Which PSU Should I Get?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
Supernova G6 is not better than RMx.

On par with G3, slightly better than GT and a lot better than G5

Buget Gold unit of the Supernova series. Better than G+, GQ and GA

PSU with outdated ACRF platform. Not worth to consider.

Same comment as G+

Good unit

HEC platform is all we know. No review available so this psu a big question mark.

Like any other brands EVGA has good PSUs but also have some crap.
Thanks! So, of the good ones above, are any of them better than the RM1000x?

Which 1000/1200W PSUs from which brands would you recommend?
 

DSzymborski

Titan
Moderator
What brands do you recommend for 1000w or 1200w?
Some EVGAs, the Corsairs we've already talked about. SeaSonic has a lot of good PSUs.

The old tier list is still useful.

 
Reactions: ReveurGAM

--SID--

Estimable
So, of the good ones above, are any of them better than the RM1000x?
Not really. Supernova P2 is about the same level.
Which 1000/1200W PSUs from which brands would you recommend?
If I need a 1200w PSU atm I would wait for ATX3.0 PSUs. The Thermaltake Toughpower GF3 for example. A very good PSU that comes in flavours up to 1650w. The 1200w is allready avaulable here (NL) for €239. Also other brands would come with ATX3.0 PSUs like MSI and Seasonic.
 
Reactions: ReveurGAM

Karadjgne

Titan
Ambassador
It's Really hard to try and figure out psu's that are actually 'better' than the RMx in its class, or the HXi in its class of psu, most have at least 1 shortcoming, a 'deal breaker', that puts them as a lesser psu, even if some specs are actually better. The RMx, HXi are just plain Solid, all around.

Seasonic and SuperFlower both have psus that rank very high, and in many respects could be considered better, but have noisy fans or shorter hold-up times etc.
 
Reactions: ReveurGAM

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
Use this one instead https://cultists.network/9412/psu-tier-list-was-updated-to-rev-16-1a/ It's the updated version from the same authors.
Thank you. I feel stupid asking this, but where are the green/gold, blue, white colors shown? I've seen nothing on the list tabs that just uses the 4 colors, but maybe it's so obvious I'm overlooking it.

It's Really hard to try and figure out psu's that are actually 'better' than the RMx in its class, or the HXi in its class of psu, most have at least 1 shortcoming, a 'deal breaker', that puts them as a lesser psu, even if some specs are actually better. The RMx, HXi are just plain Solid, all around.

Seasonic and SuperFlower both have psus that rank very high, and in many respects could be considered better, but have noisy fans or shorter hold-up times etc.
I'm not even sure which tier those two Corsairs are in, but it sounds like you're saying they are holy grails of PSUs. ;)
 

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
1) I guess it's important for me to know, given my setup, if y'all would recommend 1000w or 1200w if I plan to add additional RAM & SSDs (and an AIO if Noctua's U12A can't cope)? I have no need for the two-bay HDD hidden down where the PSU goes. I don't plan to overclock or do anything heavy-duty like CAD or 3d videos. I'd rather spend a little bit less, but only if it's not going to result in additional risk to my PC.

My setup:
  • Corsair iCUE 5000x case (225mm max PSU length)
  • Intel i7-12700k CPU
  • Noctua NH-U12A CPU air tower cooler
  • ASUS TUF Gaming H670-Pro w/wifi D4
  • Patriot 64GB Viper Steel DDR4 3600 MHz UDIMM Memory Kit (2 x 32GB) PVS464G360C8K (not yet tested by ASUS) (will probably add more RAM later)
  • T-Create Classic 1 TB PCIe 4x4 M.2 NVMe SSD TM8FPH001T0C611 (will probably add a bigger one)
  • XFX Speedster Swift 319 RX 6800 XT 16 GB Model RX-68XTAQ
  • I'll be upgrading from my Dell P2411H monitor at a later date, and using a DP to VGA adapter until then.
  • I also have external USB drives for backing up data I don't need to access regularly, the largest (Seagate 6TB Expansion) are self-powered, so their draw through USB would be the tiny amount needed for data transfer, I'd think. Those would be the only ones I'd plug in on a regular basis as the USB-powered ones are already full.
Some EVGAs, the Corsairs we've already talked about. SeaSonic has a lot of good PSUs.
2) Would you be referring to EVGA T2, G7, G6 >=850W, and/or P6 >=850W?

