[EOL] PSU Tier List rev. 14.8

[Nirurin]

1 minute ago, Juular said:

There was a brief YT review by Aris but no article yet, although from the video it sounded like Aris mostly liked the unit. It has problems with OTP (it's either set higher than 200°C making it essentially useless or doesn't work at all), and OPP (which is set way too high, especially for SFX PSU), but it's probably not a bomb, assuming you really need a SFX(-L) PSU with more than 750W then it should be fine.

I edited my original post with this - https://www.thefpsreview.com/2021/06/17/silverstone-sx1000-1000w-sfx-l-power-supply-review/ 

Seems it got released today (wasn't sure if you read my post before or after I edited it)

[MultiGamerClub]

45 minutes ago, Juular said:

RM-x 850W is cheaper than ASUS version of Seasonic Focus Gold 750W ? And you're asking what's better ? RM-x obviously, even if they were the same price and wattage really.

Browsing through prebuilts w psu's i should go for. Currently a white Corsair RM850x or a ASUS ROG Strix 650 at the moment.

Edited June 17, 2021 by MultiGamerClub

[Guest]

Do we have a review or anything with internal shots of an EVGA GA Series?

 

I have a couple here and they say "made in Taiwan" and AFAIK, Andyson doesn't have a Taiwan facility.  So I want some pictures to make a comparison to see if there's bait and switch going on.

 

EDIT:  Never mind.  Found pictures and opened mine up.  It's definitely an Andyson PAD01.

 

[LukeSavenije]

10 hours ago, jonnyGURU said:

So I want some pictures to make a comparison to see if there's bait and switch going on.

wouldn't be the first time... cough B3 and GD

[Guest]

2 minutes ago, LukeSavenije said:

wouldn't be the first time... cough B3 and GD

Dude.. I see it so often I see the insides switched out, it actually surprised me that it WAS an Andyson when I opened it up.

 

[LukeSavenije]

1 minute ago, jonnyGURU said:

Dude.. I see it so often I see the insides switched out, it actually surprised me that it WAS an Andyson when I opened it up.

 

boy that's a concern on it's own... but EVGA is a mess with over 40 models registered at this point

[IIIIIIIIII]

Silverstone SX1000 review (Tom's)

 

I wonder if this PSU will be able to fully support the RTX 3090

 

[Lecom1020]

im getting a new psu have to many problems with my dark power pro 11 1200w   fan is done and it has some failures sometimes what would you recommend 

currently using a Ryzen 9 5950x with an X570-F Gaming mb and Asus Strix RTX 3090 OC no hdd only m2 nvme  

i was thinking about getting another be quiet unit but couple of my frieds said they have issues with blinking red lights on their gpu and after getting a seasonic or corsair its gone 

 

single rails  multi rail?

 

should have headroom for future upgrades and should last a while

[Juular]

4 minutes ago, Lecom1020 said:

im getting a new psu have to many problems with my dark power pro 11 1200w   fan is done and it has some failures sometimes what would you recommend 

currently using a Ryzen 9 5950x with an X570-F Gaming mb and Asus Strix RTX 3090 OC no hdd only m2 nvme  

i was thinking about getting another be quiet unit but couple of my frieds said they have issues with blinking red lights on their gpu and after getting a seasonic or corsair its gone 

 

single rails  multi rail?

 

should have headroom for future upgrades and should last a while

[zoidzoid]

This chart is bogus.
Be quiet E9 grade D?

Its a gold rated fsp psu with nothing but glowing reviews. It was a very popular psu with 4.9stars and 137 reviews.
Mine is running for 6 years powering decent gpu, the 400w version.

I think I sense a bias in this chart.
Seasonic = good.
FSP = junk

Without any legit reasoning, the psu are only suitable for low end/igp system??? Even though the 400w model comes with 2xpci-e connectors. You put well capable psus there just because its FSP. Here in translation of a very details review of the E9 400w. Does this sound like a grade d psu only suitable for IGP systems?

"Plus:

• Very good workmanship of the outer skin, valuable feel

• robust high quality paintwork

• very high efficiency in almost all load ranges • low voltage tolerance values

• very good tension stability and very good support times

• Very good ripple and noise values

• quite high power reserves (up to 546 watts)

• Correctly responding extensive protective circuits

• Satisfactory to good active PFC values

• No interfering noises from the power supply electronics

• effective cable shielding and insulation

• sufficiently long cable strands (partially)

• Satisfactory component quality, high-quality multifunction ICs

• Very quiet, excellent fan • good to very good soldering quality

• satisfactory accessory package

• good price-performance ratio (approx. 63 €)

• Long warranty period: 5 years (on-site exchange service within the first year)

 

Minus:

• Elko assembly can be optimized

• Wrong warranty information on the packaging (instead of 5 years only 3 years printed)

• Cable harnesses could be longer

• No shrink tubing insulation on the cables to be soldered"

[Juular]

12 minutes ago, zoidzoid said:

I think I sense a bias in this chart.

