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.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: pgerassi who wrote (76405)	4/3/2002 6:10:40 PM
From: 5dave22	Read Replies (2) \| Respond to of 275872

AMD-Intel benchmark war escalates Wed Apr 3, 5:43 PM ET By Matthew Broersma, ZDNet News Intel has begun offering a "training class" to channel customers, directly targeting rival AMD's new policy of emphasising model numbers over clock speeds. The class charges that AMD's Athlon XP 2000+ chip does not live up to its model number ranking. AMD launched its "True Performance Initiative" with the Athlon XP processor last autumn, in order to shift consumers away from the perception that a chip with a faster clock speed offers better performance. Athlon XP chips come with a model number that indicates their performance relative to older Athlon chips, so that the Athlon XP 2000+ (at 1.67GHz) would offer roughly the same performance as an earlier Athlon clocking 2,000MHz (2GHz). Intel and its allies claim the model numbers are designed to invite comparison with Pentium 4 chips of a similar clock speed. Last week Intel began advertising an online seminar to its customers, aimed at presenting its point of view on AMD's model numbers in greater detail. The move followed the release of an Intel-funded report from Aberdeen Group that criticized the True Performance Initiative as "bad science" designed to fool "inexperienced buyers". The class, offered twice a day through the end of next week, appears to use a variety of benchmarks to show that the Athlon XP at 1.67GHz does not compare favourably to a Pentium 4 2.2GHz. The class is not available to the public, but an enthusiast Web site called AMDZone published what appears to be a presentation summarising the seminar's main points. Intel's benchmarks are the main bone of contention. Two main sets of benchmarks are presented, one set from SPEC--an industry-standard ranking--and another set testing features such as content creation, gaming and video rendering. In all but the Business Winstone test, Pentium 4 is shown coming out on top. AMD dismissed the SPEC benchmark as largely irrelevant. "It doesn't relate a lot to either consumer or business use, unless you're in the scientific community," said a spokesman. He said that AMD had been careful in the set of benchmarks it had used in establishing the True Performance Inititative, audited by Andersen Consulting, so as not to lay itself open to charges of benchmark fixing. "Benchmarks can be made to say whatever you want, really," he said. Intel's use of differing system configurations is likely to arouse controversy. The Intel benchmarks compare a Pentium 4 "Northwood" core with Rambus memory, for example, while the AMD core uses DDR (double data rate) DRAM, which does not allow the same bandwidth for memory-intensive applications such as video rendering, but is much cheaper. However, most Pentium 4 systems on the market do not use Rambus memory, with system makers overwhelmingly preferring the less-expensive DDR or SDRAM options. In tests where both systems were outfitted with DDR memory, ZDNet found that the 2.2GHz Pentium 4 Northwood often could not keep up with the Athlon XP 2000+. Intel also takes one or two other pot-shots at AMD, such as by comparing the power consumption of Northwood, manufactured on a .13-micron process, to the Athlon XP, which is manufactured on a .18-micron process. The Athlon XP consumes 70W "maximum" power, according to the Intel presentation, while the Pentium 4 consumes 55.1W of "Thermal Design Power" (TDP). TDP is a figure used by OEMs to design cooling units, but doesn't include spikes in power that only last a few milliseconds, where the "maximum" power may include such spikes, according to AMD. "It's not fair at all," an AMD spokesman said. "They're effectively different units." In the first half of this year AMD is to roll out a new core called Thoroughbred, based on a .13-micron manufacturing process. The process shrink will reduce the die size, reducing power consumption and allowing higher clock speeds. Intel declined to comment except to confirm that it does offer training to its customers, including training on "competitive issues".

To: pgerassi who wrote (76405)	4/3/2002 7:34:01 PM
From: Monica Detwiler	Read Replies (3) \| Respond to of 275872

pgerassi - You wrote They were at around 1% in units and 0.5% on revenue depending on how you define the market. They may be at 5 times that now and that is not essentially zero Well, Pete, by the same stroke of imagination, AMD might really be back at zero % market share since, during the intervening time, Intel has introduced low power Pentium 3 blade server products and 2-way Xeon server chips up to 2.2 GHz and 4-way Xeons with large L3 caches up to about 1.6 GHz. Right? AMD might not be able to supply 100% of the WW desktop market, but they could supply 100% of the server market. AMD can only supply zero percent of the 4-way and above server market. Ask yourself this - why would a server manufacturer want to use AMD for only a very small part of their server line (no low power versions remember ?) when they can get everything from Intel - low power, blades, large cache (up t 2 Meg L2 or 1 Meg L3), high speed, 2,4,8,16 and 32 -way....? Well, you get the picture, don't you Pete? Monica