So I've got all my computer's mining right now and I would like to get my gaming rig back. To do this I'm thinking I could put all my mining GPUs one one system. The problem with this is I only have a seasonic 850w PSU. With three cards underclocked and undervolted it pulls 650w at the wall. I figure if I add another card it would be at 800w. What are my odds of not burning out the PSU doing this 24-7?
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Running PSU up to max rating.
- Thread starter travm
- Start date
Ocellaris
Fully [H]
- Joined
- Jan 1, 2008
- Messages
- 19,077
What cards are they and how many do you have? You can’t just pile as many card as you want into a system and expect it to game better.
It's for mining.
I get my gaming rig back by taking out the mining cards and putting it in the other system.
Has no thing to do with my question, but since you asked, they are rx580s. Im not using them for gaming, at least until they have paid for themselves.
To game I'll just put my 7870 back in. It's worthless for mining.
I get my gaming rig back by taking out the mining cards and putting it in the other system.
Has no thing to do with my question, but since you asked, they are rx580s. Im not using them for gaming, at least until they have paid for themselves.
To game I'll just put my 7870 back in. It's worthless for mining.
You have to remember that PSUs are not 100% efficient. If the PSU powering those three cards is pulling 650W at the wall, then the cards are actually using around 520W assuming 80% efficiency.
Most reputable PSUs have protections that shut the supply down if an overload is detected, so you should be fine.
Something I've always wondered, the 850w rating is the output? As opposed to consumption? So realistically I could get at the wall readings up to almost 1000w without exceeding the 850w rating of the PSU?
Tsumi
[H]F Junkie
- Joined
- Mar 18, 2010
- Messages
- 13,755
Theoretically, yes. The PSU rating is the output rating, not the input. If you have a modern Seasonic, it's likely somewhere around 85-90% efficient, depending on model. Theoretically you can pull up to 1000 watts, assuming 85% efficiency. However, the kill-a-watt is not completely accurate.
As far as sustained load goes, Seasonic PSUs are generally rated for their load rating sustained in a 40 or 50 C environment.
As far as sustained load goes, Seasonic PSUs are generally rated for their load rating sustained in a 40 or 50 C environment.
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
You make a very good first point but the second one only works on a short circuit or major overload.
You can push a PSU to its absolute max such that the voltage rails start to sag but it wont shut down.
Also...
PSUs are only capable of their max stated output when brand new.
Over time and with use their max output drops as the capacitors age.
So maximum power now is higher than in a year.
And the harder you push a PSU the faster it wears out.
Pushing a PSU near its max output is a sure way to run out of power fast.
I buy a PSU that will supply double what my PC will need when not clocked.
By the time its clocked it will be using closer to 2/3 of the PSUs max power and only while gaming, not all the time. Years of safe use.
My AX750 is 7 years old running 24/7 at near idle and sees plenty of gaming or HQPlayer use (a good load on the GPU). I only need around 550W max.
If you use a PSU hard all the time you should seriously over rate it if you want it to last the warranty period.
You could use another PSU just for your gfx card.
Power that on before you power the PC up.
Reverse the process to turn it off.
Last edited:
Yeah it's a brand new seasonic Prime 80+ gold. Maybe platinum even. I bought it only planning to put 3 cards in it, but I don't want to lose hashrate and I'd like my gaming system back. I might have to break down and buy another higher rates psu
D
Deleted member 278999
Guest
I usually take the output and multiply it by the efficiency and stay under that travm
Yeah that's usually in the ballpark I prefer to be in too.
I should have bought a bigger PSU.
D
Deleted member 278999
Guest
Aye,
I never buy anything but Samsung SSD's. Even then, I've had one go bad and backup important stuff.
Hopefully you have a lightning fast connection and can grab everything off of Steam again quickly.
Definately do not. But I do have 3 PC's that I can steal the data from.
I think I might have lost the factory bios for one of my GPUs though..... Hoping I backed it up to the cloud.
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
You might be able to locate your bios here. A fantastic resource where people offer up their bios for a large collective.
https://www.techpowerup.com/vgabios/
D
Deleted member 278999
Guest
You can find it again. It's been a couple of years since I flashed my 980 Ti, but somewhere in out there in the great beyond is a very large repository of BIOS' for pretty much every GPU variant.
D
Deleted member 278999
Guest
I did something similar once.
Mine was a gaming PC. I Kill-O-Watted it, stress testing I could never pull more than 380-400W. So I put in a decent Seasonic 450W PSU.
It worked fine. For about 3 years.
I remembered there was an efficiency loss, and that was accounted for. What I didn't exactly account for was degradation over time.
If your only planning on doing it for a while, it will work. It won't be a permanent years-long solution though.
Mine was a gaming PC. I Kill-O-Watted it, stress testing I could never pull more than 380-400W. So I put in a decent Seasonic 450W PSU.
It worked fine. For about 3 years.
I remembered there was an efficiency loss, and that was accounted for. What I didn't exactly account for was degradation over time.
If your only planning on doing it for a while, it will work. It won't be a permanent years-long solution though.
Nenu said:
Complete nonsense.
