Only for cards from the Geforce 6 on up that I am aware of that are accurate and not based off whole system consumption. For everyone thing else on down accurate figures are harder to come by and one has to sometimes judge by other means. For a Riva TNT that figure would be around 8 to 10w as an example, average load will be lower and idle load will be pretty low. For boards that were well known for VR issues with the AGP slot the limit would be around 15 to 20w.

Check out this info here:

http://forums.atomicmpc.com.au/index.php/topi … equirements-v2/

There isn't much to go on for vintage parts. Based on what those charts show, you can safely assume that GF4 uses about 40W, GF3 uses about 35W, and GF2 and earlier probably draw 30W or less.

I haven't tested my Voodoo 5 systems, but I would guess that the V5 uses about 25W.

Awhile back, I tested my old GF3 Ti500 and a Voodoo 3 3000:

Stress-Testing a PSU

Stress-Testing a PSU

Based on my own observations, the GF3 Ti500 used 22W more than a Voodoo 3 3000. Sadly, my Ti500 has since developed RAM issues and I doubt I'll find another to replace it. Ti500s rarely show up on eBay. 😢

A few weeks ago, I tested a GF4 Ti4800SE (Ti4400 8X) on the same system. At full burn, it used 32W more than my Voodoo 3 3000.

KT7AGuy your numbers make sense. They definitely started drawing more and more power over time. With the GeForce FX and Radeon 9700 cards the top models started using external power connectors, something not really needed since the Voodoo 5. Floppy > Moles > Single 6 pin > Double 6 pin

Looks like a PCI card has a limit of 25W power consumption. Something like a RIVA TNT, which typically has only a small heatsink, is going to be around 10W.

According to anandtech: Geforce1 was about 18w and geforce2 gts is half that so around 9-10w. Voodoo3 is 15w.

I've read somewhere that a good ss7 board can do 6A on agp (3,3v x 6a = around 20w max). The cheaper ones (and the popular fic 2013) i believe could only do half that or even less, that's why boards got fried when using a v3.

Agp on 440BX does much better and i don't know of any problems.

meljor wrote:

Agp on 440BX does much better and i don't know of any problems.

nvidia driver documents complained that the gigabyte 6bx7 has poor agp power supply, causing heavy voltage drops and results in many cards malfunctioning. it listed this problem in the "known issues" section and claimed that its not nvidia's fault. 🤣

440BX boards can also be quite tricky. I remember a Gigabyte BX2000 board that included a jumper that needed to be shorted in order for TNT2 and Voodoo 3 to work properly. It still caused me headaches as far as I remember. 🤣

xbitlabs has some very good power measurements for some cards. Roughly starting around the 5000 or 6000 series I think. They have (had?) a much more sophisticated testing method than any other site I've seen. In many cases they even were able to break down the consumption by voltage rail, and how much was sourced from the card slot vs the auxiliary connectors.

I was recently unable to find xbitlabs results for a GT610 or GT520, which I wanted to compare with their numbers from the 210. This makes me wonder whether they still do those detailed power tests anymore. They did them for a while, anyway.

shamino wrote:

xbitlabs has some very good power measurements for some cards. Roughly starting around the 5000 or 6000 series I think. They have (had?) a much more sophisticated testing method than any other site I've seen. In many cases they even were able to break down the consumption by voltage rail, and how much was sourced from the card slot vs the auxiliary connectors.

I was recently unable to find xbitlabs results for a GT610 or GT520, which I wanted to compare with their numbers from the 210. This makes me wonder whether they still do those detailed power tests anymore. They did them for a while, anyway.

They have them for the GeForce FX 5950 at least, showing its overall max draw being something like 75W. Something else useful is they also showed the 6800 Ultra having similar power consumption, but drawing almost all of it from the Molex 4-pin connectors (it only ever draws 4-5W from the AGP slot itself), which reduces load on the motherboard.

I don't know how accurate their numbers are, but the Thermaltake (now eXtreme) Power Supply Calculator lets you select Nvidia cards from GeForce3 Ti on up, and ATI cards from Radeon 7500 on up. Might be worth a look:

http://www.extreme.outervision.com/psucalculatorlite.jsp

Wikipedia starts to list TDP only since 8xxx series..

But it wouldn't be bad if while you are at it you could note down your findings 😈

GF1 : 16W
GF2 : 8W

sunaiac wrote:

GF1 : 16W
GF2 : 8W

Watts of what exactly?

I do not understand your question.
So I'll answer : watts of electric power.

sunaiac wrote:

I do not understand your question.
So I'll answer : watts of electric power.

He/she likely means which voltage rail (5V or 12V) has to constitute the majority of the power draw. 😊

No, I just could not believe such reduction between first and second GeForce was possible. So I thought the 8 watts has to be for example only chip TDP or something.
Well, here is what I gathered from internet: http://www.vintage3d.org/rgraph/single/cons2.php

I wouldn't put too much faith in that chart - they've got some identical models listed multiple times with different values (e.g. 6800GT), and others listed with higher values than have been either provided by the manufacturer as max TDP (e.g. 7950GX2), than have been measured as max consumption in reviews, or than are physically possible for the card to consume with all of its connections at 100% draw in some cases (e.g. 290X).

Guaranteed power delivery is not a hard limit. Card like 290X can be pushed very far.

Putas wrote:

Guaranteed power delivery is not a hard limit. Card like 290X can be pushed very far.

290X is an easy target - it's relatively new, and has been reviewed and measured quite a bit. Here's three examples:
http://www.tomshardware.com/reviews/radeon-r9 … ew,3650-29.html (290W max/218W gaming)
http://www.guru3d.com/articles-pages/radeon-r … chmarks,10.html (286W max)
http://www.techpowerup.com/reviews/AMD/R9_290X/25.html (282W peak (actual gaming test), 315W "maximum" (which they define as furmark stress-testing))

We could probably assume that chart is using TPU's furmark data, but there's no attribution or source citation, so it's impossible to say. That's another criticism for that chart - nothing is cited, no measurement procedures are explained, etc. So who knows where they stole their data from, and if whoever originally tested this stuff was doing a good job, or if the numbers are even directly comparable. 😵

Like I said, I wouldn't put too much faith in it - it's lazily slapped together at best, and downright inaccurate at worst. If they can't even take the time to qualify where they're getting data from or what exactly they're showing with something as well measured and documented as the 290X, how can we trust their results for something older, rarer, etc? For example they claim FX 5800 Ultra is 74W TDP, just like 6800 Ultra, but from my own testing the 6800 Ultra draws more power in the same systems under the same working conditions. I've also never seen a published review showing power measurements for 5800 Ultra, so where are they getting their data from?