# [Q] About "GeForce FX 5700 Ultra goes GDDR3"

##### Distinguished
Sorry, I have a little question about the power consumption measurement's topology in the article "GeForce FX 5700 Ultra goes GDDR3" (http://www20.tomshardware.com/graphic/20040405/gddr3-11.html).

According to the description, THG use "Conrad Elektronik Energy-Check 3000" to do power measurement. I check the "Energy-Check 3000" on web site (http://www.elektronik-online-bestellen.at/energy_check_3000/125319.html), and this one seems can only measure the PSU¡¦s AC input power, not the real system power consumption.

Because the PSU also wastes (consumes) some power during AC-DC transformation, there is a term called "PSU efficiency". A simple way to express this phenomenon is:
"Real System Power Consumption" = "AC Power Consumption (in Watt, not VA)" x "PSU efficiency"

That is, if the real system (MB+CPU+Graphic+HD...etc) power consumption is 200W, and the PSU efficiency = 70% at this moment, than we need about 285W AC power. Every PSU has different "PSU efficiency", and different load has different "PSU efficiency" too. (ATX spec. has some requirements about the efficiency in different loading circumstances. http://www.formfactors.org/).

Because some PSU's efficiency are good (~80%), but others probably don't. (for example, there is some efficiency data here: http://www.oc.com.tw/article/0307/readcmtarticle.asp?id=1637). Using AC power measurement data probably mislead the reader about the real system power consumption to a bigger value.

My question is very simple, if THG measure the ATX's AC power input, then we must know the "efficiency" to get the real system's power consumption value. Or, THG use some other ways to measure the real system power consumption, like "dc current meter"?

PS:
Because the article doesn't provide much detail, please forgive me to waste your time to clarify my confusion. Thank you!

#### Borsti

##### Distinguished

you´re absolutely correct. I just meassure the pure AC Power consumption. I thought about considering the PSU efficiency (We´ve tested the Antec 550 TC with 69% http://www4.tomshardware.com/howto/20040122/power_supplies-22.html) but I felt that it makes not much of a difference for the article.

First, the AC Power Consumption is the number you have to pay anyway. Second, the numbers are very theoretical (only valid for my very own system configuration).

My idea was to give the reader an impression of how the graphics card influences the power demands. So I showed several cards in relation. It´s just for getting a "feel" for it.

But next time I´ll talk about the PSU efficiency as well and explain how these numbers are meant!

Lars

#### TheGreatGrapeApe

##### Champion
Thank you very much for this. I'm not sure if this had anything to do with our request from a long time ago, but I'm certain it will be appreciated by a great number of people here.

Personally I was amazed at the large power draw by the R9800XT from the start, and only under load did the FX 5950 pull ahead.

I would have been interested in seeing an R9600Pro alongside an R9600XT (at pro speeds) and an R9600XT at full speed to see the effect of the LOW-K process on power consumption.

Thanks as always!

- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK

#### Borsti

##### Distinguished
Thanks! The 9600 numbers are a great idea. I´ll try to run those numbers later today or tomorrow.

You mean the Pro at 400/600 and the XT at 500/600 or both at the same level?

I think I´ll include the power numbers in the next VGA charts update as well.

Lars

#### TheGreatGrapeApe

##### Champion
If you run both at the same level (400/300[600]) that would expose the effect of low-k, you currently have a baseline R9600XT at stock speed if you haven't tweaked since the test. If you've got a good R9600Pro you could overclock it to XT levels. I overclock my well above for stable runs to check new benchies and such. When I initially got mine, alot of us BuiltByAti R9600P users were setting the 'standard' benchmark level at 500/350(700) because almost all of us could reach that level without any problem, and often go well beyond. So 500/300 on the R9600PRo may be possible, but no point in risking a card if a 400/300 XT will accomplish the same feat. Hopefully the lower speeds will suffice.

Whichever way works best for you, either way, should provide us with a great deal of insight (one would hope).

The power numbers in the next chart would be great. Many people here have PSU issues, and questions, and I'm sure that would answer many of their initial concerns.

I chose the R9600Pro for power and heat concerns. And I know a few others who were so inclined.

Unfortunately I think the next gen will not be so frugal, and a substantial PSU and good cooling will be the norm for all but the lowest end (at least until the 0.11 vpus/gpus arrive). Maybe a Sapphire UltimateEdition R42X or GainwardNV4X with heatpipe will at least remedy the scond situation and reduce noise without a blower .

Thanks again, and great idea to have it in the VGA charts.

- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK
<P ID="edit"><FONT SIZE=-1><EM>Edited by TheGreatGrapeApe on 04/06/04 01:38 AM.</EM></FONT></P>

##### Distinguished
Dear Borsti,

Your measurement data is definitely very useful to lots of people (Although I think it would be better to make a note inside the article. ^_^ ) .

Lately, WinPC had made some power measurements in their magazine (2004'04). It appears that they used similar way. There web site has an article with part of their measurement result (they want user to buy their magazine, so the power measurement data was not full disclose on the website: http://arena.nikkeibp.co.jp/winpc/col/04/20040203cpu/)

BTW, thanks again~

#### phial

##### Splendid
If you run oth at the same level (400/300[600]) would expose the effect of low-k

intesting idea

-------

#### TheGreatGrapeApe

##### Champion
Of course point out my spelling error before I can fix it!

- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK

#### phial

##### Splendid
ROFL oh yea look at that, didnt even notice

now it shall stay there! bahah

-------

#### pauldh

##### Illustrious
Yes thanks for these results. I would love to see the numbers in the next VGA charts. Curious as to R9800 pro compared to 9800XT being as it is one of the most sought after cards right now.

I too was amazed at the R9800XT's numbers.

Thanks again.

ABIT IS7, P4 2.6C, 512MB Corsair TwinX PC3200LL, Radeon 9800 Pro, Santa Cruz, TruePower 430watt