Help Buying Psu

Gian-027

Prominent
Jun 18, 2017
3
0
510
Only Psu left for my new system and since i dont know much about them i will appriciate any recommendations.

CPu: i7 7700k
Cooler: Cryorig h5 universal
M/B: Asus Z270h rog strix
GPU: Asus 1080ti rog strix
SSD: Crucial 525gb m2
Ram : 16 gb 3200mhz G-skill\
Case: Cougar Panzer Max

Budget for Psu : 140-150 €

 
Solution
Trust me, I know exactly how PSUs work and what's really needed. The 1080ti is a 250W card. Lets pretend it's an OC'd card that goes up to 300W. He has a 7700K, which is another ~100W. I toss another 50W in for the rest of the system. That gives us 450W, probably less seeing as gaming doesn't load the CPU and GPU to 100%. Again, 450W. Here is our sister sites review of the 1080TI FE.

http://www.anandtech.com/show/11180/the-nvidia-geforce-gtx-1080-ti-review/16

Measured 394W for Furmark and 403W for Crysis, so we know I'm figuring correctly though a bit high. If the load is 400W and you don't want to go above 80% load on a PSU, that means you need a 500W PSU to run one at 80% load. An 850W PSU wouldn't even be loaded to 50%...

4745454b

Titan
Moderator
850-1kW? For a system that draws <400W? You won't even load that to 50% while gaming, and idle will barely register at all. Horrible choice of PSU as it's much to large for the system.

A good quality 500-550W will do fine. I'd only go with a 750-850W if you are running SLI.
 

YoAndy

Reputable
Jan 27, 2017
1,277
2
5,665


:no: Lol Calm down your horses. His average system load at the moment is around 500w-550w, if he decides to overclock and run another 1080ti he will be close to a load of 750w. You should know that running a power supply at his maximum capacity is not a good idea.

ACCORDING TO NVIDEA GEFORCE The Recommended (MINIMUM) System Power (W) If using a single GTX 1080Ti is 600W

Here is Guru3D's power supply recommendation: Nvidia GeForce GTX 1080 Ti - On your average system the card requires you to have a minimum of 600~650 Watts power supply unit.

LETS MAKE THIS CLEAR, A MINIMUM OF 500W IS REQUIRED WITH A GTX 1080. HE HAS THE 1080 Ti THAT REQUIRES A MINIMUM OF 600-650W. But when it comes to power who wants their new system running withing minimum specks?
 

4745454b

Titan
Moderator
Trust me, I know exactly how PSUs work and what's really needed. The 1080ti is a 250W card. Lets pretend it's an OC'd card that goes up to 300W. He has a 7700K, which is another ~100W. I toss another 50W in for the rest of the system. That gives us 450W, probably less seeing as gaming doesn't load the CPU and GPU to 100%. Again, 450W. Here is our sister sites review of the 1080TI FE.

http://www.anandtech.com/show/11180/the-nvidia-geforce-gtx-1080-ti-review/16

Measured 394W for Furmark and 403W for Crysis, so we know I'm figuring correctly though a bit high. If the load is 400W and you don't want to go above 80% load on a PSU, that means you need a 500W PSU to run one at 80% load. An 850W PSU wouldn't even be loaded to 50% while gaming, and would be sub 20%. This falls below the target needed for 80+, and you aren't guaranteed 82% efficiency. A 500-550W PSU would be fine for this system provided it's a REAL 500-550W PSU. 850-1kW is simply way more PSU than needed. Even under load it won't hit 50% of the output, and efficiency won't be as good as it can be. They suggest 600W PSUs and what not to cover themselves for people who get those crap $20 "600W" PSUs that have a single 6pin plug and you have to use adapters to power the card. I've proven over time I know what I'm talking about. I highly suggest waving your finger at someone else who is more clueless.

i Think i will go with a 750-850 one
EVGA g3 750w will be ok?

See above. The G3s are fine, but 750-850W is simply too much PSU for your system.
 
Solution

YoAndy

Reputable
Jan 27, 2017
1,277
2
5,665


Yes a 750 w PSU is great for a GTX 1080Ti with room for future upgrades, 600w would be the minimum. Your system is highly overclockable I wouldn't go less than 700. Always remember that "Heat kills" The less you stress your PSU the less heat it generates and the more it will last.
 

