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: kash johal who wrote (57220)	5/4/1999 10:35:00 PM
From: Gary Ng	Read Replies (2) \| Respond to of 1572369

kash, Re: Why do you have such a problem with this picture. I don't know about Saturn V, to me, it sounds too good to be true :-) Gary

To: kash johal who wrote (57220)	5/5/1999 3:20:00 PM
From: Saturn V	Read Replies (2) \| Respond to of 1572369

Ref- "Why do you have a problem with this picture" The K-6/Celoron Scenario will be repeated for the following reasons. IT IS HIGHLY UNLIKELY THAT THE K-7 WILL BE PERCEIVED AS EQUIVALENT TO INTEL'S BEST PRODUCT. A. The K-7 will only be marginally superior to the PII at the the same clock speed. The K-7 CPU is a significant evolutionary enhancement over the K-6/Pentium II.It has 3 ALU units allowing an integer operation in 0.33 clocks as opposed to 0.5 clocks. The Dual FPU will allow a floating point calculation in 1.5 clocks as opposed to 3 clocks. Unfortunately for most work the CPU is not the bottleneck. A L2 cache miss gives a penalty of 30-60 clocks. So the more elegant CPU microarchitecture is for naught. It will give a marginal improvement at the same clock rate for some floating point intensive benchmarks. I asserted this in December 98, and rumored K-7 benchmarks fit my beliefs. B. The K-7 will be perceived as inferior to PIII. The SSE instructions can reduce the incidence of L2 cache miss, and so on benchmarks written for the PIII, the K-7 will be toast[ eg Photoshop] The intense Intel media campaign will further feed this perception, and given the large amount of announced software for SSE, the K-7 will be fighting an uphill battle, and may end up as perceived as only equivalent to the existing PII/Celeron. C. The Intel 0.18 micron process has already shipped production quantities of Dixon. So the Coppermine will have a smoothly running process.[ The Coppermine is the cache Celeron equivalent of PIII]. If this design ramps up as smoothly as the cache version Celeron, Intel will be shipping a PIII at 700MHz by fall with an on chip cache. AMD is at least 6 months behind production of 0.18micron. D. The new infrastructure for K-7 will have have the inevitable start-up technical glitches compounding AMD's problems. It is very unlikely that AMD will make significant penetration in the server marketplace for a year or so because of the new infrastructure. THE UNIT COST [ VARIABLE COST ] FOR THE K-6 WILL BE HIGHER THAN FOR COPPERMINE A. The K-7 will have a cartridge and off chip cache. The Coppermine will not. Add $20-40 to AMD's cost. B. Intel has leveraged its economy of scale, is further along the learning curve and will continue to have lower chip costs than AMD for the immediate future. C. The new K-7 infrastructure will inevitably be more expensive because of lower volumes and being behind on the learning curve. THUS THE K-6/CELERON BATLLE GETS REPLAYED. SINCE THE K-7 WILL BE PERCEIVED AS INFERIOR, AMD IS FORCED TO CUT PRICES OF K-7 AND ITS SUPPORTING INFRASTRUCTURE SIGNIFICANTLY BELOW THAT OF INTEL. UNFORTUNATELY INTEL's COSTS ARE A LOT LOWER, ALLOWING IT LATITUDE IN SETTING PRICES AT A LEVEL WHERE IF AMD GETS ANY SIGNIFICANT MARKET SHARE, AMD WILL BLEED TO DEATH.