psu calculator. sites that have their personal interests vs honest sites

[igormp]

ATX spec 3.X are otherwise the spikes can trigger a conservative set OCP depending on the exact unit as some brands/oems did make changes to their line but did not necessarily label them so the brands could sell out old inventory

The old inventory issue is fair, but one can also do their due diligence and search for the most appropriate models, and avoid the models that had such transient issues.
Also since those issues took place I'm not aware of any new (so, actual new product name, instead of a new revision) good quality units that faced such transient problems, and we have tons of those new products (ATX 3.0 or not).

Once again with a ATX 3.X sure but you are seeing 1000w PSU being flat our recommend for gaming builds with no known hardware. You see "just get a 1000w unit and you are good" but if the person in question is using a CPU that tops out at 100w and a GPU that tops out at 150w than your 650w would be more then enough (even a 550w if no future upgrades will be done). Especially when a lot of these people don't understand the cost and follow someones link to a $180 unit only to see that Apevia 1000w unit for $90 and say "well that's a 1000w unit for half the cost, I'm getting that one!" when they actually could have gotten a quality 650w for that cost that would likely 1) last them longer and 2) perform better.

I agree on that, but that's an entirely different subject IMO. I believe most folks here would always go for a reasonable quality unit, be it 450 or 1500W.
I also expect users in this forum to be tech savvy enough to not blindly buy stuff without proper research/discussion beforehand.
Given that, calculating the minimum wattage needed should be obvious to anyone here (no matter if you use a calculator or not), and the amount of extra leeway is up to one's own personal preference (and how deep their pockets are), so it's moot to discuss that, given those assumptions.

 

[dirtyferret]

one can also do their due diligence and search for the most appropriate models

most everyone here has stated that multiple times

I'm not aware of any new

I'll occasionally see it on a reddit post or somewhere else, for the most part newer units deal better with modern GPUs although if rumors are to be believed in upcoming GPUs" all SKUs would have “some increase in power consumption, with higher SKUs increasing more"

I believe most folks here would always go for a reasonable quality unit

That entirely depends on the whose use of the definition "reasonable" we are using. I've yet to meet a single PC gamer who wasn't using a reasonable piece of hardware regardless of the hardware.

blindly buy stuff without proper research/discussion beforehand.

I think this would surprise you but in all honesty I don't care what anyone buys for themselves. You want a 1500w PSU to run an AMD APU system, knock yourself out and I'm all for it. If you state you need a 1000w PSU to run a system that will only pull 200-300w gaming than I have a major issue and yes I do see that and have seen it posted too many times on forums over the decades.

calculating the minimum wattage needed should be obvious to anyone here

should be but too often is not

amount of extra leeway is up to one's own personal preference

I agree, you want 1000w head room? sure. That 750w cost just $5 more than the 650w version, heck yeah get the 750w. You hate noise and want the 850w version just to make sure the PSU fan on semi passive doesn't kick in? Make sense to me. All good personal reasons yet none should be used to recommend a wattage to someone without first stating this is my personal reason for getting this unit and you may or may not agree with it. To often you see the goal post of accepted PC performance moved (shockingly ) to match the exact level of the person giving recommendations.

 

[igormp]

most everyone here has stated that multiple times

Welp, I guess no one disagrees with this idea lol

I'll occasionally see it on a reddit post or somewhere else

As in new models, that supposedly have good quality still facing those issues? If so, interesting and a bit worrying. Do you remember any of those models out of the top of your head?

although if rumors are to be believed in upcoming GPUs" all SKUs would have “some increase in power consumption, with higher SKUs increasing more"

Doesn't really change much about the whole transient spike scenario. A reasonable PSU should handle those. Newer GPUs also seem to be more tame in that regard, as in, the 4000 series had way more reasonable spikes than the 3000 series. I'd expect the 5000 series to follow suit.

I think this would surprise you but in all honesty I don't care what anyone buys for themselves. You want a 1500w PSU to run an AMD APU system, knock yourself out and I'm all for it. If you state you need a 1000w PSU to run a system that will only pull 200-300w gaming than I have a major issue and yes I do see that and have seen it posted too many times on forums over the decades.

