How Much Power Does Your Graphics Card Need?

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
Thank you Tom's Hardware for another great article!

I have been an avid reader of your articles for many years. This article deffinately helps me at my job. Many times I have my development staff requesting high end graphics cards and it is very helpful to know what kind of PS I require instead of just guessing.

-Randy
 
So what about the Physx card that ATI fans will need to get the same great gameplay that Nvidea owners get? How much power do they consume? Just because a brand of video card appears more efficient, it may not be as good.
 
IMO power supply is the most importnt component of each and every PC.

In the past between 1990 and 2004 all mass produced PCs have had underrated power supply units at least 50%!
Their Chineese made U$ 5.00 power supplies was the main cause for so many PCs failures!
Many PCs manufacturers were usuing power supplies between 200W to 300W while in reality we needed at least 450 W.
That was as I believe a major reason for PCs failure and for BSODs.

Despite using Enermax models EG465AX, EG651AX, and EG701AX, I had a very high rate of their failure either due failure o their capacitors (?) or possible overheat.

ENERMAX was very kind with their warranty, but I could still not understood why I had such great rate of failure.

Now I am with 350 U$ with Enermax Galaxy DXX EGX1000EWL 1000 W (for AMD Quad, and if this unit fails me I will likely switch the brand and try Corsair HX1000W.
I am NOT PC Gamer!

Despite that intel i7 has positive results personally I preferAMD processors architecture and for my needs sufficient performance.

I am NOT PC Gamer, but I need a very powerfull and stable crash free PC, or it costs me money.

When calculating Power supply demand I always used a multiplier of 35% for future growth and expansions.

so if I needed 385 watts I would multiply it by 1.35 and end up with 500 Watts power supply unit.

Despite that there is a load factor and all equipment is not runing at full power capacity the higher power demand could result in PC crash or toasted DVD, while there was a power shortage.

So I am convinced that the bigger is better and of course quieter!





 
Excellent article. It clearly shows that GPU manufacturers overstate power requirements, but I think they are trying to account for all the POS PSUs out there that can only provide half their rating. Get an Antec, and an Earthwatts 430 is enough. Choose Whatever-Max, and you better go for 800W to be safe.
 
In the Power Consumption chart, you have GeForce 9800 GTX (512 MB) listed twice, should one lower one be the GeForce 9800 GT (512 MB) or the upper one have more memory?
 
So if my calculations are correct You could build a monster deserving of a 1.1 kW PSU by using the following power hungry components:

60.00 nVidia based Motherboard
147.00 AMD Phenom x4 9950 Black with a 10% OC
25.20 8 x Memory Modules with a 10% OC
20.00 2 x Optical Drives
80.00 8 x 3.5" Hard Drives
488.88 GeForce 8800 Ultra 3SLI with a 10% OC

821.08 Total Power Consumption in Watts

Ok, I know I am making a lot of assumptions here about the ability of these components to Overclock, etc... but I'm just having a little bit of fun with this, so indulge me...

If you had these components and you wanted to keep the Load to 75% or less of the PSUs rating...

You would need a PSU rated for 1094.77 ala 1100 Watts or 1.1 kW 😀
 
Excelent article! I see no 8800GS or 9600GSO but it doesn´t matter.
One thing to keep in mind, those watts are from the wall outlet, not from the PSU to the system. In order to know the true power your PSU needs to handle you have to take care of how efficient the PSU is?

The differences are commonly between 20% to 10% depending of the PSU and the load.
 
[citation][nom]Gaea[/nom]Excelent article! I see no 8800GS or 9600GSO but it doesn´t matter.One thing to keep in mind, those watts are from the wall outlet, not from the PSU to the system. In order to know the true power your PSU needs to handle you have to take care of how efficient the PSU is?The differences are commonly between 20% to 10% depending of the PSU and the load.[/citation]
Please read the article before commenting. PSU efficiency was taken into account for actual graphics card power consumption, but not for electricity cost (because you pay for the inefficiency as well).
 
PSUs are the most underrated part of a rig. Getting the game crashing blues, just when you are owning the crowd frustrations?

A solid PSU is the basis of your rig, chaps....read up and don't be a miser. Rated watts can be very deceiving. Get something that will do the job and a bit more.
NO need for a house warmer, but go cheap and you will be frustrated.





 
THANKS A MILLION Tino Kreiss all the way to South Africa your the first person to lay out in detail in an very easy way to understand PC power requirements. REALLY THANKS for your time and effert you put in into the article. (ps reading tomshardware for years now this is the first time i felt like saying something) GRAET Website keep up the good work.
 
It'd be nice to see a test of where the cards are taking their power from - how much from motherboard and how much from pci-e connectors. That would be useful information indeed for multi rail PSU users.
 
[citation][nom]quotetracker[/nom]IMO power supply is the most importnt component of each and every PC.In the past between 1990 and 2004 all mass produced PCs have had underrated power supply units at least 50%!Their Chineese made U$ 5.00 power supplies was the main cause for so many PCs failures!Many PCs manufacturers were usuing power supplies between 200W to 300W while in reality we needed at least 450 W……….[/citation]
Or you could simply buy the Antec case that came with dual hotplug PSUs. Why go cheap if you need semi-pro stuff?