If I understand correctly, the Corsair AX Tit., HX Plat, HXi 2022, RMx, & RMi, and the be quiet! Pro 12 are the best for multi-/+switchable 12V rails. I'm only interested in 10+yr warranties, so I didn't list the Enermax Revolution DF, for example.
2a) If 1200w, the tier list suggests that I should only get switchable or multi-rail...? It's my understanding that fully modular is best...Thoughts/suggestions?

If y'all tell me that single rail is perfectly fine at 1000-1200W, then I'll look at that part of the list. Thanks!
 

Karadjgne

Titan
Ambassador
A rail is a circuit that outputs power. Simply, In a single rail psu, there's one giant capacitor, one giant circuit, rated for the entire load. In a multi rail there's 2-4 capacitors, smaller and in parallel splitting the load.
When talking about 1000w+ psus, a single rail design would require a huge capacitor, huge wires, and that takes up a lot of space, 2x 500w capacitors are easier to manage physically, so multi rail makes more sense. But all that's internal, doesn't really affect what's on the outside.

Modular means every cable is connectable, has a plug. An advantage is you only plug in what you need, the wiring can all be done during the build, the psu swapped easily, at any time. But, every connection is a potential weak point with electricity and modular plugs often get stressed with corners or shoving wires under shrouds etc.

Semi-modular has certain wires permanently attached. 20+4 mains and eps. The pcie and Sata are modular. You need eps and mains always, so eliminates potential weak point there, but is still modular in the accessory wiring. But, there's no easy wiring like with full modular, the psu must be built into the pc as the EPS and mains are permanently attached.

Non-modular has all wires permanently attached. No potential weak points from a connector. But, you have to deal with all that wiring whether used or not. Has none of modular or semi advantages.

Modular, semi, non modular only affects how you build the pc or tear it down and related aspects, doesn't affect how it operates. Rails affect the psu operation, not how you build the pc.
 
Reactions: ReveurGAM

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
A rail is a circuit that outputs power. Simply, In a single rail psu, there's one giant capacitor, one giant circuit, rated for the entire load. In a multi rail there's 2-4 capacitors, smaller and in parallel splitting the load.
When talking about 1000w+ psus, a single rail design would require a huge capacitor, huge wires, and that takes up a lot of space, 2x 500w capacitors are easier to manage physically, so multi rail makes more sense. But all that's internal, doesn't really affect what's on the outside.

Modular means every cable is connectable, has a plug. An advantage is you only plug in what you need, the wiring can all be done during the build, the psu swapped easily, at any time. But, every connection is a potential weak point with electricity and modular plugs often get stressed with corners or shoving wires under shrouds etc.

Semi-modular has certain wires permanently attached. 20+4 mains and eps. The pcie and Sata are modular. You need eps and mains always, so eliminates potential weak point there, but is still modular in the accessory wiring. But, there's no easy wiring like with full modular, the psu must be built into the pc as the EPS and mains are permanently attached.

Non-modular has all wires permanently attached. No potential weak points from a connector. But, you have to deal with all that wiring whether used or not. Has none of modular or semi advantages.

Modular, semi, non modular only affects how you build the pc or tear it down and related aspects, doesn't affect how it operates. Rails affect the psu operation, not how you build the pc.
Thank you for the concise and informative explanation! I already understood most of that, but it's still great for reference.

What I infer from your answer is that if my case space is quite small, a fully/semi- modular PSU would be advantageous because I would only install the cables I need. Aside from that, it doesn't much matter. Is that correct?

I think you've probably over-simplified rails, but it's been a VERY long time since I studied circuitry. What I'm thinking of here is partially redundancy and also that splitting the load across multiple rails would eliminate certain problems other than physical size.

So...The convenience of FM/SM isn't necessary and can introduce hotspots and, if I'm not going to frequently be adding and removing things, doesn't serve much use other than to make it easy to replace damaged cables. I'd imagine that NM is cheaper, but did the tier list makers use FM/SM/NM as a factor that decided tier/color? They colored NM red in the sheet.
 
Last edited:

DSzymborski

Titan
Moderator
They didn't include modularity as part of the tier list; the goal was to functionality of the PSU itself, something which is hard for the average person to evaluate. Whether they need or don't need modular cables is something a person usually can evaluate on their own, so a tier list is less helpful.

However, you'll generally find a loose correlation between modularity and PSU quality (note that I say a loose correlation). The cheapest PSUs squeeze every penny until it screams, so it's not something they prioritize, and in the ultra-cheap market, the price battle is fierce. Plus, modular vs. non-modular costs don't really scale up with price, so it's a smaller % of the total cost to the company in the high-end stuff.
 