You say you don't care, yet you say that you have a major issue? I don't get it.
I also don't get what this has to do with the original point of PSUs being able to handle transient spikes.

should be but too often is not

As you can see, I'm new to this forum. Should I just lower my expectations for the average user here in this case?

All good personal reasons yet none should be used to recommend a wattage to someone without first stating this is my personal reason for getting this unit and you may or may not agree with it.

Agreed, but so far in this topic all I've seen was people stating exactly what you said. Stuff like "I like to add this much extra overhead, me thing". No one is recommending PSUs here anyway, just stating how they like to pick theirs.

 

[dirtyferret]

Do you remember any of those models out of the top of your head?

I don't recall the exact models. I'll read a reddit post now and then but posting there, it's like bringing a globe to a flat earther convention.

Newer GPUs also seem to be more tame in that regard, as in, the 4000 series

you can see the tests I linked

You say you don't care, yet you say that you have a major issue? I don't get it.

I don't care what people buy, I do care what they recommend to people looking for information if they don't clearly state their reason for purchase.

As you can see, I'm new to this forum. Should I just lower my expectations for the average user here in this case?

I tend to give people the benefit of the doubt when it comes to their ability to turn on a computer, log in and type and from there we can work on future expectations

No one is recommending PSUs here anyway, just stating how they like to pick theirs.

Some did, some came to see if they could get Dr. Dro to Venmo money to them

 

[igormp]

I don't care what people buy, I do care what they recommend to people looking for information if they don't clearly state their reason for purchase.

ah, I guess you meant if ones "states to another". I assumed it was just one saying what they prefer. Got it (and agreed).

I tend to give people the benefit of the doubt when it comes to their ability to turn on a computer, log in and type and from there we can work on future expectations

I expected this forum to be somewhat made of tech-savvy members that can do some basic research (apart from the odd one), I'll keep this in mind to see if I need to recalibrate my expectations.
In all honesty, I expected the average user here to be a bit more knowledgeable than, let's say, a LTT average forum user.

you can see the tests I linked

They do point to the 4090 being more tame than the 3090ti, even though the 4090 has a higher power limit.
This goes in hand with what I saw from igor's lab.

 

[Chrispy_]

Rule of thumb for a gaming system; Add your CPU PPT and GPU TDP together, then double it.

If I have a 142W Ryzen and a 220W 4070S, then I'm getting a "724W PSU".

That's:
~150W for the 20ms spikes from the Ryzen (the RAM is powered by this too)
~300W for the 20ms spikes from the 4070S
~40W for motherboard, fans, SSDs, LEDs
for a total of 490W, which is comfortably less than the 724W by the rule of thumb, and that system is going to use about 340W in a typical gaming scenario which is close to the peak efficiency of 50% of a hypothetical 724W PSU

The rule of thumb also works for power-spiky systems using badly-behaved components. A 13900K and 3090 are notorious for power draw spikes grossly exceeding their PPTs. A 253W CPU and 350W 3090 shouldn't need more than a 650W PSU - but both components are greedy, spiky, hungry little bastards. Let's check the rule of thumb...

(253+350)*2= a "1206W PSU".

Realisically,
~360W for the 20ms spikes from the 13900K
~485W for the 20ms spikes from the 3090
~40W for motherboard, fans, SSDs, LEDs
for a total of 885W, plenty of headroom left in the rule-of-thumb 1206W PSU, and that system will likely use about 550W in a typical gaming scenario which is pretty close to the peak efficiency at 50% of the hypothetical 1206W.

So yeah, double your CPU PPT and GPU TDP hasn't failed me yet. Maybe in the future PSUs will be efficient at >80% loads, and CPUs/GPUs won't have these massive power spikes that can trip OCP or cause 12V rails to droop enough to crash your system. Until then, we need rules of thumb.

 

[Frick]

LOL I think it so funny, in a sad, if not pathetic way, when someone claims it is a matter of semantics after they are called out for being technically incorrect. This isn't a Facebook chat page. This is TPU, a "technical" support site. We should be technically correct when posting technical information. That is me being me, and proud of it.

