Original URL: https://www.theregister.com/1999/11/18/why_intel_coppermines_are_like/

Why Intel Coppermines are like hen's teeth

There's not mushroom for doubt about yields

By Mike Magee

Posted in On-Prem, 18th November 1999 08:20 GMT

Column The mushroom technique is a tried and tested formula in the wonderful world of executive suits. Essentially, it can be expressed in these few words: "Keep'em in the dark and feed them dung". This is what makes many of us mushrooms when faced with the full force of the Intel spin paramedics. A spate of articles on the Web appeared yesterday after local New York newspaper The Wall Street Journal got wind of a series of stories the not-so-venerable Register wrote a day after Intel's Coppermine was introduced. A day after Intel played the fanfare and beat the drums about the advent of Coppermine processors, we discovered that the company would be unable to deliver the full range of Coppermine processors it introduced for some weeks, nay months. We had internal Intel documents to this effect and our representatives here in Blighty, faced with these facts, said: "Yes, we're bang to rights, it's a fair cop guv, and supplies are tight" -- or words very much to that effect. Nevertheless, given the fact that Intel was on the record some months earlier that its Coppermine .18 process would not crank up until the year 2000, we find it rather surprising that more of us mushrooms didn't make a break for fruit status from our mycelium roots a little bit earlier. Here are some facts about Intel Coppermine production, kindly contributed by a reader and which add up to the fact there's not mushroom for .18 process high yields quite yet. He points out that when US spin paramedics say millions of Coppermines will be shipped in this quarter, it is not as impressive as it sounds. As the chip is just over 100 square millimetres, an eight inch silicon wafer holds between 250-275 each, with, say a good yield being 200 chips a wafer. It only needs 10,000 wafers to produce two million Copperminos -- which means that Intel can make that many in just two weeks with only one of the four fabs they have producing the microprocessors. Each fab can make around 5,000 wafers a week, amounting to 60,000 wafers a quarter, if we make the probably mistaken assumption that Intel shuts down for Yule. There's no point asking Intel about yields. Its policy is to never, never, ever talk about how many wafers it produces. It would not warm the cockles of people's hearts if such information became generally available. So what's going on? We understand that there are no process problems with Coppermine production and we're inclined to think that this is a marchitectural rather than an architectural problem. Most of Intel's .25 micron production is still rolling, rolling, rolling, and there's no way it's going to throw the switch to off on those chips until it's ready to transfer all of its production to the .18 micron process. In the meantime, it will continue to sell .25 micron chips to the world+dog. A couple of other loose ends to tie up here. The people in the PR pit who feed us the dung to help the mushroom grow are part of Intel's sales group, and hence belong to the lowerarchy rather than the hierarchy. We've noticed a growing tendency for the US press to start naming these spin paramedics and in Blighty we think this is really a breach of journalistic etiquette. Mike Sullivan, for example, worked in the UK for a while and is a very nice chap, putting the best spin possible on a story in the worst of all possible situations. "Dancing" Dan Francisco is a smooth Intel operator and could probably do with a pay rise too... And hen's teeth? Well, both male and female chicks have them at birth, so they can peck their way out of their eggs. But the tooth gradually disappears. In the case of Coppermine, these little chippy chicks are only just beginning to peck their way out into daylight. Intel's mistake -- and one of rather too many it made in 1999 -- was to shout too loud about Coppermine processors without explaining how its factories work. And our mistake, as journalists, was to believe the dung heaped on us mushrooms. ® On Coppermines, see also Pentium III-800 brought forward to Q1 next year So why is Intel's Coppermine good? Coppermines rise from the channel shallows Intel beset by further Coppermine, chipset delays Intel's Coppermine fails to dampen AMD's Athlon squib Huge shortages, technical problems hit Intel Coppermine On Intel's fabrication process see also Intel is a twelve incher Intel makes The Register sweat Part III