This is a nice article...I've always wondered how much power modern day video cards consume, since we hear quite abit about it on CPUs.

I'm guessing Nvidia cards will be in PART II ???

Maybe in PART III, you can have budget/mainstream class of video cards being measured for power consumption?

I like to see how much the Radeon 9200SE chews compared to GF4 MX and GF FX5200 series.

Finally soneone does this properly! Your efforts are much appreciated!

I realize it takes a lot of effort for each card, but I would have liked to see a couple of toasty older generation top dogs like the 9700pro in there as well to provide a reference point for us slow upgraders.

Low-to-midrange comparison would also have some interest for those who are building things like HTPC:s and seeking the ideal balance between performance and low thermal output. Subject matter for a third article in the series? :)

Thanks for the great article! The power figures for ATi's newer cards, the 9600XT and X800s, are very impressive in idle modes, especially since they don't seem to drop their clock speed, like nVidia. I'm really looking forward to analysis of nVidia's cards, especially the 5700U vs. the 9600XT and the 6800GT vs. the X800XT.

Excellent article. Here in Europe the cost of powering one of these cards can have a significant impact on its Total Cost of Ownership (TCO). Running a 115W video card for 3 hours a day can cost 120 euros over two years (probably the life of the card), making Nvidia cards a LOT more expensive than ATI...

Just passing by-

I think you can work out the optimal HTPC card from these very results, who would put anything below a 9600 se in anything these days considering the price differential?

Very good article, I look forward to more test results in Part II.

Very interesting article. I was just wandering how the mobile chips compare to these analyzed in the article? Or maybe similar measurements have been done already and I haven't heard of them?

This is an interesting article. But I would have liked to see a 9700 pro in there also. I have an antec 400W psu that does not supply enough power to run my system with a 9700 pro in it, yet it runs with a 5900ultra overclocked just fine. Had to go to stronger psu to get the system stable with the 9700pro.

#7 You must have a mobo problem not a PSU, your board isn’t delivering enough power from AGP slot, maybe Ati requires more power from the AGP slot over nvidia.

Excellent review!!!,
I agree with #4 I am European, and for us the power of the washing machine, dishwasher, freezers, .... are very important that's why I don’t go for NVIDIA(6800) and Intel(prescott), just for that.

Funny is that two 9600 would consume less over one 9800. Maybe dual 9600 is a good idea, I don’t know what would cost more a dual chip graphic card (with two 9600), or a new card with other more powerful chip (one 9800).

Good article. Never seen another on the subject... quite informative. You should repeat it on Nvidia cards to see if the huge fans on the 5xxx series were really needed.

Usefull article, but a summary table, or graph showing each cards consumption would make it more usefull.

Any chance of testing a Matrox card P650 perhaps?

This test is so great! Wish to read the Part II ASAP!

Very appreciate your hard work.

Bing on part two :)

This is a very thorough review, I am very impressed by the lengths you have gone to produce it.

The only thing I felt missing was a review of the lower end x600 and x300 cards from ATI.

Once again many thanks.

Everybody knows, that it is possible to overclock gr. cards. But is it also possible to underclock them? And what about power consumption then?

That small power consumption decrease when moving from air to water cooling system... I wonder, whether the reason for that strange decrease was that there was no power fed to the GPU fan which also could not run on love alone but consumed a watt or a few? Were did the pump for the water cooling system took it's power from?

nice article, keep up the good work!

i'm going to put a GF 6800 into a sff from shuttle. the 240W PSU seems fine with the current ati 9800 pro but i'm pretty sure the 6800 is a bit more power hungry. Time will tell ;)

OK - that's all about power :) - but i still don't know about heat dissipation of the cores. Can you tell me how many heat generate 9700pro (all default - i'll calculate rest :D). I'm going to mount TEC on it but I still don't know is this good idea THX

An excellent article! Thank you for spending the time and effort to analyze the power consumption of these video cards. Its too bad that none of the manufacturer's like to advertise their power consumption numbers. Hopefully there are more reviews of other video cards to come, particularly lower to mid-range video cards.

It would be amazing to have a comparison to find the video card with the lowest power consumption. Or perhaps a video card with the best power consumption-speed ratio. And of course underclocking video cards would be an interesting topic.

The method is sound enough....... but surely, a SHUNT is a Parallel connection...... isn't it?

Hello!

You guys did a great Job, but you have just forgot about a little Thing:

The Power-Consuption of an such Chips is in Square to the delivered Voltage.
If you place a Shunt in Front you can calculate the failure from the Voltage-Drop back.

You used: P = V5 * (dV5/R) + V12 * (dV12/R) and so on for the AGP-Voltages...

You should use: P = V5 (dV5/R) * ((V5+dV5)/V5)² + V12 * (dV12/R) * ((V12+dV12)/V12)² and so on for the AGP-Voltages

This elliminates (not exactly, but better then without it) the Calulating-failure from the Voltage-drops on the Shunts.

If I calculate the Powerconsumption from the overclocked X800 on Load:

You get: 55.8871
You would get: 67.6696

Its not much, but this are about nearly 2W or 2,7% more ;-)

If you want you can use this an calculate your Results new!

Regards, Easy

I wrote a little fault .. its not 55.8871 ... its from 65.8871 to 67.6696 ;-)