SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?


To: TGPTNDR who wrote (70958)2/7/2002 6:15:35 PM
From: Joe NYCRespond to of 275872
 
TG,

I made my comment about Duron based on the assumption that there would not be any new speed grades coming out of Austin. As long as Austin can produce them, by all means, AMD should continue.

I don't think it would be a good idea to start making Morgan Durons in Dresden (except mobile).

Joe



To: TGPTNDR who wrote (70958)2/7/2002 6:15:37 PM
From: kapkan4uRead Replies (5) | Respond to of 275872
 
Floating-body SOI DRAM. This could probably be used to build huge L3 caches in future generations of Hammer.

eet.com

Toshiba cuts capacitor from DRAM cell design
By Yoshiko Hara

EE Times
February 7, 2002 (2:40 p.m. EST)


TOKYO — Toshiba Corp. has developed a one-transistor, no-capacitor cell structure that it claims solves the difficulties encountered in producing DRAMs in sub-0.1-micron process technology. The company presented the approach at the International Solid-State Circuits Conference in San Francisco this week.

A conventional DRAM cell consists of one transistor and one capacitor. But when a DRAM is processed at below 0.1 micron, this structure becomes a bottleneck. Even if the transistor shrinks, the capacitor has to maintain a certain capacity. In order to do so, the capacitor has to be formed deeper and deeper in a trench-cell structure, or stacked higher and higher in a stacked structure.

"To solve this bottleneck, new approaches have been proposed, but those approaches require new materials or complex cell structures," said Takashi Ohsawa, senior specialist of the advanced memory design group at Toshiba Memory Division. "The cell structure named floating-body cell that we've developed has a simple structure in the smallest size of 4F2, does not require new material and is scalable, though it is not nonvolatile," Ohsawa said.

The floating-body cell (FBC) is formed on a silicon-on-insulator (SOI) wafer and consists of one MOSFET, whose body is electrically floating. Making use of the floating body, a charge (holes for nMOS) is stored in and drawn out from the body, which functions like the capacitor in a conventional DRAM cell.

"To design circuitry on SOI, a new design was needed because of this floating body, which was a kind of barrier to shift SOI. On the contrary, the FBC makes use of the floating body. The FBC is quite a suitable cell structure for SoC [system-on-chip] on SOI," Ohsawa said.

FBC is an nMOS transistor fabricated on an SOI substrate, which consists of a p-substrate, an n+ diffusion layer, a buried oxide layer and a silicon layer on the top. The source of the MOS transistor formed in the top silicon layer is tied to a grounded source line, the drain to a bit line and the gate to a word line. Polysilicon pillars standing inside the shallow trench isolation on the n+ diffusion layer are biased at a minus voltage to make holes in the bodies. Those pillars serve as stabilizing capacitors coupled to the body to enhance the signal and extend retention time.

With this structure, a signal threshold of 250 millivolts is observed without a capacitor. In the future, it may be possible to take away the pillars and the n+ diffusion layer to make the structure simpler, Ohsawa said.

The FBC cell showed a data retention time of several seconds at a temperature of 30°C and 100 milliseconds at 85°C. "This figure is not satisfactory for the requirement of a standalone DRAM, but enough for embedded DRAM requirements," Ohsawa said. "For general-purpose DRAMs, the retention time needs to be more than one second at 85°C. We are working on it and have a strategy to solve this problem," he said.

The first product that Toshiba engineers are targeting is a system-on-chip device that integrates the FBC DRAM using a 70-nanometer (0.07-micron) process. The first product will have several hundred Mbit FBC DRAMs embedded with logic and will be offered in 2004.



To: TGPTNDR who wrote (70958)2/7/2002 6:16:35 PM
From: ptannerRead Replies (1) | Respond to of 275872
 
tgptndr, re: "Keep running the aluminum Durons, Paint them with whatever speed grade is necessary to sell, and phase out the eqivalent speed grade(s) of the Athlon. The XP 1500+ is, after all, only $12 more than the Duron 1.3G on pricewatch."

IIRC, all T-Birds are no longer in production and the XP1500+ is also no longer on the official price list.

re: "Given that a Duron 1.3G($83) is going for more than an Athlon 1.3G($88)on pricewatch"

You reversed your numbers but I the point is valid if both are Fab 25 products (and the latter was discontinued sometime in the 4Q01).

-PT



To: TGPTNDR who wrote (70958)2/8/2002 12:39:25 AM
From: PetzRead Replies (3) | Respond to of 275872
 
So, do the 1.3 GHz Durons made in Austin use the Morgan core, i.e., Palomino with 64K L2? Then they have the big advantage of lower power consumption and probably have performance better than the same speed Athlons. Also they would have SSE. I would then agree that non-XP and non-Dresden Athlons should be canned.

Petz