430w not enough for little boost on 3570k?

[coffeejunkee]

With a 5-6 year old psu the problem was most likely not degradation but the distribution of power across the different voltages. Those old psu's had relatively weak 12V rail compared to current day psu's. And since modern hardware uses mostly 12V rail adding a gpu was probably too much for it. Possibly in combination with a bit of degradation as well, ofc.

TS: Furmark stable does not mean all stable. Your gpu might be stable in 100's of games, but 1 particular one might stress the gpu in such a way that it will cause instability. But enjoy your new psu anyway.

 

[zon2020]

You are correct that they lose power. Your Math for 2% to 4% loss is wrong.

at 2% 6 years = .98 x. 98 x .98 x .98 x .98 x .98 = .8858 or an 11.42

percent loss not 18%

at 4% .96 to the sixth power is .78275 or a 21.72 percent loss not 24%.



my guess is 1 to 2 percent is more likely then 2 to 4 percent so after 6 years you may be at a 5.8 percent loss to 11.4 percent loss.

of course really pushing the gear will kill it quicker

Why is time the determining variable as opposed to load or, more likely, heat? For which of course it would not be possible to create some blanket rate of degredation to be applied to all dissimilar systems operated in dissimilar fashions in dissimilar conditions.

 

[2is]

Why is time the determining variable as opposed to load or, more likely, heat? For which of course it would not be possible to create some blanket rate of degredation to be applied to all dissimilar systems operated in dissimilar fashions in dissimilar conditions.

Everything posted here, in regards to degredation is guesswork.

 

[philipma1957]

psu supplies age. the rate varies. parts quality . time used. load percent . heat. also quality of ac going to the psu.

so no one really knows all these factors. so guessing is all we got. of course you could clock your psu for a few years and let us know. or pick .5 to 5 % a year.

 

[zon2020]

But then that provides no basis on which to adjust your original projections regarding required PSU size.

And frankly, the greater likelihood is that the falling power demands of successive generations of modern cpus, gpus, motherboards, hard disks, memory, and other components (all of which have been dropping at an astonishing rate) will likely more than offset any theoretical but incalculable degradation over time in the PSU's output.

 

[philipma1957]

But then that provides no basis on which to adjust your original projections regarding required PSU size.

And frankly, the greater likelihood is that the falling power demands of successive generations of modern cpus, gpus, motherboards, hard disks, memory, and other components (all of which have been dropping at an astonishing rate) will likely more than offset any theoretical but incalculable degradation over time in the PSU's output.

PSU degradation is not theoretical and it is not incalculable.

It is a given that is difficult to calculate.

Truth be told as long as you don't slam your psu with silly high loads in hot boxes. (sounds pornographic ) Lasting 5 years is common .

I have a good tool a kill a watt lets me ballpark the load on the psu. A steady 60% load that never changes 24/7 in a cool system is not as bad as a n up and down load ranging from 5% to 85% in a hot system.

Underclocking a system to keep the psu at a safe load is dangerous I have killed a few psu's that way. If the underclock fails wattage can double. I am running a 24/7 bitcoin pulling 230 watts and doing 750 hash. If the msi underclock resets I pull 450 watts and do 850 hash. I have a 660 psu because I want overhead if I lost the underclocking over night.

If I had a 430 watt psu and underclocked to 230 watts it looks safe lots of overhead but I can't set it and forget it.

 

[tomoyo]

Thing is most psus can at way above their maximum rating. So the so-called yearly power loss doesn't even affect the max power usage you can draw. You should be able to draw maximum rated power all the way until the end of the warranty period and probably far beyond. There is no bs about derating from the rated power, no good quality psu works only at the rated power on the first year. For examples of this, see hardwaresecrets.com tests, a lot of times they draw 20-50% more than the actual rated power successfully.

 

[tomoyo]

Also as mentioned by others, 99% chance of stability problem that is unrelated to psu, buying a psu will cure nothing most likely. Your O/C technique seems poor so far.

The proper way to oc when you run into an issue is to oc only one item with a small increase. Then test this for a few days. Than o/c something else. Continue until you find your best point.