You do. Against group regulated PSUs. Which bqt SP E9 is, as most of the rest of PSUs in tier D. Which aren't suitable for a modern high power draw PC because they'll struggle with keeping voltages in specifications (5%), not to mention in tighter 1-2% which any modern DC-DC PSU will be able to. And you would've knew that if you've read the methodology first. Also, this PSU is 10 years old so it wouldn't be recommended for purchase anyway. And we have a bunch of PSUs from Seasonic there in tier D, group regulated and not. As we do a have a bunch of FSP-branded/made PSUs in tier A among others.

[zoidzoid]

25 minutes ago, Juular said:

You do. Against group regulated PSUs. Which bqt SP E9 is, as most of the rest of PSUs in tier D. Which aren't suitable for a modern high power draw PC because they'll struggle with keeping voltages in specifications (5%), not to mention in tighter 1-2% which any modern DC-DC PSU will be able to. And you would've knew that if you've read the methodology first. Also, this PSU is 10 years old so it wouldn't be recommended for purchase anyway. And we have a bunch of PSUs from Seasonic there in tier D, group regulated and not. As we do a have a bunch of FSP-branded/made PSUs in tier A among others.

I found a Silverstone ST60F-ESG on Amazon for £36 new recently. Wasn't going to buy it as knew it was older model, but for 600w 80+ gold, I couldn't turn it down. The reviews would seem to imply it is based on a slightly older/worse FSP platform than the E9 yet in this chart the Silverstone essential gold is grade C, and Be Quiet E9 is grade D. Can you explain why the Silverstone is superior?

Edit nvm, I see the Silverstone is in grade D as well and its an sfx version in grade C.
Really, even though the psu has 4 x 8pin pci-e, I should only use it on an a very cheap IGP system? I think this is a bit over exaggerated. I would like to see real world examples of group regulated psus being unsuitable for use in a modern system. I mean, when these were released, there were still very power hungry gpus like gtx 580 / 680 and I don't recall any reports of mass failures. It seems to be something discovered only in the past few years as newer models came out and tests became more advanced? But in the real world does it actually matter?

[zoidzoid]

3 hours ago, Juular said:

You do. Against group regulated PSUs. Which bqt SP E9 is, as most of the rest of PSUs in tier D. Which aren't suitable for a modern high power draw PC because they'll struggle with keeping voltages in specifications (5%), not to mention in tighter 1-2% which any modern DC-DC PSU will be able to. And you would've knew that if you've read the methodology first. Also, this PSU is 10 years old so it wouldn't be recommended for purchase anyway. And we have a bunch of PSUs from Seasonic there in tier D, group regulated and not. As we do a have a bunch of FSP-branded/made PSUs in tier A among others.

Is there really any reason to believe a half decent group regulated psu would be 'unsafe' outside of lab tests? It seems such psus would only go out of spec if under really unusual loads or pushing the 12v to its absolute limit. I think practically all consumers would be running their psu at normal load ranges. Isn't it still recommended to choose a psu which will be at approx 50% load when gpu+cpu is stressed, in that case voltage will stay well within spec.
For the Silverstone ST60F-ESG at least, they do specify haswell is supported and gpu support up to rtx 3070 for it on their 2021 gpu support list. Their gpu recommendations would point towards them allowing higher gpu power on newer non group regulated psu of the same power limit, but I dont see no igp only being mentioned. Why would they say this if its dangerous and really should only be used on a cheap igp system?
So any real world evidence for why such psus cant/shouldn't be used? Would appreciate it so dont end up damaging something with this outdated pos

[Juular]

8 hours ago, zoidzoid said:

It seems such psus would only go out of spec if under really unusual loads or pushing the 12v to its absolute limit.

12V-heavy crossloads aren't unusual at all in a modern system. In a PC without RGB and HDDs the power draw on 3.3V and 5V rails would be minuscule, while the rest of power draw would be on 12V rail.

