You guys will just have to visit the RMBS thread. I can't keep this up. :8)
Monday, Jul 31, 2000 10:33 PM ET
To: jim kelley who wrote ( 48565)
From: NightOwl
That's one way of looking at it.
But I see Sony's decision to go with DRDRAM in the face of DDR's capability and cost advantage as based on two factors:
1. The willingness of RMBS/INTC to fund development costs to date through cash payments and equity distributions rather than passing those costs through the market chain; and
2. Sony's institutional memory of the VHS/Beta Max war which they lost some time ago.
As I recall they went with the better technology there and got trounced because the "best technology" wasn't the generally accepted industry standard. At the time of the design win for PS2, INTC was talking and acting as though DRDRAM would be the next industry standard.
Sony took that bate and now INTC appears to be switching. If INTC's reconsideration of first PC133 and then DDR had been made, open and above board, back in 1998 instead of in July, 2000, I doubt seriously that Sony would be pushing a DRDRAM based com server or anything else today.
...But on the bright side, these events are pressuring the development of DDR2. So I won't complain too loudly.
But another thing I don't understand is INTC's switch on the previous DRDRAM only future.
With memory bandwidth expanding at the current rate, as I understand it won't be long before a 256 bit wide FSB of the CPU is unable to handle a full speed exchange between the CPU core and memory. INTC must be considering a method of addressing this future bottleneck.
If they are planning to attack the problem by making the bus wider, then I suppose DDR type memory makes sense. But if they could find a way to employ the RMBS bus IP for the FSB itself, wouldn't they be doing that now?
I mean they are the ones who said back in 1997 that DRDRAM was needed to keep up with the then future speeds of the CPU clock. Wouldn't you think that a simple "elegant" solution to their and DRDRAM's problems would have been to build the CPU's FSB to the narrower, "streaming friendly", and "high speed" RMBS specs?
They've been dealing with RMBS since at least '95. But I still haven't heard an announcement of a Rambus interface for the CPU in any subsequent product or product announcement.
Doesn't this strike you as a little peculiar?
0|0 |
