Silicon Investor (SI) -- The First Internet Community

STOCKTALK

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.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Daniel Schuh who wrote (94298)	2/19/2000 9:25:00 PM
From: kash johal	Respond to of 1574472

Daniel, WOW LOOK AT SOME OF THIS DATA ON AMD FLASH: AMD Flash article/Electronic News AMD hopes to add enough capacity to keep pace, and more By Tom Murphy Cellular handset OEMs need to look to NOR flash memory capacity agreements -- such as the AMD/Samsung deal last week -- before demand for parts heats up to boiling point levels. Or so says one analyst, after Advanced Micro Devices (AMD), Sunnyvale, Calif., said it would give Samsung Electronic Co. a three-year supply of NOR flash for $400 million. "Smart OEMs would want to make more of these type of agreements," said Jesse Huffman, senior analyst, Cahners In-Stat Group, Scottsdale, Ariz. "If they don't get their act together, they could find themselves limited or in a situation where they can't acquire NOR flash." Some OEMs were caught in that very situation at the end of 1999. Not only does Huffman expect demand to outweigh capacity at the end of 2000, but the popularity of Internet appliances could also catch fire. These devices will use 16Mbit NOR flash parts and place an even tighter strain on the supply. "Clearly this long-term relationship is in the interest of both companies," Huffman said. "AMD will be able to get the price they want and Samsung will be able to stabilize their supply for their cell phone business." On the other side of the equation, not only does AMD get the cash up front, but it plans to invest the revenues into capacity expansion by building new fabs and expanding existing ones, said Kevin Plouse, AMD's director of technical marketing. The growth rate for flash memory is expected to reach 70 percent per year in bit growth over the next few years and AMD plans to keep pace with capacity expansion. So much so, the company plans on being the major player in this field. "We want to drive the flash market," Plouse said. AMD said this is the first capacity agreement the company has made regarding NOR flash and it is lining up 15 more companies with similar agreements, a spokeswoman said. Interestingly, AMD gets to sell flash memory to one of its competitors, according to Bruce Bonner, principal analyst at Dataquest, San Jose. While Samsung is a player in the flash market, it is mostly concentrated on NAND flash for PC memory storage. "What this announcement is saying is that Samsung is really getting serious about cell phones," Bonner said. "Cell phone demand is straight up. This agreement is also a coup for AMD." An AMD spokeswoman said that the company has been predicting shortages in flash since May and began arranging agreements for customers to guarantee them a certain amount of supply. Flash revenues are predicted to grow to $5.1 billion in 2000, up from $4.1 billion in 1999, according to Cahners In-Stat Group. In 2001, flash sales are expected to reach $6.4 billion. AMD has a joint agreement with Fujitsu for the manufacturing of flash memory. Two foundries in northern Japan are used for that purpose and Plouse said AMD hopes to bring an Austin fab on line in 2001.

To: Daniel Schuh who wrote (94298)	2/20/2000 12:56:00 AM
From: Scumbria	Read Replies (1) \| Respond to of 1574472

Dan, do you know anything about self timed logic? I think it's considerably more complicated to design than clocked logic, but it's been used for computers in the (well, really distant) past I think. From the little I know, it's sort of cool, things go as fast as they can go. Is "clock skew" a speed of light thing, or do you start seeing transmission line effects? Maybe the chips Most processors use some self timed logic, where data is sampled based on some local timing rather than a global clock. TLBs are a good example of this. It would be a daunting task to build a completely asynchronous processor however. IBM attempted this about 5 years ago in a PPC design, and discovered that the tool support was inadequate to verify it. Scumbria

To: Daniel Schuh who wrote (94298)	2/20/2000 1:12:00 AM
From: Scumbria	Read Replies (1) \| Respond to of 1574472

Dan, Is "clock skew" a speed of light thing, or do you start seeing transmission line effects? Maybe the chips will have to start running little Rambus thingies internally. Processors have long wires distributing the clock around the chip. Delays through the wires cause the clock to arrive at different times in different places on the chip. Because of this, a safety margin must be built into the timing analysis to make sure that an early clock doesn't sample before a late clock has resolved. Typically this clock skew is about 200-300 ps, which would use up the entire 3 GHz cycle time. It will be impossible to go past 2 GHz without a solution to this problem. Scumbria