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
They didn't include modularity as part of the tier list; the goal was to functionality of the PSU itself, something which is hard for the average person to evaluate. Whether they need or don't need modular cables is something a person usually can evaluate on their own, so a tier list is less helpful.

However, you'll generally find a loose correlation between modularity and PSU quality (note that I say a loose correlation). The cheapest PSUs squeeze every penny until it screams, so it's not something they prioritize, and in the ultra-cheap market, the price battle is fierce. Plus, modular vs. non-modular costs don't really scale up with price, so it's a smaller % of the total cost to the company in the high-end stuff.
Which returns me to the question of 1,000W or 1,200W, multi or single rail?

If there's only a loose correlation between price and modularity, then it sounds like I don't need to worry about it, especially since I have a large case - I'd need to be more concerned with making sure it'll have enough of the connectors I'll need. Is that correct?

How BIG is the step down from gold to blue, in real terms? I definitely want high quality.

Since the spreadsheet itself doesn't show tiers and colors, I have to go by the list...If you can answer the questions in post #34, that would be super.
 
Last edited:

WrongRookie

Commendable
Oct 23, 2020
485
25
1,740
5
Seems to me that an 850w is ideal for your build although you could go for more if youncan afford it.

Personally i have Seasonic Focus+ PX Platinum 850w. Pretty good PSU compared to the Corsair VS650 i had earlier which was faulty.
 
Last edited:
Reactions: ReveurGAM

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
Seems to me that an 850w is ideal for your build although you could go for more if youncan afford it.

Personally i have Seasonic Focus+ Platinum 850w. Pretty good PSU compared to the Corsair VS650 i had earlier which was faulty.
Why do you say 850? That's what's recommended for just the GC.
 

zszabo

Distinguished
Dec 18, 2008
6
2
18,515
0
RTX 4090 AIBs recommend 1000W PSUs for anyone buying an RTX 4090. Your GPU has half the maximum power requirement of the 4090, which itself is an outlier of a card, in terms of its absolutely brutal power demands. With electricity prices rising precipitously in Europe (and likely the U.S. as well, in the future; odd how I just phrased that, considering I live in the U.S.), any further increases in power consumption will only be increasingly disincentivized, going forward. CPUs has been following this trend for some time, where power requirements went up and up and then eventually plateaued.

In reality, 1000W and above PSUs came into existence mainly for situations like multi-GPU configurations. This mainly only changed because of the recent surge in power requirements for GPUs. For example, I currently have a seven year old 980 ti in my system, which is similar in TDP (if not performance) to your GPU, at 250W vs. 300W. Nvidia recommends a 600W PSU for it, meaning an 850W is likely to provide all the headroom you will ever need, if you plan on not running 2+ GPUs. That's the only future proofing you need to worry about.

As for me, I'll be needing to upgrade the 980 ti soon—and it isn't due to insufficient performance, as much as it is about inadequate memory. And that is primarily because my preferences changed in the seven years since I bought that card, to include 3D-related work, which generally requires vastly more VRAM than gaming (with texture sizes typically at 4096x4096 pixels, 24 bits (or 30, or even 36 bits, for HDR images used in image based lighting (IBL)), and multiple such textures used per shader (e.g. a diffuse texture, a normal texture, bump map, transparency map, roughness map, and up to and including several more such maps for specular, metallicity, anisotropy, effects, which is not counting the sub-surface scatter maps to simulate realistic skin). Anyway, just something to think about when it comes to "future proofing." In my case, the 6Gb VRAM was probably above average at the time (I'm guessing 3-4Gb was closer to average, although the Titan X, which I in hindsight should have bought, again not for the marginal extra performance, but rather the 12Gb VRAM it came with, which if I had today I wouldn't even be participating in this thread today, no offense, and yes, biggest run-on sentence ever, probably, or maybe def.).
 
Reactions: ReveurGAM

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
RTX 4090 AIBs recommend 1000W PSUs for anyone buying an RTX 4090. Your GPU has half the maximum power requirement of the 4090, which itself is an outlier of a card, in terms of its absolutely brutal power demands. With electricity prices rising precipitously in Europe (and likely the U.S. as well, in the future; odd how I just phrased that, considering I live in the U.S.), any further increases in power consumption will only be increasingly disincentivized, going forward. CPUs has been following this trend for some time, where power requirements went up and up and then eventually plateaued.