YoAndy

Reputable
Jan 27, 2017
1,277
2
5,665


I'm not an expert in every single thing but I'm an engineer and Power Supplies are not that simple to understand. The ratio of output power to input power expressed as a percentage. Because a power supply has a rating of 600w doesn't mean that can be loaded that much. For example, a power supply that produces 350W output but requires 500W input is 70% efficient. In general, a good power supply is between 70% and 80% efficient, although efficiency depends on how heavily the power supply is loaded. Calculating efficiency is difficult, because PC power supplies are switching power supplies rather than linear power supplies. The easiest way to think about this is to imagine the switching power supply drawing high current for a fraction of the time it is running and no current the remainder of the time. The percentage of the time it draws current is called the power factor, which is typically 70% for a standard PC power supply. In other words, a 350W PC power supply actually requires 500W input 70% of the time and 0W 30% of the time.
Combining power factor with efficiency yields some interesting numbers. The power supply supplies 350W, but the 70% power factor means that it requires 500W 70% of the time. However, the 70% efficiency means that rather than actually drawing 500W, it must draw more, in the ratio 500W/0.7, or about 714W. If you examine the specifications plate for a 350W power supply, you may find that in order to supply 350W nominal, which is 350W/110V or about 3.18 amps, it must actually draw up to 714W/110V or about 6.5 amps. Other factors may increase that actual maximum amperage, so it's common to see 300W or 350W power supplies that actually draw as much as 8 or 10 amps maximum. That variance has planning implications, both for electrical circuits and for UPSs, which must be sized to accommodate the actual amperage draw rather than the rated output wattage.

High efficiency is desirable for two reasons. First, it reduces the electric bill. For example, if your system actually draws 200W, a 67%- efficient power supply consumes 300W (200/0.67) to provide that 200W, wasting 33% of the electricity you're paying for. An 80%- efficient power supply consumes only 250W (200/0.80) to provide that same 200W to your system. Second, wasted power is converted to heat inside your system. With the 67%-efficient power supply, your system must rid itself of 100W of waste heat, versus half that with the 80%- efficient power supply.
 

Wizard61

Prominent
Jun 22, 2017
89
0
660
Personally I would go with either the EVGA Supernova 750 or 850 T2 Titanium PSU's. My reason for this is that at 94+% efficiency you will actually save money in the long run over a lower cost and less efficient PSU. Are these PSUs overkill? Yes, but you will have headroom for further expansion and you will have far less heat to deal with as well.
 

4745454b

Titan
Moderator
In general, a good power supply is between 70% and 80% efficient...The percentage of the time it draws current is called the power factor, which is typically 70% for a standard PC power supply.

Your numbers are so off it's nearly comical if it wasn't for the fact that you think you are right and are spending someone's money. (For the record your info is right if it was ~20years ago.) Base PSUs are at least 80+, which means they are 80% efficient. Those are the base low end ones. The unit I have in my machine is an 80+ gold, which means as long as you are running it at 20% load it's 87%, and goes up from there.

https://en.wikipedia.org/wiki/80_Plus

Nearly all PSUs sold today have a PF of 95 or more. Most of them run .99. Here is a random review I picked at Jonyguru which is considered THE site to test PSUs.

http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story4&reid=521

Notice that normally the efficiency is above 85% with most of the real loads showing above 90%, and the PF is also normally above .98. (I'm ignoring the CL, that doesn't normally happen though CL2 could.)

I find it funny that I just went to jonyguru and picked the first review, which happened to be a 1050W PSU. This is at the upper end of what you suggested, so I can show what happens when you pick such a unit. Lets look at what happens when he's just surfing the web or watching netflix/hulu. The basic "idle" power draw of a system is ~100W. Or ~10% of the output of the PSU. What's the efficiency? What's the PF? The first test they run the PSU is outputting 108.7W, for a 87.5 efficiency number. PF is also the lowest you see at .958. If he bought a smaller more appropriate unit? (Had to search for a platinum model seeing as the first one I linked was a plat.)

http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story4&reid=465

At the ~100W load the numbers are now 90.4% and PF .981. Not only would this 550W PSU power his system, it would be cheaper out the door and cheaper to run at some loads. Why spend more money on a large PSU if it doesn't work better at all loads?

Rather than continue to argue about this and ruin this guys thread, why don't you just tell me what load you find appropriate to load a PSU up to? From what you suggest, you are fine with a sub 20% idle load, and a sub 50% gaming load. Which to me is crazy. Most people suggest or shoot for loading a PSU to 50% while gaming, in this case an 800W PSU. (806W if you want to just double that 403W Crysis load I linked earlier.) If you go back to the Jonyguru pages I linked earlier both PSUs start to lose efficiency between tests 4 and 5, which is around 75-80% so that's what I shoot for. As I showed earlier this means you need a ~500W PSU, not 1kW. (400 is 80% of 500. Go 550W if you want it closer to 75%.) Perhaps if you explained why loading to even 50% is bad your choice would make more sense. Because many for that figure when buying a PSU. Yes it's max efficiency, but your idle figures will be horrible.
 

