550w 80+ Gold or 750w 80+ silver (originally 650 but explanation below)

[Jordan_C21]

build will contain in the near future, r5 3600 (overclocked) 2060 super (overclocked), or 5700xt (overclocked) mobo, 16gb ram, all other pc things. 

 

First deal I saw and its the most selling on newegg for the fantastech sale, the 650w Corsair vengeance, its rated B on the PSU forum list didnt find out till later. Then theres the SeaSonic Focus Plus Gold 550w, now im not sure if those components with all the OC going on if 550 can handle that and be good for the future, and since the 650w sold out, since it was like $10 more I got the 750w corsair vengeance and ik 750 is MASSIVELY overkill but why not yanno. Well newegg wont let me cancel the order but I could just return it and say get the 550w if it better suits my needs. Sorry about all the posting but yanno better safe then sorry my friends. So the real question is, do I return and get the 550 or stick with this and be good.

[APasz]

For such a system with overclocking, 700 (edit: meant 600, I love my typos) is a minimum.

The thing about PSU's is that you want 'overkill' (at least a little), assuming you don't want to be buying a new one every few years.

The rule of thumb I follow is, CPU + GPU tdp + 100 (+20 on Intel), then you want that to be 50-70% of your total PSU output.

[Jordan_C21]

13 minutes ago, APasz said:

For such a system with overclocking, 700 is a minimum.

The thing about PSU's is that you want 'overkill' (at least a little), assuming you don't want to be buying a new one every few years.

The rule of thumb I follow is, CPU + GPU tdp + 100 (+20 on Intel), then you want that to be 50-70% of your total PSU output.

so if im overclocking everything in general, i should just keep it? okay dude wonderful thank you. Since your bronze im sure you would know the psu tier list, its ranked B this technically is "budget" but ive been told to not skimp out on psu obviously, is the corsair vengeance 750w good or should i swap it out?

[seon123]

13 minutes ago, APasz said:

The thing about PSU's is that you want 'overkill' (at least a little), assuming you don't want to be buying a new one every few years.

Why is that? You do realise there is more to a PSU than just the wattage? It's better to get a high end 450W like the Whisper M 450W than e.g. an EVGA G1 750W for a 2060 Super system, even assuming overclocking. 

Even within high end PSUs, you want to avoid PSUs like the EVGA G2 1600W or G3 1000W, as those are really noisy and lack protections, compared to the Whisper M 450W. 

15 minutes ago, APasz said:

CPU + GPU tdp

CPU TDP and GPU TDP values are not the same thing, and are wildly inconsistent. E.g. Intel uses a random number from a RNG, while AMD averages a few different loads. Nvidia uses the GPU core's power consumption, while AMD uses the whole board's power. 

17 minutes ago, APasz said:

then you want that to be 50-70% of your total PSU output.

Where did you get that random number?

[Jordan_C21]

5 minutes ago, seon123 said:

Why is that? You do realise there is more to a PSU than just the wattage? It's better to get a high end 450W like the Whisper M 450W than e.g. an EVGA G1 750W for a 2060 Super system, even assuming overclocking. 

Even within high end PSUs, you want to avoid PSUs like the EVGA G2 1600W or G3 1000W, as those are really noisy and lack protections, compared to the Whisper M 450W. 

CPU TDP and GPU TDP values are not the same thing, and are wildly inconsistent. E.g. Intel uses a random number from a RNG, while AMD averages a few different loads. Nvidia uses the GPU core's power consumption, while AMD uses the whole board's power. 

Where did you get that random number?

sounds like you be knowing more, back to my question, should i return and then what (hasnt even delievered)

[seon123]

2 minutes ago, Jordan_C21 said:

sounds like you be knowing more, back to my question, should i return and then what (hasnt even delievered)

Nah, you're more than fine. Iirc, the new Vengeance Silver has a switch to select single or multi rail. Make sure it's on multi rail. It should be by default. 

[APasz]

1 hour ago, seon123 said:

Why is that? You do realise there is more to a PSU than just the wattage? It's better to get a high end 450W like the Whisper M 450W than e.g. an EVGA G1 750W for a 2060 Super system, even assuming overclocking. 