In reality, 1000W and above PSUs came into existence mainly for situations like multi-GPU configurations. This mainly only changed because of the recent surge in power requirements for GPUs. For example, I currently have a seven year old 980 ti in my system, which is similar in TDP (if not performance) to your GPU, at 250W vs. 300W. Nvidia recommends a 600W PSU for it, meaning an 850W is likely to provide all the headroom you will ever need, if you plan on not running 2+ GPUs. That's the only future proofing you need to worry about.

As for me, I'll be needing to upgrade the 980 ti soon—and it isn't due to insufficient performance, as much as it is about inadequate memory. And that is primarily because my preferences changed in the seven years since I bought that card, to include 3D-related work, which generally requires vastly more VRAM than gaming (with texture sizes typically at 4096x4096 pixels, 24 bits (or 30, or even 36 bits, for HDR images used in image based lighting (IBL)), and multiple such textures used per shader (e.g. a diffuse texture, a normal texture, bump map, transparency map, roughness map, and up to and including several more such maps for specular, metallicity, anisotropy, effects, which is not counting the sub-surface scatter maps to simulate realistic skin). Anyway, just something to think about when it comes to "future proofing." In my case, the 6Gb VRAM was probably above average at the time (I'm guessing 3-4Gb was closer to average, although the Titan X, which I in hindsight should have bought, again not for the marginal extra performance, but rather the 12Gb VRAM it came with, which if I had today I wouldn't even be participating in this thread today, no offense, and yes, biggest run-on sentence ever, probably, or maybe def.).
Thank you for the very informative response, even if the sentences were...ummm...lengthy. This topic has had me rather confused, so I'm hoping you (or anyone) can clarify it for me. What I am very confused about is why, if the GPU needs 500W (something I just noticed said 255W MAX) then why does my card's documentation say that XFX recommends 850W and the bare minimum is 750W? If the GC actually uses 850W at times (assuming 90% efficiency and factoring in the other components of my system), how would it be possible to run my system on 850W without causing brown-out/throttling conditions? Following this chain of thought, I find myself unable to figure out why I could run my system with only that much. It has been suggested to me that certifications don't really matter, but they give us the efficiency of the unit - although I've read that 80+ certs are on the lax side. If a unit has 90% efficiency at 1000w, that means it actually produces 900W, , 94% begets 950W, etc., right? Thus, an 850W PSU at 90% only produces 765W, and I would have to look for at least a Platinum cert. from 80+ or Cybenetics Labs to make sure I'd get that much. Since the tiers list doesn't include CL, I'd have to compare my short list to CL's list.

I would really appreciate it if you could explain this to me.
 

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
Previously I was very interested in the HX1000i, but that was because a previous Corsair rep had told me that their iCUE software was the only way to monitor and control their products. Now that I know that there are a lot of programs out there that can do that, I don't need the HX1000i. Thus, the RM1000x (or the be quiet! Pro 12, or a high-quality single-rail PSU) seems like a really good option, especially since a few people here have already suggested it.

Getting back to the rails question...The tiers list suggests that >=1000W should use a multi-rail configuration. As I mentioned, I studied circuitry ages ago, so I'm pretty vague on the reasoning. I learned from a Corsair support person that the load will be balanced across all the rails, but he didn't feel that multi-rail is necessary (odd, coming from their agent).

zszabo has suggested that I only need 850W, but I don't understand why. Let's assume that, unless someone can help me to understand why 850W is enough, I'll get a 1,000W to hedge my bets and prepare for the future. The question is: Should I get a single-rail, multi-rail or multi-+switchable PSU?
 

Phaaze88

Titan
Ambassador
According to this news article, AIBs are suggesting up to 1200w, not 1000w.
https://www.tomshardware.com/news/1200w-power-requirement-rtx-4090

Now, you have to know that they exaggerate to cover their own butts and because the power use from someone's PC can't be predicted.
Nvidia's own recommendation is an exaggeration based on a system with a 5900X in it - quite an efficient cpu in its own right.
If the AIBs are recommending up to 1200w on some of their 4090s, power management involving manual overclocks+higher power limits must be more ridiculous than Ampere was.
You will have to wait for reviews to see what it's like. Gamers' Nexus will definitely cover it, and TechPowerUp likely will as well.