YoAndy

Reputable
Jan 27, 2017
1,277
2
5,665


That's a good idea.
 

YoAndy

Reputable
Jan 27, 2017
1,277
2
5,665


Not all power supplies are 80+ Bronze, Gold or Platinum, and even if it states that if is not a reputable brand is most likely a lie. Trust me I have tested a few I have equipment at work and home , I have tested one made by Apevia 80+, well long story short was rated for 700w my max tested output was 625. Died after a year of regular use.
2202-IMGP1921.jpg
2202-IMGP1919.jpg
 

4745454b

Titan
Moderator
Good catch madmatt. I saw the earlier newegg links and just looked for a more appropriate one there. 650W is still a bit high for my tastes, but it's not totally out of line like the other suggestions. That's a 62% load on the PSU so not horrid. I personally am not a fan of bronze units anymore though I understand outside the USA bronze units might be the best choice. The OP should seriously consider that unit.
 

Rogue Leader

It's a trap!
Moderator


Actually Apevia is 80+ certified

https://plugloadsolutions.com/80PlusPowerSuppliesDetail.aspx?id=1376&type=2

Yes some do lie, and Apevia is garbage otherwise. I don't see how in this entire discussion low end junk PSUs even factor into the equation. 4745454b has posted clear evidence about the power supplies in question, and I know for a fact he has a lot of experience working with PSUs outside of this site even. You have been posting statistics from websites that are known to be overblown. In my own experience which is not nearly his I can tell you that system even under the heaviest load will just about hit 500w, on a spike. Most of the time it would be well within the efficiency range of a good 550w unit. And GOOD units like we have discussed are running well over 80% efficiency. Heck even mediocre units these days are well over 80%.

Finally why would you post 2 photos from a PSU testing website, those are not your photos, and you gave no context as to why they were posted. To try and boost your answer's validity? We all know what a load tester and power meter are.
 

There is so much complexity that nothing can be generalized. The only thing that can be looked at is one specific unit, and the only way to say something about that one specific unit is based on a professional review of it or one of its siblings based off the same platform. In the Protection Features Evaluation of the Seasonic Prime Titanium 650W review on Tomshardware, it is loaded to a whopping 925W before overpower protection kicks in, and I quote, "with all of its rails staying within the ATX standard's specifications". Really, the way these companies rate their power supplies depends on a lot of factors. But there is no standard by which rated wattage is derived. For scammer companies they write too high of a number, for companies with a good reputation they write a lower number because then it tests better when tested within those boundaries.

For example, if they had rated that Seasonic Prime Titanium as 750W instead, it would perform slightly less well on ripple and voltage regulation tests (and other stuff, like holdup time tests). So it is in Seasonic's best interests to rate the power supply at a point like 650W where it can do exceptionally on reviews and also have a ton of leeway for large spikes of power, going up to 925W before shutting down.


That's not true. Look at modern day power supply reviews and you will see this.


Correct, and 80+ unfortunately is an old standard being replaced by http://cybenetics.com - fortunately. 80+ only tests efficiency at 4 load levels, so low-load efficiency and crossload situations are never taken into account. Thank goodness for Cybenetics, which is founded by Tomshardware's great PSU reviewer Aris.


Most all modern PSUs have power factor correction, and it's generally 97% and higher all the time. See the Jonnyguru hot testing of the new Corsair TX850M. Those power factor values are very high. Additionally, the power factor is not the ratio if output / input power. It is the ratio of active / apparent power. Source: http://www.hardwaresecrets.com/everything-you-need-to-know-about-power-supplies/10/



You are misunderstanding power factor correction. From this source on Jonnyguru I get this information: "Power factor correction won’t make your power supply more efficient (convert more DC output power with less AC input power), but can allow for more devices to be plugged into the same circuit.". A low power factor does not increase input power, that's not how it works. I'll admit I'm not an expert on power factor, it's darn confusing, but I also know what you're saying is not entirely true with all due respect.


A load table on a power supply does not say anything like this. It merely talks about continuous or peak (hopefully coninuous) power output in amperes on the rails.



True. The fan and design of a power supply also plays a role and such.