Will I need a bigger PSU?

paradoxeternal

Distinguished
Oct 30, 2011
219
0
18,680
Hi everyone,

My tax return seems to be pretty beefy this year (between 250-500 $$) so i decided that i would get a new graphics card. I currently own a radeon HD 6870 but considering that i'm spending the last two years of college in a dorm (probably not having a job) and away from home and my normal 2 jobs, i decided its best to get the top of the line now while i have the money, so i've decided to buy an Nvidia Geforce GTX 680.

i'm not sure if my PSU will be big enough. I currently have a Corsair TX 650; the rest of my computer's specs are listed below in my signature, but i will also be adding a 256 GB SSD pretty soon.

Do i need a bigger PSU for an upgrade to the GTX 680?

i've done some research, and according to anandtech, my current GPU under load consumes about 287 watts; a 680, on the other hand, consumes 362. i also used the extreme power calculator, which i'm pretty sure is a reliable tool, and it said that with my current GPU i would need 537 watts reccomended while with the GTX 680 i would need 571. The numbers don't match up and since i'm not an expert on PSUs or GPUs or system power usage idk why.

i don't have the money for a new GPU AND PSU because i will also be upgrading to new case and putting in a new SSD, so if my PSU can't handle the 680, could i run the AMD Radeon HD 7870?

So, if you guys would know anything else or have seen something i haven't looked at, please let me know. thanks for your help!
 


Anandtech measures system power consumption not GPU power consumption.
The system they used to test the GTX 680 had an i7-3960X, the system they used to test the 6870 used an i7-920. They also use different programs to achieve the load power consumption figure. With the i7-3960X system they used Metro2033, with the i7-920 they used Furmark.
So the 75W delta between the two cards is created not only by the difference in power consumption of the two cards, but also the rest of the system and the different testing methodologies.

Personally I think if they are showing an entire system with a GTX 680 drawing 362W I wouldn't worry if my 650W PSU (a high quality unit which can deliver 650W of power continuously at 50C) could handle it.

The only sites that I know of which actually tests the power consumption of the cards themselves, not including the rest of the system is Techpowerup.
http://www.techpowerup.com/reviews/HIS/Radeon_HD_6870.html
Shows a stock 6870 drawing 163W under maximum load.
http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_680.html
Shows a GTX 680 drawing 228W under maximum load.
 
You'll be fine! It's recommended a 550w PSU for a GTX 680

I hadn't even thought of checking NVIDIA's website. Duh! Thanks.

Anandtech measures system power consumption not GPU power consumption.
The system they used to test the GTX 680 had an i7-3960X, the system they used to test the 6870 used an i7-920. They also use different programs to achieve the load power consumption figure. With the i7-3960X system they used Metro2033, with the i7-920 they used Furmark.
So the 75W delta between the two cards is created not only by the difference in power consumption of the two cards, but also the rest of the system and the different testing methodologies.

I didn't even realize that it said on the website that the numbers were system usage. and yeah the different intricacies with different PC builds and everything else can be confusing. Thanks for the clarification; i feel more confident about upgrading the GPU now.

Though i really suggest you crossfire your 6870, instead of spending all that money for newer cards.

You make a good point. The only reason I wouldn't just go get another 6870 is because i don't know if it's as powerful as the 680. considering that my financial situation won't be the same for the next 2 years, i want as much power as i can get so that i can still play games coming out 2 years from now with decent settings. i'll probably post a question about this on the GPU forums. thanks though, you most likely saved me $350.