- Aug 28, 2010
- 1,551
- 203
- 106
I might buy a gtx980Ti.
Reference gtx980Ti's are using maximum 250W.
Factory-overclocked after-market gtx980Ti's are using around 275W.
I guess if you overclock yourself, and increase TDP and voltage even more, the wattage can go up. To 350W even.
My gtx680 is supposed to use 175W max.
I now wonder if my PSU can deal with the extra 75W-100W if I buy a GTX980Ti.
Current system:
i5-3570K@4.0GHz, ASRock Z77 Extreme4, gtx680, 8GB of ram.
PSU: Super Flower Golden Silent 500 W
I have an almost silent system. I don't want to change that.
So I have a water-cooled CPU and GPU.
And a fanless PSU.
The PSU is supposed to be pretty good.
Two reviewers call this the best fanless PSU they ever tested. And one of the best PSU over all. Over 90-93% efficiency at all power usages.
Tech PowerUp review: http://www.techpowerup.com/reviews/SuperFlower/SF-500P14FG/9.html
SilentPCReview: http://www.silentpcreview.com/Fanless_PSUs_Kingwin_STR-500_Silverstone_ST50NF
(Note, the Kingwin Stryker STR-500 is basically the same PSU as the Super Flower).
I measured the power usage of my current system.
I used am Energy Savings Meter. Plugged it in my wall-socket. Plugged my system in the meter.
Power use at idle: 60W-70W.
Power use when loading the CPU (e.g. Prime95 on 4 cores): pretty stable, ~145W.
Power use when playing games (e.g. GTAV or TW3): 150W-250W. Peaks of 350W some times.
Power use when running Prime95 + FurMark (1.15.2.2): 250W-350W. Once a minute I see a peak over 400W. Usually 410W-420W. Record was 440W.
Suppose I upgrade my gtx680 to a gtx980ti, to a slightly overclocked one. (Like EVGA SC running at 1,102GHz/1,19GHz). That one should use 275W max. That's 100W more than my current gtx680.
One think to note is that measured wattage at the wall-socket is higher than the actually wattage delivered to the pc-components. Let's assume efficiency is 93%. (It's very close anyway. 91.12% at 110V, which will be close to 93% on 220V. I am on 220V, btw). That means that if I see 538W at the wall, my system will be using 500W. 538W is the max for my system.
So the gtx980ti might use 100W more than my gtx680. When taking efficiency into account, that means my system might draw 108W extra from the wall. Let's modify my currently measured values with that extra 108W.
Idle: is now 60W-70W. Will probably stay similar, as Im not using my videocard. In the worst case, 170W-180W. Way under 538W. Not a problem.
When running cpu-intensive applications: is now 145W. In the worst case, 250W. Not a problem.
During games: 150W-250W usually. Peaks to 350W.
This will now be: 260W-360W. Still not a problem.
Peaks will go from 350W to 460W. Still under the 538W max. Should not be a problem.
During stress test with Prime95 and Furmark: 250W-350W, peaks to 440W.
This is where the problem is.
New average usage will be 360W-460W. With peaks to 550W.
550W is more than the theoretical 538W that my PSU can deliver.
However, there are 2 factors that might make this less bad than it looks.
1) This high wattage is only reached during artificial stress-testing. If I never run FurMark, and only play games, it looks like I will always stay under 460W.
2) The problematic max is only during peaks. I see those peaks maybe once a minute at the max. PSUs are supposed to be able to deal with short peaks in power use.
Look at this page:
http://www.techpowerup.com/reviews/SuperFlower/SF-500P14FG/5.html
Look at the chart: Voltage Regulation and Efficiency Measurements
The Tech Powerup tester has tested the SuperFlower up to 600W ! That's internal use. The use at the wall-socket was 660W. Apparently the PSU was able to deal with that without any problems. If it can deal with 660W wall-usage over a longer period of time, it should be able to deal with 360W-460W and peaks of 550W. It seems that even during the hardest stress-tests, I will still stay 20% under the max capacity of the PSU. During regular gaming, I even stay 40%+ under max capacity.
However, I am not an electrical engineer.
And I am not sure this simple math is correct.
And I might overlook something fundamental.
So, what do you think ?
Would my 500W SuperFlower Fanless PSU be good enough to replace my gtx680 with a gtx980ti ?
Thanks in advance.
--
(And for the record: no, I will not replace my PSU. And also: I know the standard answer will be: "just buy a 800W PSU, so you'll be safe". This time I am measuring things, and I want to talk more precise numbers).
