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.

Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: pgerassi who wrote (139139)	7/12/2001 6:49:02 PM
From: Mary Cluney	Read Replies (1) \| Respond to of 186894

Dear Pete: <<< It is you that is dense! Most servers come with less than the full processor complement as well. Many 4 way servers come with one or two CPUs. You always think that servers will have max CPUs and little disk. I have stated that it is the other way around. Most systems come with lots of disk, peripherals, and other things. >>> You want to talk dense? Let's talk dense. When you talk about real world applications, you are talking about systems that have front end, mid tier, and back end stuff (servers). You have to log in and get to the back end to interorgate the huge databases and then complete some transaction. You are right in that most of the volume is in the front end where Intel now dominates. But, you are wrong about the topology. In the front end, you are likely to see in the near future 1U or 2U form factors. In these chassis you are likely to see eight servers and two switches. Each of these servers will have either a PIII or P4 processor, 1 gigabyte of DRAM with ECC protection, dual 10/100 Fast Ethernet NICs and maybe 30 gigabytes hardware storage. This will also include power supplies, fans, and switches in redundant setup. All of those components will be hot-pluggable and hot swappable. These types of servers are, indeed, well suited to the front-end that you like to keep reminding us is so ubiguitous. It's like packing a lot of servers into a dense configuration. No matter how hard you wish it to be so, the silicon part of these servers come out to far more than 2% of the total cost for the server - no matter how you define server and no matter how you wish it otherwise. Now that is dense. <<< $60B in servers means about $1B in CPU chips. At the typical Intel ASP of $150 or so, that is 6 million CPU chips or enough to power 4 million servers>>> You are in denial when you ignore the Xeon servers that have established itsself as the guts of real workstations and servers, the DP and MP servers that use the Intel Xeon processor. That is okay because even Drew Peck couldn't understand this market, but he did admit about three years ago that Intel sells a ton of these chips and make a lot of money doing it. You are only a few years behind Mr. Peck. (Next generation Xeons, Foster, I beleive, are on the way - but I could be wrong). Intel gets about $1k+ for each of these and Fosters will cost a lot more. But, before Mr. Peck gets too comfortable with the Xeons, he has to look over his shoulders over the next thee years and ANALYZE THIS . Intel is targeting the Itanium processor for back-end applications. It will be aimed at that big stuff on the back-end - like, you know, large databases, mechanical CAD, encryption, de-encryption, vector segment technology, and biometric identification algorithms? All this using a choice of Oses, from HPUX, AIX 5L, Linux, or Modesto. <<< Now if Compaq, IBM, and HP sell 100K RISC chips for $1k each (their cost), that still leaves 4.6 million CPU chips for Intel. Still enough to cover all of their servers. Of course that does not mean that OEMs do not pocket $4 to $8B for the CPU modules. >>> Obviously, we are not speaking the same language. I don't know how to say dense, in your language. Mary