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: andreas_wonisch who wrote (27036)	1/31/2001 6:38:09 PM
From: fyodor_	Respond to of 275872

Andreas: I'm considering building an ultra-quiet Athlon PC for things like hearing music, watching DVD, TV etc. and consider to downclock it aggressively. 10-15W should be probably low enough? I'd recommend getting a large fan and setting it at a very low RPM (e.g. using a resistor ;)). Even a modest amount of air flow can improving the cooling quite significantly. On a side note, AMD common has separated nicely from the other tech stocks this past week: chart.bigcharts.com -fyo

To: andreas_wonisch who wrote (27036)	1/31/2001 6:55:33 PM
From: Joe NYC	Read Replies (1) \| Respond to of 275872

Andreas, You can just take the power consumption of Athlon 700 as your starting point to get a more accurate number (it is listed in the table.) BTW, does anybody know at what power consumption you could run an Athlon/Duron without a fan (big heat-sink only)? I'm considering building an ultra-quiet Athlon PC for things like hearing music, watching DVD, TV etc. and consider to downclock it aggressively. 10-15W should be probably low enough? You may want to stick with a big heatsink, and a very quiet fan like these: phamcomputer.safeshopper.com Some of the ADDO fans are really quiet. 16.1 dBA for the 60mm and 22.5 dBA for 80 mm one. Apparently sound of whisper registers at 23 dBA. Joe

To: andreas_wonisch who wrote (27036)	1/31/2001 6:56:14 PM
From: Cirruslvr	Read Replies (1) \| Respond to of 275872

OT Andreas - RE: "BTW, does anybody know at what power consumption you could run an Athlon/Duron without a fan (big heat-sink only)? I'm considering building an ultra-quiet Athlon PC for things like hearing music, watching DVD, TV etc. and consider to downclock it aggressively. 10-15W should be probably low enough?" I have turned off the fan on my Duron 600 because it was annoyingly loud. I have also "under-volted" it to use 1.3v which makes the processor run cooler. Turning off the fan and lowering the voltage hasn't forced the computer to crash even though the processor runs hotter overall. Since this is a Duron and not a PIII it should be able to handle the extra heat, at least I hope! I don't know the name brand of my heatsink, but if you do this with a faster Duron I would recommend something like a Taisol CGK742xxx or one of the other good heatsinks Kyle at HardOCP tested last week or so. My power supply is also kinda loud but I'm should be receiving an Enermax power supply by the end of this week which should solve that problem.

To: andreas_wonisch who wrote (27036)	1/31/2001 7:53:28 PM
From: TechieGuy-alt	Read Replies (1) \| Respond to of 275872

Good point. I guess it did not occur to me to start off with that 1 known point and extrapolate One lives and learns :) Thanks. That's pretty amazing (the low power). But remember, AMD may not have been getting such good results with earlier steppings. Additionally, as I mentioned in my original post, I did not test to max ambient temps etc. This was hardly a scientific test :) But, at least I think that it gives more credence to the theory that AMD can do mobile Athlons pretty soon. TG

To: andreas_wonisch who wrote (27036)	1/31/2001 8:46:39 PM
From: Petz	Read Replies (3) \| Respond to of 275872

andreas, "Power consumption scales linearly with frequency" & "Power consumption scales as quadratic of voltage" Not exactly true for frequency, as I plotted the numbers for peak power and "normal"power for the TBird as specified by AMD at frequencies from 650 to 1200 MHz. At zero frequency, it appears there would still be power dissipation of about 8 watts, though the number could be anywhere between 3 and 10 watts. That makes sense, because the TBird draws 5 watts in "STOP GRANT" mode, which is, basically, zero Hz for most of the CPU. Now what about "Power proportional to voltage squared" Turns out, it should be a little better than that. The reason is that most of the current is consumed by charging and discharging capacitors, so it is the voltage difference between the logic 1 state and the logic 0 state which is the relevant voltage, not the absolute voltage. So, if a CPU is running at 1.3 volts, the logic '0' voltage may be 0.5 volts and the logic '1' voltage may be 1.25 volts, a difference of 0.75 volts. Increase the Vcc to the standard 1.75 volts (for TBirds), and we're now swinging from 0.5 volts to 1.70 volts, a difference of 1.2 volts. Current consumption would go up more than linearly. We increased voltage by 35% and the current consumption went up by 1.2/0.75 = 60%. Therefore power consumption goes up (and down) by more than the voltage squared. Now I don't know what the logic '0' voltage is, but its probably at least 0.2 volts. (The larger this number, the greater the power reduction achieved by lowering Vcc) If we assume it is 0.2 volts, the predicted TBird power as a function of frequency (F, in MHz) and voltage (Vcc) would be P(max)= (VccVcc/3.0625) 4.3 watts + (Vcc/1.75) * [ (Vcc-0.2)/1.55 ] * 0.0509 * F If the logic 0 level is 0.4 volts, the predicted TBird power as a function of frequency (F) and voltage (Vcc) would be P(max)= (VccVcc/3.0625) 4.3 watts + (Vcc/1.75) * [ (Vcc-0.4)/1.35 ] * 0.0509 * F I used a linear regression on the spec sheet data to derive the quiescent power (4.3 w) and the 0.0509 slope. I made an assumption that the quiescent power would be directly proportional to the square of the Vcc. Since at this point I forgot what we were trying to estimate, I'll leave it to someone else to calculate the estimated power consumption. One last thing, "average" power for TBird seems to consistenly be 90% of maximum power. Petz