The Founder's Editions stand to offer the best performance per currency, or cost per frame - yes, I know, 'value and 4090, PFFT' - for each gpu tier by far(that they exist in), and that value even includes one's psu options.
[2 generations ago, ending with Turing, that was not the case; Founder's were about 100 quid over MSRP.]
So zszabo may be right about being able to work with 850w units - at least with Founder's 4090, but if you're gunning for 4090 DarkOmegaEdgelordAbsoluteChaosPower model... well, you may have to step it up a notch.

The boost algorithm already does its own overclocking, and look at those psu recommendations compared to Founder's... up to a 350w gap?! These cards become woefully inefficient from just 'a little push'.
Really though, I feel it's gotten to the point that manual overclocks and higher board power limits than Founder's needs to come to a close, or become exclusive to an AIB's highest product tier; the folks getting those should already know what they're getting into.
Though that does mean AIBs would mostly be selling on looks, cooling, and sound profile. Continuing to sell on performance comes with significant costs.
 
Reactions: ReveurGAM

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
According to this news article, AIBs are suggesting up to 1200w, not 1000w.
https://www.tomshardware.com/news/1200w-power-requirement-rtx-4090

Now, you have to know that they exaggerate to cover their own butts and because the power use from someone's PC can't be predicted.
Nvidia's own recommendation is an exaggeration based on a system with a 5900X in it - quite an efficient cpu in its own right.
If the AIBs are recommending up to 1200w on some of their 4090s, power management involving manual overclocks+higher power limits must be more ridiculous than Ampere was.
You will have to wait for reviews to see what it's like. Gamers' Nexus will definitely cover it, and TechPowerUp likely will as well.

The Founder's Editions stand to offer the best performance per currency, or cost per frame - yes, I know, 'value and 4090, PFFT' - for each gpu tier by far(that they exist in), and that value even includes one's psu options.
[2 generations ago, ending with Turing, that was not the case; Founder's were about 100 quid over MSRP.]
So zszabo may be right about being able to work with 850w units - at least with Founder's 4090, but if you're gunning for 4090 DarkOmegaEdgelordAbsoluteChaosPower model... well, you may have to step it up a notch.

The boost algorithm already does its own overclocking, and look at those psu recommendations compared to Founder's... up to a 350w gap?! These cards become woefully inefficient from just 'a little push'.
Really though, I feel it's gotten to the point that manual overclocks and higher board power limits than Founder's needs to come to a close, or become exclusive to an AIB's highest product tier; the folks getting those should already know what they're getting into.
Though that does mean AIBs would mostly be selling on looks, cooling, and sound profile. Continuing to sell on performance comes with significant costs.
Amidst all of that, I didn't actually glean whether or not I should get 1000w to hedge my bets, or take zszabo's advice, nor did I see any preference for rail(s). .....?
 

Phaaze88

Titan
Ambassador
Amidst all of that, I didn't actually glean whether or not I should get 1000w to hedge my bets, or take zszabo's advice, nor did I see any preference for rail(s). .....?
KIS-M: Keep It Simple Mate/Mister/Madam.
You either:
A)Wait for reviews. By then, we'll all have the knowledge necessary to make an informed decision. All this speculation(that can stretch to multiple pages) is getting nowhere - happens every blooming time before a new product launch; folks are asking for info the general public doesn't know/have access to yet. You are one of many to do this.

B)Go and get/order an 850/1000/1200w psu right now depending on the gpu model you're targeting, and take a chance on whether or not you'll get bitten for not waiting.


IF I were in the market for a 4090/Ti, I'd be doing A. I'm using an AX850 since Jan of 21 - the transient power issues concerning RTX 30 hadn't come up yet.
In the Cultists' psu tier list, this unit is based on the older Seasonic Prime platform and may be susceptible to the spikes. The AX850 probably won't cut it with one of those RTX 40s either - well, maybe the Founder's is ok, but that's just more speculation.
 

ReveurGAM

Great
Sep 28, 2022
160
6
85
0
@Phaaze88 I'm really not sure why you keep talking about a card that I don't have (4090, 5900). I thought you were using it as an example about wattage, but now I think you might've gotten my setup confused with someone else's. I have an XFX Speedster Swift 319 Radeon RX 6080 XT 16 GB GC. In fact, I am unaware of anything in my setup that is so new that I qualify even vaguely for the statement that I'm asking for info that's not available. Am I correct that you mixed me up with someone else, or is my GC truly that new? If it's the latter - I truly had no idea that I had a card with an "older" GPU on it that is new.
 

ASK THE COMMUNITY

TRENDING THREADS