When capacitors get old, they do not have a spigot that magically reduces the amount of current they can deliver. They do lose capacitance and have an increase in ESR, which causes the output to get increasingly out of regulation and can cause the PSU to fail. But this is completely fixable by recapping the power supply, which I've done with dozens of PSUs over the last 15 years to get another 7-10 years of service out of them. The key is you have to replace all of the capacitors, even the really tiny ones buried in the hard to reach areas, else you'll end up with whatever component it powers causing an avalanche of problems for the rest of the supply.
If you look at the ripple test here:
https://www.hardocp.com/article/2015/03/09/silverstone_olympia_1000w_power_supply_7_year_redux/5
You can see the capacitors have lost their ability to filter properly, which usually means a decrease in capacitance. The ESR probably didn't increase that much because in the load testing, the voltage regulation barely budged. If the ESR did increase substantially, it would throw the voltage regulation way off.
Nenu said:
Thanks for saving me the time of debunking your nonsense, you have a 7 year old PSU which you use near max capacity with no issues.
The only thing double or triple rating your PSU is going to do is waste money, capacitors are going to have the same age related issues no matter the capacity of the PSU you buy.
Active PFC, when it works properly, fakes the behavior of a resistor.
Input current should draw a sine wave, in time with the voltage sine.
Power available at any moment is proportional the square of a sine.
However, output loads require power all the time. Voltages are steady.
DC load current will vary, but this happens a much faster time scale
than the sine waves coming in.
Kirchoff's Current Law states that current going into a wire must equal
the current coming out. Unless you are doing something funny to store
and release electrons, which is exactly what the Bulk Capacitor does.
Lets say you are drawing 10A 110VAC from the wall (1100W supply).
In reality this is 14.4A of current ripple. Active PFC kicks the voltage
up to around 380VDC, and reduces the current ripple to around 4.2A.
This much current has to be stored in the cap and released from the
cap 120 times per second.
So, its that 450V horsekiller cap in the middle, yeah the big one. Has
the hardest job of it. Probably would be the first to show signs of age
and resistance to AC current. That 380V is not steady 380V, but has
the current ripple superimposed on both capacitance and resistance.
Voltage wobbling about the target by a wide margin is normal and
expected at this energy storage midpoint of the supply. But not too
much wobble. Cause the peak could exceed 450V and risk futher
damage, or dip too low to properly feed the output supply, or both.
The capacitance and low resistance required here are so great that
the job almost cannot be done with anything but an electrolytic cap.
Those definately have a limited lifespan, since the electrolyte must
remain wet.
I've seen a few low power supplies that have managed to replace
the bulk cap with a dry style. But the voltage rises to around 800V.
And we are only speaking of about 150W supplies now.
---
Using a supply below its rated value reduces the AC current draw,
and the ripple current in/out of the bulk cap. I'm not saying that cap
would dry out any slower, but at low load you might not notice there
was much of a problem. Till you add a Vega or something...
---
Energy stored in that cap could be as much as 450V*450V*470uF/2.
Thats around 48 Joules. About the same as being shot by an arrow.
Constantly being replenished, so you can be shot over and over.
I don't recommend you try to measure it.
Input current should draw a sine wave, in time with the voltage sine.
Power available at any moment is proportional the square of a sine.
However, output loads require power all the time. Voltages are steady.
DC load current will vary, but this happens a much faster time scale
than the sine waves coming in.
Kirchoff's Current Law states that current going into a wire must equal
the current coming out. Unless you are doing something funny to store
and release electrons, which is exactly what the Bulk Capacitor does.
Lets say you are drawing 10A 110VAC from the wall (1100W supply).
In reality this is 14.4A of current ripple. Active PFC kicks the voltage
up to around 380VDC, and reduces the current ripple to around 4.2A.
This much current has to be stored in the cap and released from the
cap 120 times per second.
So, its that 450V horsekiller cap in the middle, yeah the big one. Has
the hardest job of it. Probably would be the first to show signs of age
and resistance to AC current. That 380V is not steady 380V, but has
the current ripple superimposed on both capacitance and resistance.
Voltage wobbling about the target by a wide margin is normal and
expected at this energy storage midpoint of the supply. But not too
much wobble. Cause the peak could exceed 450V and risk futher
damage, or dip too low to properly feed the output supply, or both.
The capacitance and low resistance required here are so great that
the job almost cannot be done with anything but an electrolytic cap.
Those definately have a limited lifespan, since the electrolyte must
remain wet.
I've seen a few low power supplies that have managed to replace
the bulk cap with a dry style. But the voltage rises to around 800V.
And we are only speaking of about 150W supplies now.
---
Using a supply below its rated value reduces the AC current draw,
and the ripple current in/out of the bulk cap. I'm not saying that cap
would dry out any slower, but at low load you might not notice there
was much of a problem. Till you add a Vega or something...
---
Energy stored in that cap could be as much as 450V*450V*470uF/2.
Thats around 48 Joules. About the same as being shot by an arrow.
Constantly being replenished, so you can be shot over and over.
I don't recommend you try to measure it.
Last edited:
Nenu
[H]ardened
- Joined
- Apr 28, 2007
- Messages
- 20,315
Last edited:
Continued nonsense.