8 hours ago, zoidzoid said:

Isn't it still recommended to choose a psu which will be at approx 50% load when gpu+cpu is stressed,

It isn't lol, it was a standard recommendation when group regulated PSUs were prevalent, they aren't today. Or rather, even in budget range (50-60$), DC-DC units are very affordable, so no reason to buy a group regulated PSU unless you want an absolute cheapest stuff and the rest of your PC is exactly as cheap.

8 hours ago, zoidzoid said:

ISo any real world evidence for why such psus cant/shouldn't be used?

It's not exactly about damage from out-of-spec voltage regulation, it's about stability and damage from lacking protections or high ripple because most group regs are either very cheap (read noname) or very old designs or very old stock (as say, bqt E9 will be if you'll find it anywhere). If motherboard / GPU VRM is even half-decent they should be able to keep up even with out of spec voltages, it's just, as i said, when the market is flooded with DC-DC units, going for a group-reg doesn't make sense unless you just want a cheapest PSU possible.

Then there's transient response, i don't have exact numbers on most units (barely anyone but Aris tests transient response and he doesn't test a lot of group regs), but i imagine in a group reg PSU on an old platform without LLC resonant primary, designed when high-transient GPUs/CPUs weren't really a thing, the transient response wouldn't be good, that combined with already bad voltage regulation in crossloads leads to voltages going way out of spec momentarily, which would if not damage the hardware, but at the very least affect stability. Then, the part of transient response is also the amount of secondary side filtering capacitance, there would be less filtering in budget units, group reg or not, and thus less capacitance.

And reports of people using their high-end GPUs with group reg PSUs, even relatively okay ones such as Corsair CV and then getting stability issues, aren't something unheard of.

12 hours ago, zoidzoid said:

I found a Silverstone ST60F-ESG on Amazon for £36 new recently.

Are you sure about that ? What's manufacturing date ?

[zoidzoid]

 

12 hours ago, Juular said:

You do. Against group regulated PSUs. Which bqt SP E9 is, as most of the rest of PSUs in tier D are. Which aren't suitable for a modern high power draw PC because they'll struggle with keeping voltages in specifications (5%), not to mention in tighter 1-2% which any modern DC-DC PSU will be able to. And you would've knew that if you've read the methodology first. Also, this PSU is 10 years old so it wouldn't be recommended for purchase anyway. And we have a bunch of PSUs from Seasonic there in tier D, group regulated and not.

 

Yes I already received it from Amazon, the Silverstone. It looks and smells brand new. I cant check the date now, obviously I don't expect it to be newly made probably just new old stock from 2013-2014. I already own the E9 400w since 2012. I also have an FSP 400w which is the exact same PSU. I guess my psu choice was not good haha.

I understand a group regulated would not be the best choice nowadays, but is it really required to class them all as only suitable for  'cheap igp system only'. If its that bad, surely someone has recreated a situation like this and proved the problem? Like got a group regulated psu powering modern systems and run it under realistic but unfavourable load for prolonged time and found higher hardware failure rate than with a non group regulated. Its nice to have perfect voltage but I find it hard to believe this is something required to not damage parts, specifications must exist for a reason. Its not like everything powered by a non group regulated will not fail ever, there are many reports on the internet of people claiming their psu failed, blew up, took parts with it and they were using top of the line corsair/seasonics. I'd like to just see a few reports of hardware failing and owner was using a group regulated psu.

 

Without any proof I would assume the main reason most psu moved to dc-dc was to avoid their product getting bad reviews. Like alot of things, RGB, psu shrouds/tempered glass cases, things become a trend and its a business decision to include in their product. If this isn't the case please just give me even a bit of anecdotal evidence that shows it might be dangerous for me to use the psu with a modern pc under normal situations.

 

 

 

[Juular]

1 hour ago, zoidzoid said:

Like got a group regulated psu powering modern systems and run it under realistic but unfavourable load for prolonged time and found higher hardware failure rate than with a non group regulated.

Again, it's not strictly about damage. Even though most group regs wouldn't be able to keep voltages in specs under crossloads (and even more so under transient load), that's not what damages hardware unless the VRM is really, very cheap.

It's about stability, there are a lot of reports of people using popular 'branded' group reg PSUs (Corsair VS/CV, EVGA W1, Seasonic S12II, whatever 50$ Gamemax, Raidmax, Aresgame, Apexgaming, Gamdias etc.) with high-end hardware and wondering why it's not stable despite wattage is enough per se. And then, damage may still occur if the PSU is bad quality to start with and has sky-high ripple, 99% of such PSUs are group reg because designs are usually decades old, DC-DC wasn't a thing back then and it's more expensive anyway.

