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: ted burton who wrote (35776)	4/15/2001 12:13:56 AM
From: Charles R	Read Replies (1) \| Respond to of 275872

Ted, <This feature offers a huge advantage, which is really only vulnerable to FUD. > I agree hundred percent with you on the power issue. It is FUD with very little understanding of enormous strides made in power management in the last decade and the benefits it offers consumers. What do you think of the other aspects of the article? I actually thought Bert McComas article was well written until he got to the power angle. The article might end up being picked up by a some opinion makers and raise the FUD level but McComas completely blew that one. Chuck P.S.: By the way, Intel marketing guys are no saints and generates a lot of FUD so don't take it too hard now that company on the recieving end on this one.

To: ted burton who wrote (35776)	4/15/2001 2:03:21 AM
From: Petz	Respond to of 275872

ted, throttling is a very good idea which AMD will, no doubt, copy. But I still think the Max Power rating in the spec should reflect the absolute maximum power which can be drawn by the CPU without throttling. It would be easy to measure the max power consumption for a few seconds while running the most pathologically designed power-monger program, before the throttling occurs. Otherwise, the temptation would be to set the maximum design power consumption too low, resulting in marginal OEM designs and frequent frequency throttling. I don't trust either AMD or Intel to honestly report max power consumption under non-worst-case conditions. Petz

To: ted burton who wrote (35776)	4/15/2001 12:58:33 PM
From: pgerassi	Read Replies (1) \| Respond to of 275872

Dear Ted: The specifications are for design engineers. These specifications are for those that build these into systems. Reducing max TDP is making the CPU look like it is better than it really is. A Thermal Management Design needs the actual worst case power and thermal output so that he can design a system that delivers the maximum performance for the given criteria (usually the requirements of the product). By lying to these people, Intel is making the CPU go into designs that are unsuitable for the claims they make. If you design to no more than the P4 specs, you will wind up with a system that will not perform in a power user environment. These units would not work even with a dumb user that uses a client like Seti that performs heavy duty number crunching. That will force the CPU into thermal overload and thus cycle all of the time (it runs in the background chewing up what would be idle cycles). This high performance system will always run at 50% of max speed. Didn't think of that nasty case? And so many now do it, even unsuspecting system administrators. By giving the real worst case power (and it is very easy to get within 95% of the worst case in many apps that stay entirely within L1 & L2). If Intel wanted to convey that typical TDP is 54W, they could show a speed vs TDP graph showing maximum, typical, and minimum TDP from 0 to 100% rated speed. And they should also specify the min/typ/max temperature that starts thermal cycling. All of this is needed for a design engineer to build a minimal cost system that delivers desired performance under worst case conditions (and these can get quite bad like max voltage, max ambient, nasty app, dusty and dirty environment). What Intel is hoping is that designers will build in at least a 40% margin so that thermal cycling never occurs. With today's typical do it on the cheap, margins are smaller than they used to be (and getting smaller). With AMD, they know that if they design exactly to spec, it will work in all conditions at rated speed. With Intel's P4 specs, a design exactly to spec will produce a system that delivers rated speed, only if run in a freezer or in other normally cold places. Users that throw their systems into hot closets and other typical places will get the shaft considering that the price difference between a 500MHz P3 and a 1.5GHz P4 is at least $500. What would you do, if your brand new car goes only half as fast as the manual states when the temperature is 75F. You ask your dealer and he states that is to prevent overheating in hot conditions. You would be pretty angry. And you would go to your State Attorney's Office of Consumer Protection and charge the Manufacturer with fraud, and anything else they can think of. This is no different. There is no excuse not to use absolute maximums and minimums in the specs. More info is better than less, but these have always been worst case numbers. This was and continues to be the correct policy. Anything else is unethical, if not downright fraud. Pete

To: ted burton who wrote (35776)	4/15/2001 11:24:03 PM
From: Joe NYC	Respond to of 275872

Ted, This feature offers a huge advantage, which is really only vulnerable to FUD. Go ahead and hype it up - convince yourselves it's a major flub-up. Don't overdo it though because its a sure bet the OEM's have already asked your beloved AMD to copy it, and they've already got someone working on it. I suspect AMD has something similar in Palomino, probably more powerful, since PowerNow can lower voltage as well. And I agree that the primary problem for Intel is FUD. That is, the only thing ging for Intel right now is MHz of P4. Suppose this single variable is undermined severely by repeatable tests showing CPU to actually run slower than nominal clock speed. If that happens, a reasonable measuring stick will become actual performance of the CPU, and this is where Athlon wins big. Suppose (and this is a long shot), one day your local computer retailer will not display MHz of the processor, RPM of the hard drive, but Winstone or Sysmark number of the system. IMO, it would be a better measuring stick than the number of different variables of the number of different components that they display today. Joe