Even within high end PSUs, you want to avoid PSUs like the EVGA G2 1600W or G3 1000W, as those are really noisy and lack protections, compared to the Whisper M 450W. 

CPU TDP and GPU TDP values are not the same thing, and are wildly inconsistent. E.g. Intel uses a random number from a RNG, while AMD averages a few different loads. Nvidia uses the GPU core's power consumption, while AMD uses the whole board's power. 

Where did you get that random number?

Yes but I didn't realise we were going to recommend crap PSU's, note that I didn't mention any specific PSU. Sorry for my assumption.

I say go 'a least a little overkill' because I've had too many cases of people thinking they only need a 400w PSU for a 100w CPU and 200w GPU. Then coming to me 6-12 months later wondering why their PC wont turn on any more.

 

Intel's TDP is at base clock. AMD is at boost. And as you said, NVidia does the GPU core while AMD does whole board but were talking tiny differences here.

 

Knowledge of what a PSU is and over a decade of experience building PC's.
Anyway it's electronics 101. Never run a power supply at full load for an extended period.

Generally speaking the optimal load for a computer PSU is around 60% give or take 15% depending on it's brand, size, wattage, certification, multi-rail vs single rail, etc.

Again it's rule of thumb with PSU's, everything must be taken with a grain of salt.

 

Please do your research.

[Jordan_C21]

19 minutes ago, APasz said:

Yes but I didn't realise we were going to recommend crap PSU's, note that I didn't mention any specific PSU. Sorry for my assumption.

I say go 'a least a little overkill' because I've had too many cases of people thinking they only need a 400w PSU for a 100w CPU and 200w GPU. Then coming to me 6-12 months later wondering why their PC wont turn on any more.

 

Intel's TDP is at base clock. AMD is at boost. And as you said, NVidia does the GPU core while AMD does whole board but were talking tiny differences here.

 

Knowledge of what a PSU is and over a decade of experience building PC's.
Anyway it's electronics 101. Never run a power supply at full load for an extended period.

Generally speaking the optimal load for a computer PSU is around 60% give or take 15% depending on it's brand, size, wattage, certification, multi-rail vs single rail, etc.

Again it's rule of thumb with PSU's, everything must be taken with a grain of salt.

 

Please do your research.

so final verdict is???
 

[Jordan_C21]

1 hour ago, seon123 said:

Nah, you're more than fine. Iirc, the new Vengeance Silver has a switch to select single or multi rail. Make sure it's on multi rail. It should be by default. 

so keep the 750w keep the overkill, enjoy the deal, turn multi rail on with switch, and we good?

 

[APasz]

2 hours ago, seon123 said:

~snip~

You like getting hung up on specifics mate.

 

Since I apparently don't know what I'm talking about, why don't you teach the class then?

 

MCE, multi-core enhancement? And? What, is that meant to be a gotcha or something?

And what do you mean sneaky? Turbo Boost across all cores is now becoming a standard feature of Turbo Boost.

 

Can you give an example on what I haven't learned?

 

A 750w PSU is rated to output 750w over a given amount of time, doesn't mean it's going to be efficient while delivering that 750w.

Sure it can output more than that but that is no good. It can cause a fire, or at the very least a short. Also remember the 400W is internal, the PSU pulls more from the wall.

"continuous" forming an unbroken whole; without interruption.

 

Please show the research that demonstrates PSU's are most efficient at their peak output.

BTW have you looked at the 80plus certification table lately?

Funny how the required most efficient load in this table is at 50%.

Now why would they require a higher efficiency rating at 50% if running at 100% is supposedly just as efficient.

Admittedly I should have said 65%, give or take up to 15% but typo, whatever.

 

I'll give you that one, not sure why I put multi-rail in there.

 

Of course the OTP has an effect on efficiency. Cause if it triggers your efficiency goes way down.

[Spotty]

@APasz I'm going to hit you up with a few quotes and address some of your comments individually, and try my best to explain what I can. Seems like some of the things you have said may not be entirely accurate or may be based on outdated information, so I'd like to help clear it up not only for yourself but for anyone else who may be reading who might not be aware. 
If you have any questions feel free to ask, and if I know the answer I'll do my best to try and explain it.