Reference gtx980Ti's are using maximum 250W.
Factory-overclocked after-market gtx980Ti's are using around 275W.
I guess if you overclock yourself, and increase TDP and voltage even more, the wattage can go up. To 350W even.
My gtx680 is supposed to use 175W max.
I now wonder if my PSU can deal with the extra 75W-100W if I buy a GTX980Ti.
Current system:
i5-3570K@4.0GHz, ASRock Z77 Extreme4, gtx680, 8GB of ram.
PSU: Super Flower Golden Silent 500 W
I have an almost silent system. I don't want to change that.
So I have a water-cooled CPU and GPU.
And a fanless PSU.
The PSU is supposed to be pretty good.
Two reviewers call this the best fanless PSU they ever tested. And one of the best PSU over all. Over 90-93% efficiency at all power usages.
Tech PowerUp review: http://www.techpowerup.com/reviews/SuperFlower/SF-500P14FG/9.html
SilentPCReview: http://www.silentpcreview.com/Fanless_PSUs_Kingwin_STR-500_Silverstone_ST50NF
(Note, the Kingwin Stryker STR-500 is basically the same PSU as the Super Flower).
I measured the power usage of my current system.
I used am Energy Savings Meter. Plugged it in my wall-socket. Plugged my system in the meter.
Power use at idle: 60W-70W.
Power use when loading the CPU (e.g. Prime95 on 4 cores): pretty stable, ~145W.
Power use when playing games (e.g. GTAV or TW3): 150W-250W. Peaks of 350W some times.
Power use when running Prime95 + FurMark (1.15.2.2): 250W-350W. Once a minute I see a peak over 400W. Usually 410W-420W. Record was 440W.
Suppose I upgrade my gtx680 to a gtx980ti, to a slightly overclocked one. (Like EVGA SC running at 1,102GHz/1,19GHz). That one should use 275W max. That's 100W more than my current gtx680.
One think to note is that measured wattage at the wall-socket is higher than the actually wattage delivered to the pc-components. Let's assume efficiency is 93%. (It's very close anyway. 91.12% at 110V, which will be close to 93% on 220V. I am on 220V, btw). That means that if I see 538W at the wall, my system will be using 500W. 538W is the max for my system.
So the gtx980ti might use 100W more than my gtx680. When taking efficiency into account, that means my system might draw 108W extra from the wall. Let's modify my currently measured values with that extra 108W.
Idle: is now 60W-70W. Will probably stay similar, as Im not using my videocard. In the worst case, 170W-180W. Way under 538W. Not a problem.
When running cpu-intensive applications: is now 145W. In the worst case, 250W. Not a problem.
During games: 150W-250W usually. Peaks to 350W.
This will now be: 260W-360W. Still not a problem.
Peaks will go from 350W to 460W. Still under the 538W max. Should not be a problem.
During stress test with Prime95 and Furmark: 250W-350W, peaks to 440W.
This is where the problem is.
New average usage will be 360W-460W. With peaks to 550W.
550W is more than the theoretical 538W that my PSU can deliver.
However, there are 2 factors that might make this less bad than it looks.
1) This high wattage is only reached during artificial stress-testing. If I never run FurMark, and only play games, it looks like I will always stay under 460W.
2) The problematic max is only during peaks. I see those peaks maybe once a minute at the max. PSUs are supposed to be able to deal with short peaks in power use.
Look at this page:
http://www.techpowerup.com/reviews/SuperFlower/SF-500P14FG/5.html
Look at the chart: Voltage Regulation and Efficiency Measurements
The Tech Powerup tester has tested the SuperFlower up to 600W ! That's internal use. The use at the wall-socket was 660W. Apparently the PSU was able to deal with that without any problems. If it can deal with 660W wall-usage over a longer period of time, it should be able to deal with 360W-460W and peaks of 550W. It seems that even during the hardest stress-tests, I will still stay 20% under the max capacity of the PSU. During regular gaming, I even stay 40%+ under max capacity.
However, I am not an electrical engineer.
And I am not sure this simple math is correct.
And I might overlook something fundamental.
So, what do you think ?
Would my 500W SuperFlower Fanless PSU be good enough to replace my gtx680 with a gtx980ti ?
Thanks in advance.
--
(And for the record: no, I will not replace my PSU. And also: I know the standard answer will be: "just buy a 800W PSU, so you'll be safe". This time I am measuring things, and I want to talk more precise numbers).
Last edited: