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: Ali Chen who wrote (44770)	1/4/1999 4:51:00 PM
From: Bruce A. Thompson	Respond to of 1571031

\|\|\|\|\|\|\|\|\|\|\|\|NEWS\|\|\|\|\|\|\|\|\|\|\|\|\|\|\| Monday January 4 4:37 PM ET Intel Unveils Faster Celeron Chips; Cuts Prices By Duncan Martell SAN FRANCISCO, Calif. (Reuters) - Intel Corp. (Nasdaq:INTC - news), the world's largest computer-chip maker, Monday introduced faster versions of its microprocessors for low-cost personal computers in the latest move to regain market share lost to scrappy rival Advanced Micro Devices Inc. (NYSE:AMD - news) Intel, which had long ceded the low-cost PC market to AMD and National Semiconductor Corp. (NYSE:NSM - news)'s Cyrix unit, was forced to introduce a chip to compete with AMD and Cyrix when the market for PCs costing less than $1,000 mushroomed in the past year. Now, PCs cost as little as $799. But investors then became concerned that Intel would not make as much money on the Celeron low-cost chips as it does on its high-end Pentium II and Xeon processors, the brains of PCs. Intel, which does indeed make good profits on the Celeron chip, managed to allay those fears at its semi- annual analyst meeting in November. Even so, Intel is seeking to boost its share in the U.S. retail PC market, where it lags both AMD and Cyrix with about 28 percent of the market. With sales of PCs costing less than $1,000 accounting for 59 percent of machines sold at retail in November, according to PC Data Corp., the 30-year-old chipmaker can no longer afford to ignore that market. ''You have to let that horse run'' in the low-end of the market, Paul Otellini, head of Intel's architecture business group, told reporters at a briefing in San Francisco. Monday, Santa Clara, Calif.-based Intel unveiled two new Celeron chips that run at 400 megahertz and 366 megahertz and priced at $158 and $123, respectively, in quantities of 1,000. Those chips use a system of pins to connect to the PC's innards instead of the single-edged slot design of the Pentium II. With these faster chips, Intel took the memory and placed it directly on the chip itself, rather than on a surrounding board. That helps cut costs and will, analysts say, make it more competitive with offerings from AMD and Cyrix. Intel also cut its Celeron chip running at 300A megahertz 21 percent to $71 from $90 and its 333 megahertz Celeron processor 16 percent to $90 from $107. Intel made small cuts on Pentium II chips running at 350 and 400 megahertz. Otellini reaffirmed Intel's intent to bring out its next- generation Pentium II chip, code-named Katmai, by the end of the current quarter. That chip, with 70 sets of new instructions, is designed for heavy-duty multimedia programs. Intel said it also will bring out a Celeron chip designed for laptop computers later in the quarter, adding that PC makers and customers can expect more frequent price cuts on the low-end of its microprocessor lineup. Intel stock closed up $2.25 at $120.8125 on Nasdaq.

To: Ali Chen who wrote (44770)	1/4/1999 5:04:00 PM
From: Tenchusatsu	Respond to of 1571031

<What are you doing in "server chipset" development at Intel if you do not know your data locality patterns?> I validate 'em, Ali. I don't do performance. <Once the data get retrieved and cached, no other processor needs this data (because other transactions most likely are completely independent from yours), and no interprocessor cache contention occurs.> Actually, there is a lot of interprocessor cache contention as the L2 cache size increases. That becomes a problem when the L2 cache becomes bigger and bigger, since then the rate of cacheline invalidates transactions increases proportionately. Why is there so much interprocessor contention? I don't know. Perhaps the data set that each processor operates on isn't as independent as we think. For example, all processors have to look up the same hash tables, indices, database query code, etc. Tenchusatsu