 

 

8 hours ago, APasz said:

The rule of thumb I follow is, CPU + GPU tdp + 100 (+20 on Intel)

This probably isn't the best way to calculate the wattage. You can kind of ballpark guesstimate where it might be from TDP, but as addressed already TDP does not equal power consumption, especially on things like Intel CPUs which can run above TDP for periods of time.
The way I personally prefer to calculate the wattage requirements of a system is to look up reviews that measure the power consumption of individual components using proper testing equipment (such as current clamps). You can put a current clamp on the EPS12V cable to the CPU and that will give you a very good representation of how much power the CPU is consuming. Likewise if done properly you can do the same with PCIe devices (graphics cards). The highest power draw components in a PC that you should be concerned about are the CPU and the Graphics Card(s). Other things such as fans, RGB, HDDs, etc do not consume a lot of power on their own, but when you add them all up it may be up to 50 or so watts depending on the system.

Alternatively some media outlets will measure entire system power consumption, which will be less exact, but still close enough to get a general idea of how much power you should expect a system to consume.

 

Basically you should look at the measured power consumption of parts, rather than trying to guess it yourself. I quite like Tom's Hardware reviews as they do a good job with individual component measurements and have an extensive catalogue of tests that will cover most hardware, however other outlets also do such testing as well and Gamers Nexus has recently moved to doing individual component power consumption testing as well.

 

 

6 hours ago, APasz said:

I say go 'a least a little overkill' because I've had too many cases of people thinking they only need a 400w PSU for a 100w CPU and 200w GPU. Then coming to me 6-12 months later wondering why their PC wont turn on any more.

A good quality power supply rated to output 400w continuously should have no issue running a system that consumes less than 400w. If they are coming back after 6-12 months with a PSU failure then it's likely the quality of the PSU being used (there aren't that many quality 400w units), and not expressly due to the load put on it. Though, I'm assuming that it was a hypothetical scenario you were referencing so I won't get too hung up on this.

 

8 hours ago, APasz said:

then you want that to be 50-70% of your total PSU output. 

6 hours ago, APasz said:

Anyway it's electronics 101. Never run a power supply at full load for an extended period. 

Generally speaking the optimal load for a computer PSU is around 60% give or take 15% depending on it's brand, size, wattage, certification, multi-rail vs single rail, etc. 

Again it's rule of thumb with PSU's, everything must be taken with a grain of salt.

There seems to be two main issues here you're touching on.

1. PSUs aren't capable of running at their rated output for extended periods of time
2. PSUs are more efficient at 50% load (or some other arbitary figure)

Both of these are incorrect.

 

Continuous Output

The output rating on a power supply is for continuous output, and is rated at a certain ambient temperature. Depending on the power supply it may be rated between 30°C to 50°C. Most "decent" power supplies will be rated at least 40°C ambient temperature, with high end units often rated at 50°C ambient. Some low end budget units may be rated at 30°C. Some really poor quality units' output may only be rated at 25°C (EVGA N1 for example), and those should most definitely be avoided at all cost.

On some particularly old or poor quality units (eg. absolute junk that claims it's a "750w PSU" despite being only able to deliver 400w on the 12v rail) may advertise what they call "Peak Output". This is basically entirely worthless and doesn't really mean anything. As far as I'm aware there's no standards or definitions on what "Peak Output" is how how long it should be able to sustain that peak output for.
Thankfully, no self respecting power supply will show "Peak Output" on its label. Anything worth buying would show "Continuous Output".

 

Continuous output is what it says - It is what the PSU is capable of outputting continuously, ie. 24/7 365. When operating within the temperature range the PSU is rated for within the continuous output range the power supply will perform within spec and will meet it's 80+ efficiency claims.

If you over load a power supply, or operate it in temperatures outside its operating range - for example operating a 550w PSU rated at 30C at 650w in a 45C hotbox, then (if protections don't trip to shut it down) you will see a range of issues with the power supply including but not limited to; de-rated efficiency, less voltage regulation (voltage drops as current rises too high), higher ripple (potentially going out of spec), and potentially even failure.

