SI
SI
discoversearch

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)
INTC 35.10+2.3%Nov 19 3:59 PM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Cycleman who wrote (43904)1/5/1998 1:42:00 PM
From: Paul Engel  Read Replies (2) of 186894
 
Geoffrey - Re: " I believe Intel is striving hard to technology-distance themselves from the field, and thereby put an AMD at both a technology and manufacturing disadvantage."

Intel's circuit designs are their own. Their existing CPU architecture = the x86 = can be duplicated and has in most, BUT NOT ALL, respects. As you note, AMD and Cyrix compete with Intel.

But how do they compete?

Intel continues to EXTEND the x86 architecture with Dual Independent Bus architectures, multiple CPU support in SMP mode built in, etc. These extensions have been more carefully safeguarded via patents and trade secrets and Intel is maintaining this HIGH END advantage over its competitors who cannot duplicate these new functions.

From a process standpoint, Intel has excellent superiority over its competitors. Intel began production of 0.35 micron Pentium chips in early 1995 and runs this process in 5 or 6 wafer fabs at extremely high yields.

Further, Intel began to ramp 0.25 micron processes in 2 fabs last year and 2 more fabs are bringing up this process now.

AMD, on the other hand, had to purchase/copy their technology from IBM and to this day does not have a fully debugged 0.35 micron process. They are trying to bring up a 0.25 micron process amid yield problems on the 0.35 micron process in their ONE AND ONLY PRODUCTION WAFER FAB process.

Cyrix is at the mercy of IBM and IBM seems to keep offering slightly "old" technology to Cyrix. Although their 6x86MX process is at 0.35 microns, it cannot achieve the speeds of Intel chips - limited currently to less than 200 MHz operation (for get what their "P" rating is. However, Cyrix does get a lot of performance out of these lower speed parts.

And now NSM owns Cyrix - so in the future, NSM will have to supprt the voracious fab appetite of these new processors. And just what is NSM's fab situation? They are only now building their 8 inch wafer fab in Maine to support 0.25 micron processes - they effectively are starting from scratch - 3,000 miles away from home!

Don't forget - AMD had to buy NexGEN to get a workable design - the K6. And now that team is beginning to break up - witness the departure of Vin Dham, their design leader. AMD may have run put of designs to "purchase".

Form a design standpoint, look at any comparison of Intel's FPU performance and MMX performance vs. AMD's K6 and Cyrix's 6x86MX - Intel is generally 25% t 50% faster. Some people claim that this is not useful. If speed is not important, then NONE OF THESE chips are useful! They are bought for speed! People don't buy cars because 2 out of the 4 wheels can go >80 miles per hour! They want ALL the wheels to go at that speed!

Further, in order for AMD to compete with Intel's process technology, they (AMD) have to run their processors at much higher voltages - 3.3 volts for a K6 233 MHz whereas Intel can deliver 300 MHz Pentium IIs running at 2.8 volts - both using 0.35 micron processes.

For 0.25 micron processes, AMD will require 2.2 volts to achieve 266 MHz K6 speeds whereas Intel will be producing 350 and 400 MHz Deschutes CPUs running at 1.8 volts.

Intel's processes are simply higher performing. AMD's technology is more dense, but as their yields have shown, it is much less manufacturable.

Intel has more development personnel and a strict regimen for process development, debug and transfer. AMD, on the other hand is on a treadmill trying to patch-up last year's process while making a stab at debugging this year's process. Their production and development lines are intertwined and it is not clear if K6 production parts manufactured this week use the same process that was used last week!

On the architecture front, Intel is extending the x86 chips with Deschutes - a faster Pentium II with 8-way SMP support and capable of handling full-procssor speed L2 cache at speeds up to 450 MHz. Intel will be revealing this device early next month at the ISSCC and products based on these will be available within one or two months after that conference.

Intel has been producing 1.8 volt Pentium MMX devices for the notebook market since September of last year - and will introduce a 266 MHz version in 7 more days.

Intel's Katmai device will extend the MMX instruction set for Pentium II/Deschutes devices and this will be available late this year.

Intel will be introducing 100 MHz 440BX chip sets to run these Deschutes chips at ever higher speeds and possibly as fast as 133 MHz before the end of the year. A new chip set - the 450NX, will support similar speeds along with 4 and 8 -way SMP processor support for server applications.

Intel's i740 3D Graphics chips supporting AGP wil also be introduced in the next month or so.

And then there is MERCED - BIG SILICON! This extension - and yet a DEPARTURE of the x86 architecture - will be the platform for performance machines for the next decade. Servers, SUPER SERVERS, workstations, etc., will use this as the common computing platform for the foreseeable future - a landmark situation.

Intel has convinced not only their customers to endorse MERCED - Compaq, Dell, Gateway, Sequent, HP and IBM, but also THEIR COMPETITORS - SUN Microsystems, SGI, IBM, HP (IBM are customers and competitors), and DEC .

This common industry architecture will revamp the Server and Mainframe business in years to come - one near-universal hardware platform endorsed and supported by all computer and software manufactures. From the software standpoint, Microsoft, SAP, Sun Software (Solaris), Oracle, SCo, etc., will be producing software to take advantage of this platform.

From a technology standpoint, the MERCED will be built on a 0.18 micron process - at a time when most manufacturers will be getting the bugs out of their 0.25 micron process! Look for 20+ million transistors on that chip.

To be sure, AMD and Cyrix are attempting to keep up - and they will find some areas of success to exploit. But will they be able to offer the depth and breadth of products that Intel offers? Not in our lifetime!

And through all this past year of "New Competition", who has made money? Cyrix - losses in 3 out of the last 4 quarters. AMD - Losses in 4 out of the last 6 quarters. ANd Intel - >1 Billion DOllars/Quarter Profit (with a capital P) for the past SIX QUARTERS.

And that profit money is being re-invested in new fabs, new process technology and new designs and new architectures and new marketing - brand awareness! (Is it Intel? No, not exactly!)

These chickens will be coming home to roost in the not too distant future.

Paul
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext