Strange whiff surrounds Intel's great i820 shambles

Dell, Samsung and Intel have their own related reasons for the delay of the i820 Camino chipset, it has emerged. Technical mistakes that were announced just a few days before the release of the i820 and Intel's own Vancouver motherboard are more to do with the chip giant's inability to design working circuit boards and with its relationship with Dell and Samsung, according to well placed sources close to the companies' plans. The problem now seems to boil down to this. The i820 chipset actually works fine. But Dell only uses Intel motherboards in its desktops and technical mistakes made by Chipzilla led to a last minute panic. The Intel mistake is confined only to its motherboards and not to third parties, but the essence of the allegation is that as a result, Santa Clara pressed the stop button on all mobos using the i820 chipset, while it attempted to fix its own problem, calling all third party mobos in for qualification. Intel had to stop the programme for another reason. It, and its major customers including Dell, would look very silly if their Vancouver solution failed to work while the i820 chipset hummed away very nicely on third party motherboards equipped with bright and shiny Coppermine .18 micron chips. There is another complication. Many readers have pointed out that buying Rambus RIMMs is not easy, and that if you can get hold of them, you have to pay quite a lot of money. Our information is that this situation is compounded by the fact that Samsung, which has received large investment from Intel, and is a senior member of the Seven Dramurai, also has its own technical problems with Rambus. Other manufacturers of the RIMMs, even if they are members of this select club, do not have these problems. In the short term, Intel has taken the extraordinary step of advising its smaller PC customers to use either the Via chipset or to use the BX chipset. However, several of these smaller manufacturers have pointed out that the Via chipset does not really fit the bill, while BX chipsets, surprise surprise, still seem to be in terribly short supply. Further, several of these two tier and three tier manufacturers have pointed out to us the strange nature of this advice, which came only a week after Intel screwed up. Had Intel, which is taking legal action against Via anyway, qualified mobos using the chipset before they advised their loyal customers to try it out? So smaller PC companies, as well as the third party motherboard manufacturers, just have to wait until Intel cleans out its own Augean Stables. That will be towards the middle or end of next month, and as we have pointed out earlier, will not leave time for children to get their i820 mobo toys in their Christmas stockings. When AMD was ramping up its K7-Athlon hype machine earlier this year, many speculated that Dell would dabble in the non-Intel waters and try the processor out. But that was always unlikely, given the preferential rates Dell gets from Intel for its processors. Within the industry, Dell is often jokingly described as an Intel distributor. It subscribes wholeheartedly to the Intel Inside campaign and it uses Intel motherboards in its successful range of PCs. Tomorrow, Intel releases its i840 workstation chipset (Carmel), described rather nicely by one online news service as Caramel, perhaps because it hits the sweet spot. But, as the Great Satan of Chips rolls out its Coppermine processors to the world+dog, it is only now become possible to piece together the i820 Camino fiasco and place it in context with the events of the last two months. Placed in context, this is a far greater debacle than the FDIV bug which forced Intel to scrap a heap of Pentiums a while back. The trouble is, that made for easy headlines such as Pentium has a bug, while this i820 story is complex and bound up with Intel's investments in Rambus Ink, Samsung and its relationships with other large memory manufacturers worldwide. We will put these questions to the parties concerned during the course of this coming week. ®