Adding another 970 in sli



Absolutely Not. I'm sorry, there's no polite way to say this, but you are simply wrong - combined two 970s plus the i5-4690k alone has a TDP of 560W - With drives that would put the total load on the power supply while in games at around 95% - not only is that egregiously inefficient, it's also just ASKING for a failure.
 
You re confusing TDP with power consumption Werberman.

Here is a single GTX970 PEAK power consumption with a factory overclock.

Yes, thats 207W @ PEAK consumption. So lets make it 420W for the SLI GTX970s.


http://tpucdn.com/reviews/Palit/GeForce_GTX_970_JetStream/images/power_peak.gif

Next the I5 consumes the below at peak.
Yes that is 77W. Lets make it a clean 80 W

http://img.hexus.net/v2/cpu/intel/Haswell/4690K/Power2.png


Now 420 + 80 = 500W.

That leaves more than 150Watt for other internal which combined should not be more than 40W.
 
I'm afraid you're mistaken; I'm not confusing anything.
Peak power cannot be used as a guideline, especially in a multi-GPU setup, as internal conditions are not conducive to effective cooling.
TDP is the max power a device can dissipate when running real-world applications. There's a reason it's always so prominently on display in specs
460W Max draw from the 970s http://www.anandtech.com/show/8568/the-geforce-gtx-970-review-feat-evga
88W for the i5-4690k
http://ark.intel.com/products/80811/Intel-Core-i5-4690K-Processor-6M-Cache-up-to-3_90-GHz
Already at 588W without adding the rest of the core components and cooling system. Not an acceptable margin.

 
Computers are seldom the lovely, sterile controlled environments we se in benchmark testing. Ignoring TDP recommendations is a great way to cause all sorts of fun little SLI bugs - Your assertions are irrelevant; the bottom line remains that the power consumed CAN hit well over 600W, much too close to the max of the PSU. Is it likely? No. But personally, I would not take the chance, especially when a sufficient PSU can go for as little as $50.
 
Just get the cards if you want them. If your computer shuts down during a stress test, then take the second card out until your new PSU comes in the mail. As long as you're watching things, ready to flip the off switch on your surge protector, nothing is going to burn up.

To be safer, you can just get a good PSU. The AX860 or AX 860i would be good and quiet for that setup, since the fan won't come on unless you're around 400 watts or more, plus they're really reliable. Good luck!
 


So the computer is so hot that it doubles as a bread toaster? Unforunately, most computer assembler have no idea how much power a computer really draws. Only recite what they have heard.

So that help lines do not waste time teaching computers assemlbers about electricity, we simply tell them to use at least double the wattage. Most computers consume between 100 and 200 watts most of the time. And rarely if ever exceed 300 watts. That means we tell computer assemblers they need 600 watts. Then profiteers tell computer assemblers that more watts means higher quality. It works in a market full of people who have no idea how to select a supply.

Is a supply too small? That can be answered in less than a minute with a $5 tool called a meter. A tool so simple that even a 13 year old can do it. Supplies are selected by a current for each voltage; not by watts. Meter can confirm that current using a method little understood by most computer techs. It is just easier to preach a Tim Allen joke of "More power". Which is why so many have massive power supplies that, if truly loaded, the computer would also toast bread.

What happens if a supply is too small? Computer simply powers off. No damage to anything - not the CPU - not a disk drive - not even the PSU. Power in PCs never even approaches 600 watts.
 

Sometimes a computer with an undersized supply continues to work. Sometimes it fails intermittently. Sometimes it does not boot at all. Sometimes it works for months and then start acting flakey. But in every case, numbers using a meter can identify an undersized supply.

 

The informed measured power consumption rather then recite hearsay as if fact. Power is nowhere near 600 watts.

In "What is your desktop power usage while browsing these forums?" :
http://forums.anandtech.com/showthread.php?p=34917091#post34917091
> 41 watts idle - 109 watts
or a largest power consumer:
> 250-300w surfing web, 325w playing a typical game

Another measured power consumption:
http://forums.anandtech.com/showthread.php?p=35547012
> I had been getting 35W idle ... it peaked at 102W booting up."

Claims based in hearsay are classic junk science reasoning. A computer consumes far less than 600 watts. Any tech with basic electrical knowledge would have measured that power before posting. Other who did discovered no computer consumes anywhere near 600 watts.

Is that 80 mm chassis fan keeping a 600 watt computer cool - as speculated? Numbers that somehow got ignored. One 80 mm fan removing 600 watts means internal chassis temperatures will be at 140 degrees F (assuming a 70 degree room).

An 80 mm fan obviously is not removing 600 watts. Internal temperatures are nowhere near 140 degrees. Only wild speculation proclaimed a fan removes 600 watts so he need not admit error.

Computer assemblers are told to buy a PSU at least double the necessary wattage. Then computer assemblers need not learn basic electrical concepts. Or learn how to measure electricity. Or perform simple multiplication for a cooling fan.

Computers consume no where near to 600 watts that we tell computer assemblers to buy. As demonstrated how many times? Honesty means also saying why a conclusion with numbers is known.