And for the record, claiming someone said, or meant something they didn't say or mean is NOT a matter of semantics. It is misrepresenting what they said, often innocently, but, sadly, often intentionally too. And when intentional, it is often with mal- or deceptive intent. In any case, it is not cool.

Yes, I said researching all components takes a long time - but you decided to leave out where I prefaced that by saying,


What I meant is what I said - much of which you decided to leave out of your reply - this after you claimed you didn't want to get into this, then jumped in.

Yes, I also said the eOVC (eXtreme OuterVision Calculator) is "the most conservative" when padding the results. But note it still pads the results! And I said, "and that's a good thing!" and I explained why. But again, you decided to leave that out in your "not getting into it" replies.

Also left out is where I clearly recommended users set "CPU Utilization to 100%" and "Computer Utilization Time to 16 hours per day" to pad the results even more.

So for the umptenth time, since you and others clearly believe you can do a better job of researching and calculating your power needs (and if so, that's great!), then don't use it! I don't care.

I mean I've read that post several times and I do get your point, but I a) think you're really fun to interact with and b) I still read that post as you saying "trust this calculator"*, which there's now examples for when you can't.

And just for the record, not every computer user uses a monster graphics card with outrageous, instantaneous power demands. In fact, not every computer user even uses a graphics card.

And here it is.

NO WHERE, NOT ONCE did I ever say users cannot add extra padding when purchasing, if they so desire. And NO WHERE did I say to "only" or "just" use "a" calculator, or even "the" eOVC, as I was falsely accused of when you were not jumping into this.

And here it is again. I mean ... you implied "use this calculator it's good and accurate" and people said "not always" and off you went and when given examples you say "Well you don't HAVE TO USE IT" and "not everyone uses GPUs" and as always we end up agreeing but you're being weird about it.

*THIS is what I mean about semantics**. You write things which sometimes don't make sense (like "This will calculate your minimum needs and recommend a suitable size for those needs." What does that even mean? Are minimum power supply needs different from maximum power supply needs?) and then when miscommunication invariably happens you at least write like you get hyper upset and call people pathetic and morons and you can never, ever be wrong, it's always the other person who is pathetic or a moron. This was just an example btw.

**DO NOTE english is not my first language so I might be using that word wrong.

 

[Bill_Bright]

But you are not wanting to argue. Got it.

I am really sorry for those of you who can't seem to comprehend how the eXtreme Outervision Calculator can be a useful "tool" . Or worse, how others can use it as a valuable tool to help them determine the "minimum" PSU size one needs.

That's fine if you can't see the benefit to you, but please don't assume others can't figure out how it can be useful for them.

If you wish to research each component yourselves, then double or triple the results, that's perfectly fine! I really truly don't have a problem with that. As I noted, there's no risk to your components in going with way too big a supply.

In fact, I used to do all that research myself years ago. Well, I didn't double or triple but I typically would add 100 watts for good measure. Then I figured out the eXtreme Outervision Calculator will do most of that work for me, including padding the results. I might still add a 100 - but that would depend on possible future upgrades and such.

And for those of you who just want to guess, then double or triple, that's fine too. But again, but please don't assume others can't figure out how to use the eXtreme Outervision Calculator to figure out their minimum requirements, for again, that just is not true. Please stop assume everyone is dumber than you.

One last thing I will note. It seems many of those who think all PSU calculators, including the eXtreme Outervision Calculator are worthless, please stop assuming every user is using the most power hungry graphics cards out there. That simply is not true either - not by a long shot.

 

[ShrimpBrime]

Rule of thumb for a gaming system; Add your CPU PPT and GPU TDP together, then double it.

If I have a 142W Ryzen and a 220W 4070S, then I'm getting a "724W PSU".

That's:
~150W for the 20ms spikes from the Ryzen (the RAM is powered by this too)
~300W for the 20ms spikes from the 4070S
~40W for motherboard, fans, SSDs, LEDs
for a total of 490W, which is comfortably less than the 724W by the rule of thumb, and that system is going to use about 340W in a typical gaming scenario which is close to the peak efficiency of 50% of a hypothetical 724W PSU

The rule of thumb also works for power-spiky systems using badly-behaved components. A 13900K and 3090 are notorious for power draw spikes grossly exceeding their PPTs. A 253W CPU and 350W 3090 shouldn't need more than a 650W PSU - but both components are greedy, spiky, hungry little bastards. Let's check the rule of thumb...