There are of course bad quality DC-DC units, and they become more frequent with each year, so DC-DC doesn't mean that the PSU is necessarily good quality either. But overall, just don't expect any group regulated PSU you can buy today to be better or at least match a decent DC-DC unit in quality and performance. The latter would also be significantly more expensive but you gotta pay for quality. But them being DC-DC is the consequence of increased quality and performance, not the (only) cause.

[zoidzoid]

17 minutes ago, Juular said:

Again, it's not strictly about damage. Even though most group regs wouldn't be able to keep voltages in specs under crossloads (and even more so under transient load), that's not what damages hardware unless the VRM is really, very cheap.

It's about stability, there are a lot of reports of people using popular 'branded' group reg PSUs (Corsair VS/CV, EVGA W1, Seasonic S12II, whatever 50$ Gamemax, Raidmax, Aresgame, Apexgaming, Gamdias etc.) with high-end hardware and wondering why it's not stable despite wattage is enough per se. And then, damage may still occur if the PSU is bad quality to start with and has sky-high ripple, 99% of such PSUs are group reg because designs are usually decades old, DC-DC wasn't a thing back then and it's more expensive anyway.

Overall, just don't expect any group regulated PSU you can buy today to be better or at least match a decent DC-DC unit in quality and performance. The latter would also be significantly more expensive but you gotta pay for quality. But them being DC-DC is the consequence of increased quality and performance, not the (only) cause.

I understand this, but from a review of this Silverstone psu I could find from 2015, from someone very harsh on group regulated psu, it would still seem to be one of the better ones and they gave it a Pass rating. If using it probably wont result in damage but stability issues, why brand them only suitable for use in cheap igp system? I think it would make more sense to not automatically put group regulated in this category but categorize based on other qualities also with an asterisk of possible stability issues (even though claims to support haswell). Especially since most are unable to buy nowadays, its not fair to just brand them as trash when they are fully capable of powering a decent system within spec.

 

Because dc-dc appears to be so prelevant nowadays I would be confident in saying there is worse quality dc-dc power supplies that have higher chance of harm someones pc,from Grade C here, than the Grade D rated Silverstone Strider Gold Essential. Eg Evga BQ. In fact I was choosing between an amazon warehouse Evga BQ 600w for £38 or the new Silverstone for £36. Silverstone is gold rated with high quality capacitors, 4 pci-e. The BQ is bronze, dc-dc, crappy caps, 2 pci-e. I still feel I made the better choice even knowing about the group regulation issue.

[Juular]

3 hours ago, zoidzoid said:

from a review of this Silverstone psu I could find from 2015, from someone very harsh on group regulated psu, it would still seem to be one of the better ones and they gave it a Pass rating.

What review are you talking about exactly ? I don't recall any review on that platform with crossload and transient response tests.

3 hours ago, zoidzoid said:

If using it probably wont result in damage but stability issues, why brand them only suitable for use in cheap igp system?

Because there would be stability issues if you use them with modern high-wattage PC ? This guide is supposed to be fool proof, so by putting all group regs in a lowest tier and telling people that these units are only suitable for iGPU systems we try to minimize the chance of these people having issues, even if some of such group regs are fine for lower-end hardware, like say, Corsair CV550 with GTX1660 Ti if you're really limited in your PSU choice or budget.

3 hours ago, zoidzoid said:

its not fair to just brand them as trash when they are fully capable of powering a decent system within spec.

Again, i repeat, they can't. Most of them at least. And there are too few detailed reviews on the rest to say otherwise. We're highlighting the units there are good reviews on and which show relatively good regulation but we assume the worst on the rest because that's how you should treat PSUs. If i were to find a review of FSP Aurum based PSU with crossloads and transient respoтse tests and in which it performs relatively favorably (i'd say 12V is still in specifications under 12V heavy crossload + transient response deviation) then i'd highlight such PSUs too.

3 hours ago, zoidzoid said:

Because dc-dc appears to be so prelevant nowadays I would be confident in saying there is worse quality dc-dc power supplies that have higher chance of harm someones pc,from Grade C here, than the Grade D rated Silverstone Strider Gold Essential. Eg Evga BQ.

Yes, there definitely are, and we try to move such units to low priority subtiers to highlight that we don't trust them, or move them to tiers D and E if there were some confirmed issues (Gigabyte P-GM among others from recent times).

[zoidzoid]

16 minutes ago, Juular said:

What review are you talking about exactly ? I don't recall any review on that platform with crossload and transient response tests.

