Hi Jdassoc; I wouldn't think it would be much more difficult to get two RDRAM channels going than to get one channel. So if/when Intel solves the Camino problem they will probably solve the other.
If someone stuck a gun to my head, and asked me to use RDRAM, I would refuse. But if I heard them pull back the trigger, I would comply. My suggestion is eliminate the RIMM modules completely, and don't let the consumer touch any part on those 800MHz transmission lines. Instead, just have all the memory soldered into the mother board.
This would put a certain amount of a crimp on the way computers are sold today, but this is not completely unheard of. It would just be the case where a certain amount of memory is available with the motherboard, and the user can then add DIMM sockets to extend memory. The expansion memory would be PC1xx, of course.
That plan would allow INTC to claim a partial victory, get rid of a certain amount of RDRAM chips, and have a system that would work more reliably. Like I've said before. If you think an RDRAM system is going to be unreliable when it walks out the door, wait till you see it after the consumer has tried to field upgrade the memory. A slightly bent insertion, and that system is going to be history.
I should mention that one of the great advances in socket technology was the addition of double "wipes", or contacts. This way, if one wipe failes, the other still provides electrical contact. My guess is that RDRAM would have trouble working with a pin with only one wipe active, as this does change the impedance of the socket. In addition, oxides are a problem, though INTC has gold plated everything.
At least all that gold plate will provide relatively high scrap values. BWAHAHAHAHAHA
-- Carl |