(253+350)*2= a "1206W PSU".

Realisically,
~360W for the 20ms spikes from the 13900K
~485W for the 20ms spikes from the 3090
~40W for motherboard, fans, SSDs, LEDs
for a total of 885W, plenty of headroom left in the rule-of-thumb 1206W PSU, and that system will likely use about 550W in a typical gaming scenario which is pretty close to the peak efficiency at 50% of the hypothetical 1206W.

So yeah, double your CPU PPT and GPU TDP hasn't failed me yet. Maybe in the future PSUs will be efficient at >80% loads, and CPUs/GPUs won't have these massive power spikes that can trip OCP or cause 12V rails to droop enough to crash your system. Until then, we need rules of thumb.

1kw has covered me very well. Imagine 13700k 2x 290x gpus draw is tremendous. I melted (real close to fire) the power cord to the PSU. 30a cord wasn't big enough.

PSU, I retired it shortly after. OCP didn't trip, and I'm not sure it even went over. But it smoked a 30a 110v cord, so I figured just in case, better junk it.

 

[Shrek]

Confused... you were running a 30A cable at less than 10A where the cable heating was 1/10th of what it could handle?

 

[bebealex]

Something that I'm quite surprised to not have seen, a UPS can be a great long-term protection. Saved me many times with the 5 minutes of battery power and allows me to use a vaccum per say on the same circuit as the computer's outlet is on.

 

[ShrimpBrime]

Confused... you were running a 30A cable at less than 10A where the cable heating was 1/10th of what it could handle?

Defective cord no doubt. It must have gotten warm, then hot and resistance went through the roof. During a TimeSpy run I think it was.

It was actually a 390 paired with a 290x.

So 390 about 300w. The 290x over 300w. These are no problem figures for these cards. 13700K we'll just guesstimate 275w.



Fairly certain I was well over 10a though.

 

[Dr. Dro]

Defective cord no doubt. It must have gotten warm, then hot and resistance went through the roof. During a TimeSpy run I think it was.

It was actually a 390 paired with a 290x. Happened during TimeSpy.

So 390 about 300w. The 290x over 300w. These are no problem figures for these cards. 13700K we'll just guesstimate 275w.



Fairly certain I was well over 10a though.

View attachment 362514

Hawaii really chugs. I miss my 290X Lightning

 

[ShrimpBrime]

Hawaii really chugs. I miss my 290X Lightning

I miss em' too, but can't hang onto stuff. Selling helps buying.

 

[Ruru]

Do NOT, under any circumstances, skimp on the power supply. Don't even think about it. Spend the extra $50, $70, whatever now to avoid having to potentially replace a very expensive component, or the entire system, in the future.

Still remember when I used a R9 290 with a (good quality) 430W unit though the card was a good undervolter which may explain why it worked like a charm.

 

[Bill_Bright]

Something that I'm quite surprised to not have seen, a UPS can be a great long-term protection.

What do you mean by "surprised to not have seen"? There have been many many discussions about UPSs on this site. And while this thread is NOT about using a good UPS, it was noted way back in post #7 how the eXtreme Outervision Calculator will even recommend a suitable size UPS too (for the inputted components - the user must, of course, add the power requirements for the monitor, network devices, etc. manually for other connected devices).

 

[bebealex]

What do you mean by "surprised to not have seen"? There have been many many discussions about UPSs on this site. And while this thread is NOT about using a good UPS, it was noted way back in post #7 how the eXtreme Outervision Calculator will even recommend a suitable size UPS too (for the inputted components - the user must, of course, add the power requirements for the monitor, network devices, etc. manually for other connected devices).

Oh sorry ! I meant in the actual post, it does seem that you're the only one who suggested it. Altough I'm still a bit sad not seeing it suggested in more answers.

Thanks and a great day,
Alex

 

[Bill_Bright]

Altough I'm still a bit sad not seeing it suggested in more answers.

Well, this thread is about power supply calculators, not UPS. So it makes sense, actually, UPS are not mentioned.

