Re: in a wringer with this 820 thing? ...
Intel has gotten the 820 to work, at least with 2 RIMMS (from everything published).
Here's a guess about what may be happening. The problem is Rambus cost. Yields are half of SDRAM and the chips are twice as big - then it costs more to test and package them. As a result, it costs an enduser $1,000 for a 128 meg Rambus module when the same size PC133 module is $225. Rambus edges out PC133 in some applications, but not nearly enough to make a 64 meg Rambus computer competitive with a 128 meg PC133 computer - which would still leave the PC133 system a lot cheaper, so that approach won't work.
$750 is a big delta when the price of a middle to upper end PC is typically $1750 to $2500.
Intel, Dell, and some others are willing to subsidize that $750 dollar difference on a few machines just to maintain the confidence of the markets - that were promised reasonably priced, high performance rambus machines. But nobody wants to subsidize millions of the things, which would cost billions of dollars.
So expect a limited number of high profile shipments at subsidized prices, then a steady de-emphasis on rambus as real prices are rapidly put into place to keep losses down.
The hope from rambus to be down to double the price by about mid 2000, and a 50% premium by the end of next year, from 4 times the price as it is now, but by then DDR will be all over the place. DDR should be quite a bit faster than rambus, but cost about the same as PC133.
The name of the game now is to get out without undermining investor and consumer confidence in Intel. Carefully handled, it shouldn't be too difficult or expensive.
Dan |