BTW, OEM PSUs are generally of acceptable quality so I am not sure where you got the idea everyone needs a 450w+ PSU.


[citation][nom]TechDicky[/nom]So if my calculations are correct You could build a monster deserving of a 1.1 kW PSU by using the following power hungry components:60.00 nVidia based Motherboard147.00 AMD Phenom x4 9950 Black with a 10% OC25.20 8 x Memory Modules with a 10% OC20.00 2 x Optical Drives80.00 8 x 3.5" Hard Drives488.88 GeForce 8800 Ultra 3SLI with a 10% OC821.08 Total Power Consumption in WattsOk, I know I am making a lot of assumptions here about the ability of these components to Overclock, etc... but I'm just having a little bit of fun with this, so indulge me... If you had these components and you wanted to keep the Load to 75% or less of the PSUs rating...You would need a PSU rated for 1094.77 ala 1100 Watts or 1.1 kW[/citation]
I don’t think so as Phenom x4 9950 would bottleneck the 8800 3SLI setup so they are not going to run at 100%. In fact, I can’t think of an instance where you need to burn 2 DVDs, backing up all drives while playing Crysis.
 
Errors in calculation, how can 8800gt with less wattage cost more to run than the 4870?
What's wrong with 8800gt 512 and 1024 wattage how is it that different ? Credibility people ....
 
[citation][nom]neiroatopelcc[/nom]Both of you misinterpreted my post. My power supply is a 620W corsair I received last year. Presumably rather efficient. My 400W came from the page where it said, that a 600W psu is effective from 390w and up to 5-something. So I'd need to reach at least a 400w power consumption on my rig to get within the generalized band of high effectiveness. I remember seeing a high power shootout, where a silverstone psu was actually more efficient at 300w than at 600w though, so I know the chart doesn't represent all power supplies correctly. But still. Anyway, I still think it's better to have a too big psu than a too small one. I still remember when my 550w psu sent flames out the back when I tried to power an overclocked northwood, a couple raptors and all the other stuff that went into the system. I don't want that again.[/citation]

Well i believe it is this where the review sets us on the wrong foot. For example; look at this . It's obvious that with a 150watt load the HX620 PSU's efficiency is still better then the 80% to 100% full load. So beside this article shows some interresting figures and gives us a good idea what is happening inside the PSU; the table for 65% till 85% i take with a grain of salt.

THe HX620 performs very well from as low as 25% of it's total Output. Possible even lower. The dropoff point lies somewhere between 50 and 100watt, i guess.
 
An 80+ certified PSU is required to be 80+% efficient across a variety of loads; I think one of them is either 25% or 40%.
---
And yes, I had a cheap "550W" PSU blow chunks on a 2.2GHz AMD "Thundering Herd" (lol, or whatever it was called; I've forgotten), with a 440MX and a couple of IDE drives. There was a stink in that room for days. The mobo and GPU were blown, replaced with a 2.4Ghz P4 on some Intel board, with a 9250SE.
 
[citation][nom]V3NOM[/nom]it must use both. otherwise it would just have a single 6 pin,seing as how most people don't have 8 pin PSU's, and those that do have "6+2" pins, which allow you to use either 6 pins or all 8.[/citation]
Ahhh my PSU has two 8 pin plugs none of this 6+2 bs. Truly prepared =] 1 kilo watt. And is truly a overkill =/ heheh
 
Oi, i have a core 2 duo 2.4 ghz with 3 gigs ddr2 800, and a GTX 260 oc, its the 192 not the 216. It's all running on a 430 watt power, sooo if you're buying a 260 oc, then the wattage consumtion is probably not the same as the table says. My computer shuts off on games like crysis, when running ultra and high def, runs great, but then crashes. At full load, by it self, the card can take 330 watts. Pece
 
It's a bit strange to test a gaming PC with 8hours/day, 365days a year.
I think no person is capable of doing such a thing.
The situation is different with desktops that are used in work environments.

It is more feasible to test a gaming pc/graphics card @ 2-4 hours a day, 300 days a year. (I mean, you gotta take a break sometime, no?)

With this test one could say his card costs him 150$ per year over another card, while in real life it costs him eg: only $40 per year more because he only plays games 2 hours a day, 250 days a year.

Then there's less the urge to buy a newer graphic card because at the end of the year he'll lose this amount anyways in electric bills.



If one is really going to game 356x8 hours/year one better buys a more expensive card, since one is playing a lot.

to some, power requirement will determine which card to take.
I mean, if could choose between 2 AMD and 2x Nvidia cards,that would be powerful enough to run my games in full detail, and where cheap enough to buy, probably the power requirement would be the determining factor after technology advancement (eg: DX9 / 10 / 10.1 card),and extra's like HDMI/DVI/...

Ps: some numbers seem a bit off on the power chart of the AMD cards, where a card used more in 2D than 3D (4550,and 3650 seem a bit off here).
 
The numbers for the 8800GTS/512 are pretty much spot on as my Kill-A-Watt measures it. So I have high hopes for the other numbers.

I overbuy my PSU, as I prefer less heat and noise in the case and 100% stability.
 
Status
Not open for further replies.