that of the K6. The K6's predictor was admitted by AMD to be overkill, and the K7's predictor was probably made simpler in order to allow for higher clock speeds. 2) MDR estimates for the manufacturing cost of K7 is $104 a pop. In comparison, the K6-2 is $47, the PII Deschutes is $60, and the Katmai is $66. I'm assuming that this assumes equal yields across the products. Couple the K7 with off-chip L2 cache, a PII-like daughterboard, and a 200 MHz bus (64-bit wide) which could be tricky to implement, and it seems that the K7 isn't going to be cheap. The initial low volumes of K7 isn't going to help, either. That second point is leading me to believe that the K7 isn't going mainstream anytime soon. It seems that AMD, with help from Compaq, might be targeting the high-end workstation market first, where a uniprocessor K7 system would prove to be one muscular number-cruncher, if it can beat out Tanner and Cascades. The speed at which the K7 is going to be introduced (even June 1999 is pretty fast in my book) means that server solutions aren't going to be available for at least six months after the intro. That puts even more emphasis on the K6-3 to carry AMD's mainstream line into 1999 and beyond. Tenchusatsu