To: Cirruslvr who wrote (46212 ) 7/5/2001 10:28:00 PM From: fyodor_ Read Replies (3) | Respond to of 275872 The FLIPFLOP nature of this thread… continues to astound me. Since AMD's profit warning, comparisons with the "K6-times" and the like have been numerous. While there are similarities, there are many, quite significant differences: 1. AMD ASPs now ~ AMD ASPs in the "K6-times". AMD had a significantly cheaper platform to leverage (the socket7 and ss7). Since the K7 has, so far, been competing almost exclusively with the PIII, this platform cost advantage is reversed. For similar system prices, this requires AMD to sell K7s at a lower price. When the competing product becomes P4, the cost advantage shifts to AMD again. (I'll leave out talk of the economic climate, although it certainly plays a factor as well) 2. Platforms - now and then. Back then, AMD was desperately clinging to an old platform that, to put it bluntly, simply wasn't up to the task. Performance was awful, compatibility likewise. E.g. Plug-n-Play never worked quite right on ss7. Today, AMD has a powerful platform that is nearing maturity. Bus speeds still have room for improvements (unlike the ss7!), stability is excellent - same with compatibility. If they should want to do so, AMD can continue to leverage this platform (EV6) for years (with slight upgrades - e.g. increased bus speed and high-speed NB/SB connection) - although it doesn't look like they will chose to do so . Regardless, the possibility exists, as was certainly not the case with the ss7 platform. 3. Performance While AMD has been left behind in terms of raw operating frequency, performance is holding up quite well. With the K6 and K6-2, AMD was sorely beaten by the P6 core in a slew of important benchmarks. Only integer and office benchmarks were reasonably close. 4. Athlon4MP ~ K6-III No way! The K6-III improved AMD's performance in all the wrong places! And not a lot, at that! Athlon4MP improves performance, sometimes quite drastically, in many of the places where the P4 outshone the K7. This is a much, much better (relatively speaking) core than the K6-III. Additionally, the relative die size increase is much, much smaller for the MP (over the Tbird). Additionally, heat production is LOWER, not higher. Bottom line(s): - The down-right shitty economy makes for tough times - both for AMD and for Intel. This directly results in either drastically lower sales volume or drastically lower ASPs. Choose your poison. - The Athlon4MP is a tremendous improvement over the Tbird in the benchmarks where the latter was being soundly challenged by the P4. Additionally, power consumption is down some 20%. There may or may not be a problem with scaling. While initial tests seem to indicate that there might be an issue, it really is quite difficult to know. It makes a lot of sense to start a new core off with a much more conservative process, tweaking it only when yields are good. There may be a problem, but it's hard to tell at this point. - In Q2, AMD presence in the notebook and server/workstation segment was virtually nil. Significant revenue will be booked for these segments in this Q (Q3). This will serve to increase market share (providing desktop market share stays the same, which is not at all guaranteed) and increase ASPs (same caveats apply). - For the next year or so (at least), AMD has a really good platform in place - in both the server/workstation and desktop markets. A really good mobile platform has yet to emerge, but there are a couple of promising candidates. Additionally, AMD could choose to do their own (basically, a .13µ, low-voltage version of their 760 chipset + PowerNow! - UMC and/or TSMC would be more than happy to do this at a very reasonable price, given their record low fab utilization rates). -fyo