550w psu for 1080ti

[Bajeetas]

I just bought an ASUS strix 1080 ti, will a 550w power supply (https://antec.com/product/power/neo-eco-modular-550-v2) be sufficient? I also have a Ryzen 7 3700x

[Needfuldoer]

The 1080ti is a 250 watt card, and Nvidia recommended at least 600 watts for the Founders Edition. It might work on a 550 watt PSU, but you're pushing it.

 

https://www.nvidia.com/content/geforce-gtx/gtx_1080_ti_user_guide.pdf

 

I'd get at least a 650 or 700 watt power supply, that way there's no question you have enough power and you've got some headroom to upgrade in the future. 

 

What's your budget?

[Blue4130]

1 hour ago, 8tg said:

550w is peak, not sustained

That is incorrect. It will peak higher than that before the OCP trips.

[Somerandomtechyboi]

3 minutes ago, 8tg said:

tldr the labeling on power supplies anywhere other than mainland china is a safe peak output, its sustained load capability is usually lower, and since we dont know its usually just safe to assume 100 watts of headroom

even on psus where the 12v rail match the psus advertised wattage?

[Blue4130]

23 minutes ago, 8tg said:

Kinda. Thats still peak, not sustained, but then it gets into how load is handled across different rails.

and higher end power supplies the difference is smaller and smaller between rated peak and sustained, so that number becomes more and more accurate 

 

this is also why some psus have ratings that don’t add up, because their label is peak and their measured per rail is sustained 

there’s not really a consistent standard for this, it’s why we error on the side of caution with overestimates 

So what is your source for this?

[Blue4130]

9 minutes ago, 8tg said:

It’s mostly inference based on how power delivery in electronics works in general, it’s very closely related to RMS output vs peak output on high voltage systems or most commonly, speakers and amplifiers. The principle is the same, there’s a rating the hardware is designed to handle sustained, and a rating they can handle as a sudden temporary spike, in any metric, amperage, wattage or voltage.

with consumer computer power supplies it gets into a gray area because they don’t really list the output with an RMS vs peak metric, it’s just comparing what the label says to what it tests as.

But it can go both ways. Something like an RM850x can actually sustain 950 watts without issue, OCP triggers at 1030w. So it’s “safe peak” is somewhere between those two numbers, but it’s RMS is 950 watts, but it’s listed as an 850w psu presumably because they might not all do that consistently even if it’s tested as pretty common.

The inverse is the risk to avoid taking, PSUs that list their safe peak as their primary naming and metric, while their RMS is lower.

Thats like calling that RM850x the RM1000x. It can peak 1000w no problems, with an RMS of 950w though. 

Thats basically most lower end and some mid tier power supplies that name like that. It’s one of the things the old orange label VS series was known for, named for their peak, their RMS was 50-100 watts lower. So paired with almost no functional protections, orange label VS series toasted themselves a bunch with 560ti SLI or whatever nonsense was going on with fermi housefire budget builds.

Ok, I just read the official labeling specs that are defined by DOE for the USA (and followed by Canada and EU) To meet spec and pass certification to be sold in the USA, the label lists average. Not peak.

 

It's dry and not very exciting, but you can start the rabbit hole here if you want.

 

https://www.cui.com/efficiency-standards

[Somerandomtechyboi]

1 hour ago, 8tg said:

Thats basically most lower end and some mid tier power supplies that name like that. It’s one of the things the old orange label VS series was known for, named for their peak, their RMS was 50-100 watts lower. So paired with almost no functional protections, orange label VS series toasted themselves a bunch with 560ti SLI or whatever nonsense was going on with fermi housefire budget builds.

looking at the wattage table for the orange label vs550 it has a 12v rail for 504w, so its actually a <500w psu?

 

then the rm850x lists 70.8a on its 12v rail but it actually handles 950w?

 

i guess it really is inconsistent but it probably just means i should buy ones with 12v rail as close to advertised wattage as possible when im buying used psus, guess this also explains why my dazumba 450w (360w) shuts down even though im just doing 3000+ ram oc on x58 and no cpu oc

[Glenwing]

13 hours ago, 8tg said:

It'll peak higher but in the US/EU the way power supplies are named is pretty vague, but its generally its "reasonable peak" if youd want to call it that.

Like a 600w psu is really a 500-550w psu, because itll sustain lower without issue, but can also handle an ~600 watt peak without issue.

PC power supplies are rated with continuous output power, not peak output power. As stated by a previous poster, this is required by law in the US since the energy efficiency testing procedure detailed in 10 CFR 430 requires that a power supply be operated at 100 % of its nameplate power output for at least 30 minutes.

 

10 hours ago, 8tg said:

Thats basically most lower end and some mid tier power supplies that name like that. It’s one of the things the old orange label VS series was known for, named for their peak, their RMS was 50-100 watts lower.

You talking about this one?

https://www.corsair.com/ww/en/p/psu/cp-9020097-uk/vs-series-vs550-550-watt-power-supply-uk-cp-9020097-uk

 

 

The only inconsistency is whether manufacturers make the 12 V rail capable of the full nameplate power by itself or whether they add in the auxiliary rails to get to that number, which cheaper supplies often do (including the VS orange). But every power and current rating in these tables is for continuous power and continuous current, not peak.

 

I should also note there is no such thing as "RMS power". I suppose it can exist in the sense that mathematically you can take the root mean square of the power and you will get a number, but it won't represent anything useful. For voltage and current we use their RMS values instead of their averages when calculating power because power scales with the square of voltage or current, so the average voltage or average current is not useful for this. If you already have power, you don't then take the RMS of that. It's not representative of anything. If you want the average over time then you just take a normal average, not the RMS. The term "RMS power" exists only in audiophile articles and blog posts from people who don't know the meaning of the terms they are using, and are just using the term "RMS" to mean "average".

[Motifator]

Yeah lol that's not true at all... even an old school GOOD 550W PSU like this one 

https://www.techpowerup.com/review/kingwin-lzp-550/5.html

Is not only rated 550W CONTINOUS, but also can peak much higher. Like the LZP-550 could actually, partially act as a 600-650W PSU even though it was a 550W in reality.

Also it's not like a single 1080 Ti and a 3700X will ever pull 600W in reality lol, more like close to half of it while chilling around in games... I say he'd be fine.

[Mark Kaine]

On 5/29/2024 at 8:05 AM, 8tg said:

Kinda. Thats still peak, not sustained, but then it gets into how load is handled across different rails.

and higher end power supplies the difference is smaller and smaller between rated peak and sustained, so that number becomes more and more accurate 

 

this is also why some psus have ratings that don’t add up, because their label is peak and their measured per rail is sustained 

there’s not really a consistent standard for this, it’s why we error on the side of caution with overestimates 

stop shopping at ali express...

none of that is relevant for quality brand psus  - the advertised power is the minimum,  max is usually around 100w higher, end of story.