Because there would be stability issues if you use them with modern high-wattage PC ? This guide is supposed to be fool proof, so by putting all group regs in a lowest tier and telling people that these units are only suitable for iGPU systems we try to minimize the chance of these people having issues, even if some of such group regs are fine for lower-end hardware, like say, Corsair CV550 with GTX1660 Ti if you're really limited in your PSU choice or budget.

Again, i repeat, they can't. Most of them at least. And there are too few вуефшдув reviews on the rest to say otherwise. We're highlighting the units there are good reviews on and which show relatively good regulation but we assume the worst on the rest because that's how you should treat PSUs. If i were to find a review of FSP Aurum based PSU with crossloads and transient respoтse tests and in which it performs relatively favorably (i'd say 12V is still in specifications under 12V heavy crossload + transient response deviation) then i'd highlight such PSUs too.

Yes, there definitely are, and we try to move such units to low priority subtiers to highlight that we don't trust them, or move them to tiers D and E if there were some confirmed issues (Gigabyte P-GM among others from recent times).

https://www.hardwareinsights.com/silverstone-strider-essential-gold-600-w-bringing-essential-to-next-level/#Introduction

They give this PSU a pass, but they fail other group regulated like Be Quiet Pure Power 9 + Silverstone Bronze Essential.

I still can't foresee any issues using this psu in a normal system. Lets say mid range cpu + mid range gpu, under normal load its only going to be max 250-300w load under gaming.  Yes there will be some spikes, lets say another 50-100w. Still under 400w. Worse case scenario in this psu review, 550w on the 12v with practically no load on 3v + 5v rail = 12v @ 11.66v. Bad but not out of atx spec, but seeing as a mid range gaming pc will be 300-450w absolute maximum it will never get this bad. Something like a 1660 + appropriate cpu will never exceed 300w no matter what and have no issue at all, this is still a capable gaming pc so to brand it as only suitable for IGP system... I think its a bit much.

 

[jeelemone]

Is there any difference between the RMx 2018 white and black versions? The white version is listed as being in tier A on the list but is there any real difference other than colour between the two? Asking as I need to buy a new PSU but for me the white and black versions are the same price so price isn't really a factor

[GoldenLag]

5 minutes ago, jeelemone said:

Is there any difference between the RMx 2018 white and black versions?

They are the same. If you read the legend you will see why its marked with "white". 

[jeelemone]

18 minutes ago, GoldenLag said:

They are the same. If you read the legend you will see why its marked with "white". 

Oh, you're right I completely missed that, apologies! Thank you!

[Juular]

2 hours ago, zoidzoid said:

They give this PSU a pass

Yeah ... while crossloads for 100% rated current are fine, 10.4% deviation on high combined load is not a pass for me.

2 hours ago, zoidzoid said:

I still can't foresee any issues using this psu in a normal system.

While high 5V load is not as frequent today as before, someone would stuff their PC with HDDs or RGB fans and it will reach 8A power draw on 5V easily, combined with a power draw in something like Blender (not everyone just games on their PCs) of 550W (or maybe even less than that seeing that it's 10%+ at 550W) from something like i9 11900K + RTX3080 and hello double-out-of-spec voltage regulation. People are cheap, they will and they do put group regulated PSUs in such high-end builds, we ain't gonna normalize that. Also, buying a PSU for best-case scenario (300-400W load in gaming from a 500-600W hardware) is not how you choose a PSU. If your PC would reach maximum 600W sustained power draw, even if that's only in stress tests, then you buy a 600W PSU, not 400W. With modern high transient-power draw hardware it's even more than that, RTX3080 is ~350W GPU if we just look at average sustained power draw, but it has almost 500W transient peaks, the difference is so high that it actually kind of justifies to overprovision your PSU wattage, but not because of reasons most people cite (efficiency golden spot, endurance increase etc.), just because your PSU would probably just trip it's UVP because the GPU will momentarily drain all it's secondary side capacitors so voltage plummets. Jon Gerow says that PCIe 5.0 specifications would also define such spikes in paper, effectively normalizing them, so we'll probably see more of such behavior from more GPUs to come.

Another thing, there are no protection tests on this or most of other group reg PSUs. But the fact that it goes double out of spec under 100% combined load tells us that OPP or OCP is not configured properly, if we would see such behavior from any PSU in higher tiers it would've been detiered to tier D, this unit is already there because really, we don't expect group regulated PSUs to have a good voltage regulation in the first place.