 

This only applies to power supplies that aren't lying on the label. If you're buying a $15 750w Power Supply that sounds too good to be true, then it will probably shut down at 300w load. That's not because power supplies can't run at their rated output continuously - It's because that PSU is a piece of junk and shouldn't be used.

 

TLDR; Any good quality power supply will happily run at 100% load continuously for extended periods of time without issue. It's only an issue if you're talking about power supplies that are operating outside of the rated operating range or are junk (lying on the label, etc)

 

 

PSUs are most efficient at 50%

The entire "PSUs are most efficient at 50%" started simply because the 80+ efficiency testing for power supplies rated power supplies for efficiency at three load levels - 20% load, 50% load - 100% load. The efficiency target for the 50% load was higher than the 20% and 100% figure, which lead many people to think that 50% is a sweet spot for efficiency where the power supply is greatly more efficient.

 

This is largely untrue, misconceived, or overstated.

The efficiency of a power supply depends entirely on the individual model and the design of the unit. Most* modern power supplies tend to peak at approximately 30% load with a very slow and gradual efficiency drop towards 100%.

 

As an example of a modern PSU's efficiency curve, here is the Corsair RM750x efficiency curve.

 

You can see a very sudden and sharp rise in efficiency up to around 10%, when it gradually increases towards 20% load before peaking somewhere around 30-40% load. Then there is a very slight and gradual decline in efficiency towards 100%.

 

The efficiency difference between a power supplies peak efficiency and other loads is highly overstated and exaggerated. This "peak" however is very slight, and the difference between 30% load and 70% load may be within a range of less than 1% efficiency. The difference between 50% load and 100% load may be around 2% or so. Something that is not going to cause a noticeable difference in your electricity bill.

 

Another thing to consider is since efficiency tends to be lowest at low loads, by buying a higher wattage unit than required you may fall in to a lower efficiency level at lower/idle loads - which is where a lot of systems spend most of their time. For example a 1000W PSU at idle using 60w may only be around 70% efficient.

 

If efficiency is important to you, then you should look at more efficient units, rather than higher wattage units. It's likely that an 80+ Bronze unit is less efficient at 50% load (85% min) than an 80+ gold unit is at 100% load (87% min).

For a system that requires 350w, for the cost of an 80+ Bronze 750w unit to reach "50% peak" you could likely afford a good quality 450w or 550w 80+ Gold unit.

 

 

4 hours ago, APasz said:

Please show the research that demonstrates PSU's are most efficient at their peak output.

BTW have you looked at the 80plus certification table lately?
 

Funny how the required most efficient load in this table is at 50%.

Now why would they require a higher efficiency rating at 50% if running at 100% is supposedly just as efficient.

Admittedly I should have said 65%, give or take up to 15% but typo, whatever.

Two things to consider when looking at the 80+ efficiency standards.
1) They chose 3 values spread out across the load of the power supply to represent the load across a full range. 20%, 50%, and 100% are just spread out to capture low load, mid load, and full load on the unit. If they only measured one point (say, 100%) then PSU manufacturers could make a PSU that is efficient only at that load, but highly inefficient at other loads. By spreading the testing out over the full range of the power supply it ensures that PSU manufacturers make efficient units at all loads.

2) The 80+ efficiency standards are over 15 years old. They were brought in when different designs and technologies were used in power supplies. At the time there weren't any 80+ units and when making an 80+ bronze efficient unit was actually considered high end. Nowadays with modern designs 80+ bronze is basically guaranteed even with budget PSU designs, with 80+ gold easily achievable with the use of LLC resonant converters and other such designs.

Back in 2005 the efficiency curve looked different than what it looks like on modern units. The "PSUs are most efficient at 50% load" may have been more true on older units, however on modern units this tends not to be the case. (Though, you still need to check efficiency of each model as depending on the designs used efficiency curves may vary)
 

 

5 hours ago, APasz said:

A 750w PSU is rated to output 750w over a given amount of time, doesn't mean it's going to be efficient while delivering that 750w.

