Hi jim kelley; Re RDRAM and graphics, some bull comments:
richard surckla speaks the truth, for once:
DDR seems to work good in grahic processors. For some reason graphic processors with RDRAM don't work well.
#reply-12038375
dmf
Re: ...high bandwith applications like graphics I expect such applications will become more pervasive.
#reply-13248264
In fact, you previously wrote:
In the future RDRAM will be used for graphics applications
#reply-13383509
Given that design engineers design for the future, not for the present or the past, don't you think it odd that RDRAM wouldn't be mentioned in an article selling memory to graphics design engineers? Seems mighty odd to me.
The new device was developed in conjunction with Rambus for use in high-speed video cards and other graphics applications
#reply-13026445
Estephen
I think that is very likely. Rambus first design wins were in video cards and it is the best solution for them. Also, I don't think the X-box will use ddr.
#reply-13239574
The basic fact is that RDRAM had the graphics market, then lost it.
The basic fact is that RDRAM had the PC market, and is now in the process of losing it. The board makers are not going to hurt their current sales by talking about a replacement technology until they have it in hand. The first prototype systems will show up this summer, and when the board houses begin to think that they are getting ready to ship volume, they will begin letting them out to reviewers. Tests will be run comparing them with RDRAM at about the same time. Jdassoc will begin reporting DDR compatible motherboards late this summer. It isn't going to be pretty.
Micron has already shown a DDR motherboard, but Micron is in a different situation than the other motherboard makers. Micron wants to support DDR, so it is letting that board out early, earlier than the other board houses. That is why the board was double size, etc., it was an early prototype. The other guys aren't showing their early prototypes, but you can bet that they have them, and they are fixing bugs and optimizing performance on them even as we speak. Where do you think all those DDR samples are going? Do you guys think that the memory houses just make those so that they can sit in engineer's desks? (Which I have seen happen!)
The reason that this isn't obvious to the majority of you is that you are situated way downstream of the guys designing stuff. Remember how Jdassoc knew that RDRAM systems weren't going to hit Xmas99? He's upstream of the average Joe who goes into a store and sees if there are any RDRAM systems available. Jdassoc is in the industry and he knows where to look to see shipments getting started, and how to interpret what he sees. He thus has about a month or two lead on the average Joe. I'm at the design end, and that leads the Jdassoc stage by 9 to 15 months. I'm seeing RDRAM not being designed into new designs. The consequences of this won't be seen until about 9-18 months after I began stating that RDRAM was "dead, dead, dead" as the designs slowly complete and go to prototype and production. Go look up my post and start counting off the months, it's a comin!
Designers decide on the memory type to use in a system very early in the process. After choosing, it is very hard to convert to a radically different type. Design is a long process, it takes a long time to get a product out for testing. Designers quit designing in RDRAM around September of last year. Jdassoc won't be aware of this until those new DDR designs start to hit the pipe late this year. Until then, it will appear that RDRAM is doing fine. But by the end of the summer the truth will be impossible to deny, and that will be caused by the coming flurry of prototype DDR reviews, around beginning to mid summer. We've already seen the first of these, there are a huge number more coming.
-- Carl |
