PSU tier list 2.0

Page 210 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Depends on the breaking point. Caps do degenerate over time, lower quality caps going much faster than high quality. Lower quality caps also have a greater failure rate due to impurities, bad manufacturing etc, all this combined with time under a consistent load which then gets added too is often enough to push the unit over the edge and it fails. If left where it was it prolly last a good while longer.
 


The way I interpreted his post seemed as if he was implying that running a GTX 960 from the get-go would result in a longer-lasting power supply compared to a system that was upgraded from a GTX 650 into a GTX 960.
 
That's entirely possible. What gets me is both the 650 and 960 have the same power requirements, 20A and 400w. What's probably different is op got a new game that's much higher demanding than what he previously had and is now driving the system much harder than what it was before
 


The 650 is a 65W GPU, the 960 is a 120W GPU. Not at all the same power requirements.
 
I said same requirements. As in 20A and 400w psu. I didn't say same consumption. But the 960 is considerably more powerful than the 650, so it's entirely possible that the op is driving the gpu much harder and using both the cpu and gpu to more excessive amounts, simply because he can now run some games on ulyra/high and good fps, where's with the 650 that wasn't possible. This can easily bump up demand on the psu, over and above what it is capable of withstanding vrs what it should be able to withstand.
 
Entire system recommendations don't make sense, that's what I was talking; how can they possibly know your system specs? I guess it's a recommendation and not a requirement technically, but it still seems silly to me. Basing the entire system off just the GPU.
 
It makes sense for the average user who is trying to figure out if they can put a GTX 960 into their Dell. Power users who are building systems (should) understand that a system with an i3 will be significantly different than a system with an FX 9590
 
Do the math. Avg. cpu is 100w. Avg. mobo is 50-100w. Max for a dual 6pin is 225w. So the entire system is capable of 425w. Avg. running would be around 250w.Then take into account the differences between psus. So you'd be looking at a 500w psu to maintain @50% efficiency during avg. use, a good quality 450w minimum would cover everything. You could run with less for sure, it's been proven repeatedly, but generally by those who have a clue exactly what their system really pulls or taking into account minimums like SSDs, i3's no optical etc but the general public is not generally that educated so recommended requirements have to take Insignia, startech, Huntkey etc into consideration so for a 2x 6pin you'll generally see a 500w or so minimum recommended requirement.
 


The motherboard number is super inflated. A locked Intel CPU under full load should result in maybe 20-25W for the motherboard. Enthusiast boards and overclocked CPUs together can indeed get into 50W territory, but your normal boards on an average person's system should be about 20W when gaming, or less, since it is directly relative to CPU load.

100W can be a little high, too, for Intel CPUs at least. Skylake CPUs should be more like 60W when gaming.

Here is the I7-6700K and I5-6600K reviews http://www.tomshardware.com/reviews/skylake-intel-core-i7-6700k-core-i5-6600k,4252-11.html

The 6600K is 69W when gaming, and the 6700K is 83W. But that's with the iGPU! WIthout the iGPU, the I5-6600K is about 52W when gaming, and the I7 more like 75W.
KpIJMu0.png


It really is incredible how efficient Skylake CPUs are; even I was overestimating that. Even under a torture test the 6700K highest single point does not surpass 102W and the 6600K also remains at about 75W. I am really anticipating to see how Zen will turn up in terms of efficiency, since AMD's CPUs are still pretty high in power.
 
Turkey, that's just the mobo, once you add 15w for the hdd, 3w for ssd, upto 6w for every fan, all the crap added to USB ports like kb's, optical mouse, Bluetooth, iPad chargers etc, the total wattage easily breaches the 50w mark and can close on the 100w top end.
 
NVIDIA actually specifies the reference system specifications:

"Minimum system power requirement based on a PC configured with an Intel Core i7 3.2GHz 130 Watt TDP processor."

15 Amps (i.e. excluding graphics card) is what they allow for that reference system. It's the figure they've used since the Nehalem Microarchitecture days.
 
I wonder....
Of all the people that go back and forth about this wattage or that....something being too low, and this being needed.

How many of you have actually monitored your Kill-A-Watt device, with your system under full load.
Gaming, Prime95, Heaven, whatever benchmarking tool you choose.
 




I don't need to, all that information is free online. Tomshardware, Anandtech, Techpowerup, HardOCP, etc. all run power tests on different configurations, or with a site like Toms and the expensive equipment even uses a scope to analyze power of a specific piece of hardware.
 


Yes, I know this.
However, watching the readout on my Kill-A-Watt was actually eyeopening.

But many, many people pull numbers outta <somewhere> and preach that as gospel, never to be contradicted.

I wanna see some real people, real world numbers.
 


Well that's why sources must be cited when it comes to this stuff. Eyeopening in what way?
 


Eyeopening in that the number pulled from the wall, under 100% CPU usage, was far less than what was expected.

If the lighting under my desk were better, and I had the time this Sunday afternoon, I'd post a pic of the Kill-A-Watt readout.
But I have an exhaust to fix, a fence to finish demolishing, and shish kebob to cook for dinner.
Maybe later.
 
And not knowing exactly what the current efficiency of the psu is under that specific cpu load, you'll still have to pull a number out of your <somewhere > to do any guessing at what the actual output of the psu is. And this can change if trying to run p95 with a gpu killer to maximise possible gaming loads. After all the fans ramp up, of course, and don't forget to plug a USB in for an iPhone.

But as basically stated, its mostly conjecture, its a vastly different scenario with a 6700k, case full of fans and gear vrs an i3-6100 in a simple office type setup. So allowances gotta be made somewhere, and further again for ppls choice in junk psus due to budget pricing
 
You buy a $39 CWT Corsair and it dies in 18 months. What are the chances you RMA it? Slim? None? Chances are it's only going 1 place, the bin. So RMA's are totally unreliable. You spend $120 for a Seasonic built Corsair and it dies in 18 months, that's a different story. It would be easy to have 100 Seasonic RMA in its 5 yr Warranty, but you could just as easy have 10,000 CWT die within its 3 yrs, only to have 100 RMA, simply cuz they died in the first week, not 18 months later.
 
i got the seasonic m12ii evo in 2015 and it is not in the list and it is unrated yet :/ what tier is this ? is my seasonic have build in surge protector ?
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom