Bitcoin Forum

I am running multiple 1200 wat corsairs HX1200i and one 1000 watt RM1000i.

Is it safe to run the 1000 unit with an outup of 950 watt (input 1040) and the 1200 units with an output of 1150 (input 1250)?

of course 24/7.

I understand psu ratings is for Output and not Input so I assume I will be safe, correct?

1000 watts psu means you can use up to 1100 watts at wall. They were built to use 1100 watts at wall 50c, server is 24/7, desktop psus don't think they withstand 24/7 using 1100w at wall for a long time, so for 24/7, i'd say 10% to 20% less watts is safe to be used 24/7.

Safe is a strange word to use in this context, because I'm honestly unsure as to what you mean. I don't think it will cause any harm to your hardware, if that is what you are asking. I had someone once that their ground went bad in the outlet they were sourcing power to their rig from and overnight they shorted out every card, circuit, accessory and anything attached to the rig. They are lucky it did not start a fire, but even the metal keyboard was outputting high voltage and shocking anyone that touched it. There are unforseen problems that nobody can predict, so "safe" may not be the most appropriate word in this circumstance.

Safe in electronics means "well within compliance" which means only a disaster would do any harm which given the probability, one in billions of cases.

Listen to Metroid. he know gpus will die this year.


The real deal is those psu will die in under a year the way you use them.

But WTF you have a warranty so why not?

I am being a bit nasty cause in my opinion you the op are not using your gear properly.

My opinion is the 1200 watt should read 900-1000 at the wall on a k watt meter

and the 1000 watt should read 700-800 at the wall on a k watt meter


killing gear and looking to get rmas hurts the gpu mining industry.

it also hurts the tax psu industry.


but safe as in burn down your home most likely your home will not burn down  .  your psu will die in a shorter time period and you burn more power at the higher  capacity

Efficiency means how much power from the wall would be drawn by the PSU at a current output power.
You must give more power spare at your PSU. Look what I've done with my RM1000i:

---

https://i.imgur.com/7V8rBWJ.jpg

---

In certain situations, your cards will draw more power from your PSU, ex. When your cards adjust fan speed to keep temperature, and it will draw more power.

Ahh, well if this is the definition we're working with in this context, then I would say that you are "safe" OP, for sure. As philipma has said, I doubt your house will burn down; I am more concerned about what philipma has said about GPUs dying this year, what exactly do you mean by that? Do you mean that the prices will become too high, or the profitibility will become too low or something else entirely?

It's never good to run gear at near full load, and I almost never push any of my gear past 75%.

Dave's opinions:

Corsair no, not safe to keep running near max.

Other better brands yes they are. Where you have them is on the border of what I would do with the ones I use, perhaps a bit over.
I have had EVGA 1300 units running @ close to 92% rated for years now with no issues.
Same with some of the earlier Seasonic

Corsair I will not run above 75%. Just have not had good luck with them in PCs or in mining.

-Dave

To be on the safe side i never max anything out.  Maybe just too cautious but when you got to ask the question or think i hope this doesnt cause any harm, i likely avoid the situation and go with a higher rated and wattage capable psu.  Why risk anything

If you plan to run them that hard you should have some spare ones on deck to swap out while you wait for your RMA to process.  Otherwise go with the 80% of max wattage rule to be safe.

I've ever tested a rig with 6 x MSI 1070 Seahawk at 100% power limit (full load), Kill a Watt indicated that the total power consumption is 1080W, and the PSU I used is Super Flower 1200W. It run only in about 2 hours than the rig crashed. I think you should not push your PSU too much, I suggest the safe operating range is <80% power limit of your PSU, because we're talking about 24/7, and the thing is PSU's efficiency is getting worse during your mining. For me, I set all of my rigs mine at 75% power limit.   

I am quoting Metroid he has a thread  that claims  gpus will die in 2018.

Back to this thread.

 I have run

seasonic
evga
corsair
thermaltake
fractal
antec
rosewill

all of them don't like to be maxed.

all of them die in under 5 years when maxed.

All of them  run better at 70%  and have less issues.

it is best to run at 60-85% of total power

I would not take this as gospel by any means. Corsair is a quality company, IMO; what sort of information are you working with besides anecdotal experiences to come to that conclusion? I can understand having a personal preference, but this seems misleading in the worst way.

You are overdue for a failure with your EVGA with the way you run it and I hope you don't then assume EVGA is a bad company, too. I wouldn't run anything above 75%, because I like some equipment to last and I treat it with some respect.

they all break when pushed hard.

Evga 1600 p2 melted running an s-9 doing 1350 watts 24/7/365 last 2 years 9 months

Seasonics
1200 plat
1000 plat   melted

Evga 750 t2 died

Rosewill 1500 watt tokamak    running at 1250  for 2 1500s
rosewill  1200 watt tokamak    all popped caps with a boom  running at 1000 for the 1200

fractal  1000 watt plat  90 days dead running at 800 watts all 3 of them

I have yet to kill about 15 corsairs but I have since learned and run them more gentle.

Wow...Phill
You have a lot experience with many PSUs brand. Yeah, agree, whatever its brands, when pushing to the limit would break faster. And some brand has excellent durability when given a bit nor middle push, and the other would get a bit damage such as melted.

man, I'd say this is bad idea to run that power supply with that power levels.
this will not end well. first I guess your rigs will start to shutdown and then start over - this will be the first sign your psus are going to die!
and if you will not change the load - they will burn out, and it is good if it will be the PSU only, cause they can affect mb and gpus as well
my advice, give them air to breathe and make the load easier

Can you share which specific model of Corsair PSU are you talking about?

I run a bunch of Corsair 1000's and have not any any issues with them. I don't push them as hard as you though, and I keep them to a maximum of 800-850 watts on input, meaning they are probably delivering 750-800 output. I also run them at 240V input, so this is a little easier on them as the input current is slightly less and PSU efficiency is slightly better.

For the OP pushing the input over 1040 watts and 950 output, while you may technically be within specifications, I think consumer PSUs are designed more for the 1000 watts to be a maximum surge current, and not a 24/7 constant load. I would look at reducing the load on it if you expect it to last and to also minimize fire risks.

Thanks everybody for the heads up.

I am on 240 V power so efficiency is  90+ even on these loads.


I will replace this 1000 unit with a 1200 one so it should be ok.


Regarding my 1200 units I suppose I should run them up to 1000-1050 input?

I am running one at 1200 input -1100 output,, do you guys think the 1300 G2 from EVGA will be significantly better than my RX1200i from corsair?

that's better


It would be safer and would keep both of your device (PSU and GPU)  and another part also in good condition, at least for an extended period.


With 1200 input and 1100 output, of course, you should choose Evga with 1300 Wattage.

I am working with only like 20 or so typical 8 gpu rigs. And PSU is the unit that dies most often. Probably because I do push those to 90-95% but probably not.
Out of 20 rigs I had
Corsair hx850 - clicking sound from psu not working
corsair 1500w - 2 units - both shut off even at 60% load - dead on arrival
evga 1000w dead fan - could be my fault
seasonic 1000w - burned sata cable - my stupidity - 3 risers on one sata cable
most of the units died because my fault or came to me doa.

I talked about this on jonnyguru forums and came to conclusion that using psus on near max load is ok if the cooling is OK and if the psu is high quality. So maybe in summer ill lower the load.

For normal uses it's fine. It would be different for 24/7-365. But I absolutely agree with "to keep PSUs cold", that's why I'm adding more cooling systems near my rig. I'm using Peltier since Air Conditioner isn't my options, lol.

---

https://i.imgur.com/jttuNwG.jpg

---

To take a page out of Metroid's book... "PSU's will die in 2018".

Death, it seems, is coming to a lot of electronics this year, according to the great oracle that we know as "Metroid". ;D

https://icondoit.files.wordpress.com/2010/01/burn-baby-burn.png

Which was why I started the post (part that you didn't quote) with "Dave's opinions"

None of us can really speak to this beyond our own experiences and what we see here.

Would it have been better if instead of "opinion" I had said, in my limited testing of less then 100 units in mining EVGA and seasonic tend to seem to be able to take more abuse then corsair.

And in my real job of IT work, with countless power supply replacements over the last 2 decades. The generic / OEM power supply units have the biggest pile in the recycling box with corsair running a distant second but still many more then any other brand.

Once again, nothing I can prove that matters. Is corsair #2 in dead units because they are #1 in replacement power supplies so by nature I am going to see more? I don't know. All you and I and probably most people here can give is opinion.

Unless you want to go out and get a few 1000 of each brand and run them at 90+% in a controlled environment for years and keep track of how long they last. Because in test sample size less then that, it's all just opinion....

What I can *also* say is that in very limited times I have done the RMA myself, EVGA is easier to deal with then anyone else. They are also the slowest.

-Dave

Not really a good idea. Although they are capable of these loads leaving 15-20% gap between rated and actual output helps your PSU to work long without any issues. Stays way cooler that way, too.

And 1 more thing that I do not see stressed enough.

BAD / dirty / not stable 110/220v *input* power will cause just as much (if not more) damage to PSUs then running them at a high percentage of the max output.

-Dave

OP, what many haven't told you is that you are at the threshold where you should seriously look into server PSUs. While EVGA and Corsair are good brands, the fact remains they are consumer PSUs not meant for 24/7 operation at or near 100% capacity.

Now this is not to say that they cannot perform this function, as many miners certainly utilize them in this manner and they are both quality units manufactured to last, it is just that wasn't their design purpose. Most of the specs are for peak loads and the actual assumption is they would normally be running closer to 50%.

Server PSUs on the other hand are definitely designed to be installed in equipment that is going to run 24/7 and probably running closer to its max rating. They will use more robust components, but they will require 240V and are very loud since they are designed for data-center usage.

There are several companies selling these (used ones are significantly cheaper than new) along with various breakout boards and even pico units to plug feed your motherboard 24-pin ATX connector. The cost of the breakout boards, cables, pico unit, and used PSU are roughly that of a new consumer PSU, sometimes even less. Also, one benefit is if the actual server PSU dies, the rest of your investment is still safe as you can simply pickup another $25 PSU as your breakout boards, etc. are still good.

One company that sells these is Parallel Miner (https://www.parallelminer.com/), but there are others with a quick google search.

It´s easy,
Normally the points of maximum efficiency are those of greater stability and duration.

https://www.adslzone.net/app/uploads/2016/06/80plus-en.png

Sorry, i don´t know how to upload the img file.

fixed it

Thanks, i´m not a lab, but i have almost 200 psus, delta, corsair, seasonic, evga etc. Under 18º celcius you can go up to 55 to 60%, over 18º it´s better to stay near to 50%-55% to keep it at their best efficient level. The problem isn´t only the power efficient, there are many factors that becomes criticals when you go to the limits, THR, harmonics, power factors, etc...

That's a very risky move. See, server PSUs don't have any built-in protection, so when there's suddenly any short etc in your GPU/MB they will continue pumping current into it. "Home" PSUs will shut down automatically.

Well, that they have no protection is not entirely true. The problem is that if you use a PSU that can support 200 Amps in 12 v, it is really hard for a short circuit of one MB to be so strong as to exceed the limits of those PSUs.

I have almost 30 Delta DSP 2400 working 24/7, at 50% load are almost undetectable under 22º celcius, over that the start to make some noise.

OK, not entirely but mostly, then :) I tried to use server PSUs a year or so ago then was annoyed with their noise plus saw a number of very bad examples.

For the server PSU, I only run them at 60% capacity. Otherwise, it is too noisy.

For one I seen many consumer grade, or "home" as you put it, PSUs that didn't shutdown automatically, and some of the cheaper ones actually caused the fire to begin with.

Please share a link where you can show that server PSUs do not have built in protection if you are going to make that claim. Most server PSUs retail for $500 on up when new and I doubt they have any less protection than the far less costly home units. I sure as hell don't want my $7k+ server busting into flames any more than you do your home computer.

Secondly, once you go the route of server PSUs, you almost certainly are going to be running many of them and with 240V power. Since you will be running 20 or 30 amp circuits, you are going to need a PDU to break this power out and distribute it to many different rigs. These PDUs range from low cost to expensive units, but they all come with built-in protection as well, that usually limits the maximum current each outlet on the PDU can deliver, which is a fraction of the units total current rating. So no, a server PSU isn't anymore likely to continue pumping current into anything than a consumer PSU is, both because of its design and because of the extra protection often placed in front of it.

Also when a short circuit situation does occur, no matter the wattage, or current available, a voltage drop will occur and it is this that the PSU circuitry detects, when this goes outside of acceptance limits it shuts the unit down, whether a home or server PSU. Can this circuitry fail, yes, but again I would wager most components in a consumer versions are actually less reliable and robust than those found in a server counterpart.

Mining can be a dangerous activity if done improperly, this is the case whether with a consumer or server grade PSU, but just because you are not familiar with one aspect of it don't go around spreading misinformation about the subject.

Thanks for detailed post, I'll stand corrected. When few years ago I was thinking of server PSUs I researched tons of forums and videos and I *did* see a few mentioning this issue. Just spent another half an hour seraching Youtube and this time didn't find any. Could they be El Cheapo Chinese knock-offs? Maybe, who knows. Anyway, as I said above, even if protection isn't an issue I was very disappointed by noise levels. My rigs are in my small house and living in a server room never appealed to my wife (nor to me tbh). I know people modify them to make them much quieter but still, they're loud. Size also matters. So back then I made my choice and am with quality ATX PSUs never loading them over 70-80%. Quiet, cool and don't take much space, although cost is higher.