[zoidzoid]

1 hour ago, Juular said:

Yeah ... while crossloads for 100% rated current are fine, 10.4% deviation on high combined load is not a pass for me.

While high 5V load is not as frequent today as before, someone would stuff their PC with HDDs or RGB fans and it will reach 8A power draw on 5V easily, combined with a power draw in something like Blender (not everyone just games on their PCs) of 550W (or maybe even less than that seeing that it's 10%+ at 550W) from something like i9 11900K + RTX3080 and hello double-out-of-spec voltage regulation. People are cheap, they will and they do put group regulated PSUs in such high-end builds, we ain't gonna normalize that. Also, buying a PSU for best-case scenario (300-400W load in gaming from a 500-600W hardware) is not how you choose a PSU. If your PC would reach maximum 600W sustained power draw, even if that's only in stress tests, then you buy a 600W PSU, not 400W. With modern high transient-power draw hardware it's even more than that, RTX3080 is ~350W GPU if we just look at average sustained power draw, but it has almost 500W transient peaks, the difference is so high that it actually kind of justifies to overprovision your PSU wattage, but not because of reasons most people cite (efficiency golden spot, endurance increase etc.), just because your PSU would probably just trip it's UVP because the GPU will momentarily drain all it's secondary side capacitors so voltage plummets. Jon Gerow says that PCIe 5.0 specifications would also define such spikes in paper, effectively normalizing them, so we'll probably see more of such behavior from more GPUs to come.

Another thing, there are no protection tests on this or most of other group reg PSUs. But the fact that it goes double out of spec under 100% combined load tells us that OPP or OCP is not configured properly, if we would see such behavior from any PSU in higher tiers it would've been detiered to tier D, this unit is already there because really, we don't expect group regulated PSUs to have a good voltage regulation in the first place.

I don't think even today that most would recommended 600w psu if usage will get near to that. I don't frequent forums as much as I used to but I have seen a few threads here and there on reddit etc and this isn't something I see. Isn't it good to have some headroom and also to achieve max efficiency? I would never consider putting a 3080 on this psu but I would be comfortable with anything lower as long as cpu is nothing too power hungry.

 

Btw that silverstone psu review in 2015 says v1.0 , my one is v1.1. Maybe its now dc-dc

For the price I got it, I will just keep it as a backup. I'm  not stuck for power supplies as I have a few others. Just bit shocked to see some of my psu which have been running rock solid with decent hardware are recommended for igp only!

[Juular]

23 minutes ago, zoidzoid said:

Btw that silverstone psu review in 2015 says v1.0 , my one is v1.1. Maybe its now dc-dc

Unlikely, while Silverstone does sometimes switch platforms completely, they change the major number if they do so. You can see for yourself, if the 12V wattage on the label is the same as on v1.0 then the platform is likely the same, or at the very least it's still group regulated.

23 minutes ago, zoidzoid said:

I don't think even today that most would recommended 600w psu if usage will get near to that. I don't frequent forums as much as I used to but I have seen a few threads here and there on reddit etc and this isn't something I see.

I don't get what you're trying to say here. I and other people do recommend 550W, 650W, 750W, 850W and 1kW PSUs alike. Headroom is good but not for efficiency reasons but to accommodate for transient power draw peaks and to shift the load range lower so the unit would be quieter. Say, if it's a gaming build with Ryzen CPU and GPU on the level of or lower than RTX3060/2070 then i'd recommend 550W PSU because it's enough for maximum load and in games it would be in the range where a PSU would still be quiet enough. From there it gets higher, RTX3070/2080 is 650W, RTX3080/2080Ti is 750W, RTX3090 is 850W. All that depending on the CPU too also, Intel CPUs, especially OC'd tend to draw more power, so if it's something like i9 10900/11900K then i'd add 100W to that. The GPU matters too, more high-end SKUs with wider power limit range would draw more, so it would be a good idea to get say, 1kW for high-end RTX3090 or 850W for high-end RTX3080. If it's not (only) a gaming build but for something like hybrid CPU+GPU rendering and such, and yet the PSU should be relatively quiet, adding another 200-300W would make sense too. On the other hand, if it's a very budget build with something like Ryzen 5 2600 / locked Intel i3/i5 and RX470 / GTX1660 then while not exactly what would i personally use, a group regulated PSU like Corsair CV450 would be just fine if low price is a priority. But i don't see the reason for group regulated PSUs with wattages higher than 550W to exist today, if your PC needs more than 450W even aside crossloads, it means that this build is relatively expensive so you can and should spare some 30-50$ for a better PSU.