Hi all; I'm a little late with the EE-Times news, and there's a lot of it, particularly about Rambus:
I finally got (2) samples of (Micron) DDR SDRAMs. My board takes four, so I hope Micron will come up with two more, and some of the other vendors (Samsung, IBM, etc.) will cough some up. No big rush, I won't need them for another couple months. I hear that Rambus samples are still hard to get as well. Anyway, the sync-linc consortium (which was a much more complicated interface than DDR) folded, and is throwing its support behind DDR. So it is Intel versus a good portion of the rest of the industry. This is not looking pretty for Intel:
Double-data-rate SDRAM picks up steam vs. Rambus
The field of organized initiatives pursuing alternatives to the Intel-backed Rambus DRAM has shrunk by one. A decision by the SLDRAM Inc. consortium to wrap up its own work and throw its support behind the emerging DDR II spec cleaves the DRAM industry into two major camps, amassed along the boundary between desktop and server: the Rambus approach, initially aimed at desktop PCs, and the server-bound double-data-rate SDRAM.
techweb.com
Like I've said before, Intel's decision to support only a Rambus memory interface on its new processor was magnificently stupid. It really stuns me that any professional who has been in the memory industry for a decade or so would do something that silly. This is a major disaster for Intel, and one which is not understood by most analysts. Notice that the marketting segmentation between Rambus and DDR mentioned in the above quote gives Rambus (and therefore Intel) the desktop market, and DDR the server market. Rambus is best for the very low end consumer machines, but that is where Intel has been losing market share drastically, and that is also a market where competition based on memory price is very significant. Those low-end machines have plenty of spare pins to get bandwidth out of. No need to have to use a very high bandwidth per pin device like Rambus, if the cost increase per added pin is low.
Here is Rambus losing more design wins to SDRAM:
Rambus delay spurs ad hoc effort in U.S., Taiwan -- Support grows for PC133 DRAM alternative
A groundswell of support for the PC133 DRAM specification is building in Taiwan and the United States. With Intel Corp. otherwise occupied with the difficult transition to the Rambus memory architecture, a growing number of memory and chip-set companies are working to establish PC133 as an alternative in the marketplace.
...
Some memory manufacturers, however, question whether the cost of shifting to Rambus can be justified, particularly if system OEMs avoid using the 600-MHz Rambus "speed bin." Too many Rambus parts are not able to hit the full 800-MHz specification, opponents point out, and DRAM makers see no ready market for the slower Rambus parts, which would deliver little or no performance gain over the 133-MHz SDRAMs.
techweb.com
The consequence of Rambus's inability to sell the sub-800Mhz parts is a low "marketable" yield. This is basically going to run the price of (800Mhz) Rambus memory chips way high, as they will get fewer of the high price chips per wafer. This is going to really give Intel's competitors a massive edge: Their computer chips will be able to get the same performance as Rambus, but with much cheaper memory chips. Say goodbye to more Intel market share...
Another memory maker has trouble making Rambus parts. I tell you, this has all the makings of a major disaster for Intel:
Fujitsu shifts DRAM focus to stress consumer apps
One issue that promised to bring a measure of margins back into the DRAM business but which is proving thorny is Direct Rambus DRAMs. Taguchi said tester and packaging problems are squeezing availability of the new Direct Rambus DRAMs, and the issue could delay a robust ramp for the devices until next year.
techweb.com
The LCD makers are hoping that they don't see prices slide any further. But if they do, we may observe the beginning of the end of the electron gun display device. It's been a great 50 years:
LCD monitor firms strive to hold line at $999
If 1999 were to follow in the footsteps of 1998, today's $999 15-inch LCD monitor would hit the "magic" $500 price point at the end of the year and CRTs would start heading toward the trash heap of display history. But that's not going to happen. In fact, the liquid-crystal display will barely make a dent in the CRT's monitor market share.
techweb.com
More trend towards cheaper computers. One of the last reasons for a consumer to buy a high-end processor is games (Yeah!). But the big guys are getting ready to run that market down to very low prices with machines designed to give the gamer what he needs:
New Celeron module gets embedded nod
The motherboards support 100-MHz SDRAMs, extensive hardware-monitoring functions and features geared for the PC-games enthusiast.
techweb.com
-- Carl |
