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 : CYRIX / NSM -- Ignore unavailable to you. Want to Upgrade?

To: Steve Porter who wrote (27163)	6/7/1998 3:31:00 PM
From: Paul Engel	Read Replies (2) \| Respond to of 33344

Steve - Re: " your thoughts on a couple of things (namely the $$ of the L2 cache in a 450Pii vs a 233Pii and secondly Cyrix slot 1)" With just a little checking around, I found a price for IBM static ram cache memory for their new 750 processor on Pricewatch. This was $35 for a full 512 KiloBytes. From a performance standpoint, I will "assume" that this is comparable to the Intel Pentium II requirements at 400 MHz (current top of the line). Now, the $35 price tag is a retail price on small volumes so let us say that Intel, which buys these wholesale in huge volumes, gets a 25% discount - or at a price of $26.25. This is probably $10 more than the approximate $15 they were paying for the older and slower 233 MHz Pentium II 512 KB cache rams. Bear in mind, the new Pentium II/400 MHz die is much smaller than the original (131 vs 203) and the ASP's are quite a bit higher due to their much higher performance. Overall, the $16.25 cost increase in L2 cache will be offset by reduced silicon costs for the CPU chip and higher ASPs. As Intel transitions to the XEON with full speed L2 cache, the 400/450 MHz SRAM chips are made internally by Intel. This die is 221 sq. mm and is made on Intel's 0.35 micron process. The device is a single chip 512 KiloBYTE chip with on board TAG Ram. This process is yielding approximately 75% good die per wafer and at $2000/wafer (an assumed UPPER BOUND FOR Si costs on this older process), Intel should make these for $22 in silicon costs. Throw in assembly, test and miscellaneous costs and Intel will more than make up for this with the very high XEON/Slot 2 selling prices of over $1000 for the 512 KiloBYTE version. As for Cyrix and Slot 1, let me say the following. If you happen to have cable TV and receive CNBC, nearly every time they discuss the impending bankruptcy of Intel and the "fact" that Cyrix and AMD are eating Intel's lunch in low end CPUs or the FTC is about to pad lock Intel's front doors, CNBC often plays footage of Intel's manufacturing/prototype lines for the Slot 1 Pentium IIs. Watch that carefully. Look at all the miscellaneous equipment and processes required to manufacture a Slot 1 device. It is not a simple matter of throwing a CPU chip on a PC board and stuffing it into a black plastic case. The infrastructure required to tool up for that operation will exceed CYRIX's current revenues. Automatic silk screening of solder paste onto PC boards, automatic feed and placement of zillions of chip capacitors and resistors, solder reflow furnaces with accurate temperature profiles and automatic load/unload robotics at each end, automatic insertion machines for PC board/cartridge assembly and fastening, overall computer control of the process floor to keep this in fine tuned -process control, etc. Intel has built three new factories recently to handle this - outside of Manila in Cavite, Phillipines, San Jose (Cost Rica) and in Leixlip, Ireland. Intel has set into place the manufacturing and automation infrastructure to build 100,000,000 Slot 1 and 2 cartridges per year and to grow from that base. All the "talk" out of Cyrix regarding a SLot 1 will fall a lot short of the enormous back-end operations required to build, assemble and test a Slot 1 cartridge. NSM will be swamped when they evaluate the enormity of the size, complexity and COST of this type of operation. On top of all that, you have the designers who have to design the chip without detailed information on the bus operation because Intel regards this as a trade secret. Ergo, Cyrix can't copy the details from patent disclosure because many of the details aren't patented. The P6 bus itself runs with a special GTL+ electrical interface which must also be reverse engineered. Couple these "minor details" with the fact that Intel continues to evolve their technology and architecture with Katmai New Instructions and you have a Cyrix rushing around and spending money (that they don't have) like a drunken sailor trying to catch up with Intel's technology from 1997. And NSM is not a company identified with extravagant spending. When the losses get reported in the next month or so, Brian Halla is going to get the Wake-Up Call of his life - and Mother Intel won't be in any mood to lend him a hand. Paul

To: Steve Porter who wrote (27163)	6/7/1998 3:49:00 PM
From: Richard Wang	Respond to of 33344

Steve, Pravin, Jim, Fuchi, Scumbria, Bobs, and others, news.com preparing its case, the FTC staff has looked at the company's relationship with Alabama computer maker Integraph and its sometimes convoluted relations with Digital Equipment, Compaq Computer, Acer Computer, and Packard Bell, according to industry sources. I am certainly no legal expert, but I can see something for Cyrix and AMD in this. If FTC can win its claim that the OEMs like Compaq were deterred from using 2nd party chips, the maker of these chips, ie, Cyrix and AMD, can then claim damage. They may claim that the predatory behavior of Intel prevented the OEMs from buy their chips. For those of us who have been following the three companies for a number years, we have repeatedly witnessed such strong hand behavior of Intel. These claims could involve big time $s. Richard