 

[cvaldes]

Since I don't bother overclocking anything anymore and I don't have some crazy configuration, I just look at the GPU manufacturer's PSU recommendation, leave it at that and move on with my life. After all, GPUs tend to be the biggest power-hog components in your typical consumer enthusiast system; address GPU power needs is generally good enough to cover pretty much everything else in all but the weirdest builds.

In 2024 it's more important than ever to emphasize that some people in online PC forums have a tendency to say things that really reflect corporate IT "best practice" ideology rather than sensible everyday consumer-grade advice.

For Joe Consumer, it's unrealistic for them to take measurements of all system components using a bunch of tools that they might not own (physical instruments) as well as software tools (which they might not understand to the extent necessary to make critical decisions).

And how does Joe Consumer know for sure that PSU A provides all of the right power at the right levels? They don't. Unless you build/maintain PCs on a very regular basis (like some corporate IT staffer or a PC repair shop technician), there's pretty no way to know if any of the numbers you read in some measurement tool are correct.

This is why I don't read PSU reviews unless I'm in the market to buy one. I simply don't have the background to understand the pros and cons of any given PSU reviewer's methodology. Realistically I'd have a bachelor's degree in electrical engineering (or the equivalent in real world experience) but I don't.

 

[Bill_Bright]

In 2024 it's more important than ever to emphasize that some people in online PC forums have a tendency to say things that really reflect corporate IT "best practice" ideology rather than sensible everyday consumer-grade advice.

Or they just parrot what they heard someone else say rather than do the actual research themselves. Or they assume they represent the norm when in fact, most users, even most PS builders don't invest $500 (or more) in a graphics card. Most don't buy i7s or i9 or AMD equivalents.

 

[wheresmycar]

Personally, i've never used online power consumption calculators and i've been happy with PCPPs conservative estimates. For good quality power supplies, PCPP estimates are overestimates and that doesn't stop me from throwing on another 100-200w on top for some long-term/upgrade piece of mind. I'm still on ATX 2.0 hence that added headroom comes handy for better handling power spikes/unexpected loads.

BTW, for a long time i've been under the impression that we hit the "most efficient" sweet spot with loads not exceeding 2/3rds (or so) of the units maximum capacity. Where are we on that with modern power supply units and ATX 3.0?

 

[dirtyferret]

Where are we on that with modern power supply units and ATX 3.0?

depends on the exact unit but the better modern platform units have flatter efficiency curves than on older DC-DC platforms
ATX 3.x has more to do with GPU power connectors and increase transient load than it does efficiency

 

[wheresmycar]

depends on the exact unit but the better modern platform units have flatter efficiency curves than on older DC-DC platforms
ATX 3.x has more to do with GPU power connectors and increase transient load than it does efficiency

that makes sense. I was thinking more about efficiency curves in general and how much they've improved with good quality mainstream units. Would you say the 50-70% load still achieves the best efficiency or has the goal posts improved dramatically? Perhaps there was a time when efficiency sweet spots were relevant or more-so with low quality units but not keeping tabs on how things have shaped up over time somehow sprung up the "want" to know a little more.

 

[Bill_Bright]

Personally, i've never used online power consumption calculators and i've been happy with PCPPs conservative estimates.

How do you think PCPP achieves their estimates? I'll give you a hint. It is on-line and it estimates power consumption.

I have quite a bit of experience with their "Estimated Wattage" feature. I would hardly call it "conservative". That is NOT necessarily a bad thing for once again, going too big hurts nothing but, maybe the wallet or budget. But make no mistake, that estimator is an "online power consumption calculator".

I will say, through my experience, it typically does not go WAY overboard. But to call it conservative? Nope.

 

[dirtyferret]

I was thinking more about efficiency curves in general and how much they've improved with good quality mainstream units. Would you say the 50-70% load still achieves the best efficiency or has the goal posts improved dramatically?

I mean once again it depends on the exact units but in general modern platform units are better. Here is the NZXT C1200 with more of a flat solid numbers across the board you tend to see now

100W	89.2
200W	90.5
400W	91.4
600W	93.2
800W	92.1
1000W	91.4
1200W	90.3