Sure it can output more than that but that is no good. It can cause a fire, or at the very least a short. Also remember the 400W is internal, the PSU pulls more from the wall.

"continuous" forming an unbroken whole; without interruption.

See Continuous Output section above.

A PSU rated to 750w continuous output can provide 750w over a continuous period, not over an "amount of time".
It's efficiency at 100% is included in 80+ efficiency testing, so it will need to meet efficiency standards for 100% load in order to use the 80+ efficiency branding and logo. As shown in the efficiency section above, there is a gradual drop in efficiency towards and beyond 100% load, however it's not a significant drop in efficiency and is not concern.

 

Any decent PSU should not catch fire while operating within its operating range - Or at all. If operating outside of a safe range it should have appropriate protections to trip and shut down the power supply before a fire is started. This obviously does not apply to junk power supplies without protections, which is why they're not recommended.

 

Operating a power supply outside of its rated operating range, overloading the unit, should not cause a "Short". A short is where electricity is bridged from one circuit to another. This should not occur if the PSU is overloaded, unless there is otherwise failure within the components of the PSU which cause the short.

You are correct that the rated output is indeed what is supplied to the system and measured after efficiency losses from the wall. So a 400w PSU at 90% efficiency will be pulling 445w or so from the wall AC power. This, however, is not an issue as power supplies are rated for Continuous Output power. They're rated at what they can output to the system, not rated for the input prior to efficiency loss.
 

5 hours ago, APasz said:

Of course the OTP has an effect on efficiency. Cause if it triggers your efficiency goes way down.

OTP = Over Temperature Protection. This will trigger to shut down a power supply if the temperature at a given point within the PSU reaches a point considered unsafe. This isn't related to efficiency, and would more likely trip due to something like fan failure or dust build up.



That was longer than I was expecting. Hopefully that explains a lot of it for you, however if you or anyone else does have any questions feel free to ask.
Also, I'm not perfect, so if I've made any errors feel free to correct me so that I too can learn from my mistakes!

[APasz]

11 hours ago, Spotty said:

~snip~

You're getting hung up on specifics as well.

I've got lots of work so I'll keep this concise, which appears to be a problem.

 

So what is the point in the TDP then? Since you seem to be saying it's useless.

The formula I gave is not the be all end all. Sorry that people seem to think so.

It's meant to be a jumping off point. Nothing more.

If I know a particular part uses a significant amount more than it's rated TDP, I'll factor that in.

Since been concise seems to be difficult. I'll write out all the formulas I use in full. +10 per HDD/SSD and +20 for Intel. +5 per fan/RGB above the first 4.

Single CPU with minimal extras. CPU TDP + 50.

Low end rig, i3 with low end gpu. CPU + GPU TDP + 75.

Average rig, i5, 2060. CPU + GPU TDP + 100.

Above average, i5/i7, 2070. CPU + GPU + 120.

High end, i7, 2080. CPU + GPU TDP + 140.

Proper high end, i9, 2080 ti. CPU + GPU + 160 (+ 30 HEDT.).

 

For my system, it works out to be 555 watts. Under typical gaming loads it gets up to about 530w but under full load on both CPU and GPU it can reach 650w.

For my system I'd get a 700-750w.

 

Yes it was hypothetical.

 

Where did I say PSUs aren't capable of running at rated output continuously?

I said that 50% was more efficient than 100%, not that 50% is the most efficient load. With modern PSUs I'd say that 75%, or even 80% is around the most efficient given how little modern hardware uses while idle but also how high it can spike.

You know sleep mode is a thing right? People who are concerned about idle power usage will put their PC to sleep.

 

The rest of this section appears to be based on a big misunderstanding. Partly my fault for not being verbose enough.

 

Would like to see the warehouse with all the 10+ yo PSUs still doing their continuous output test.

 

I don't expect anal people to get such jokes.

[Matsozetex]

6 hours ago, APasz said:

--snip--

This is a help fourm, if you are going to spread mis-information and make jokes, you don't belong here... Sorry.

[APasz]

It seems a few posts have disappeared.

Yay for mods.