Power Supply Draw SC846

[malcolmputer]

I have a SC846E16-R1200B chassis with the dual PWS-1K21P-1R power supplies. I currently have it loaded up with a X10SLL-F with 32GB of RAM and a E3-1241 v3. I have the full 24 bays loaded with mixed HGST HDS72202 and WD WD10EAVS drives.

My question is, my power draw from the wall is ~220W. Does that sound right? According to the 80 PLUS test report (http://www.supermicro.com/products/powersupply/80PLUS/80PLUS_PWS-1K21P-1R.pdf) I should be about 89% efficient at that range.

Is it possible to switch out the supplies in this chassis with the more efficient 920W platinum rated ones? I don't see much info in the chassis manual from supermicro, but I assume it would be ok.

 

[malcolmputer]

Welp, I did a little math, and the drives themselves under use should pull on average 157W, and idle should be 106W, so that explains some of it.

I'm still curious about possibly reducing the draw some by adding a higher efficiency power supply.

 

[jgreco]

As far as I know, yes, that should be fine. If you're going shopping anyways, also be sure to get the 920P SQ ("super quiet") which is platinum rated.

The quickest way to boost your efficiency is to pop out one of the PSU's. You'll probably drop about 20 watts.

 

[depasseg]

I can't imagine the savings gained from using new/higher-efficiency PSU's will offset the cost.

 

[jgreco]

I can't imagine the savings gained from using new/higher-efficiency PSU's will offset the cost.

That's quite possibly correct, though if you're buying a SQ on eBay and you sell your existing modules on eBay, who knows, it could very well work out.

Some of us are willing to trade capex for reduced opex, though, even if it doesn't show a clear obvious win when you project into the future.

For simple opex savings, popping one of the supplies out and keeping it as a cold standby is going to save money.

 

[malcolmputer]

> For simple opex savings, popping one of the supplies out and keeping it as a cold standby is going to save money.

I hadn't thought of that. It will be the first thing I do when I get home.

As for the return on investment. It looks like without reducing the load any (by idle-ing the drives or somesuch) the difference between 89% and 94% at 200W is around 10W. At 10c/kWH that's $8.75/yr difference. Depending on the deal I get one the power supply the payback period would be too long (though, if I attack it from a noise perspective it might be worth it).

 

[jgreco]

Which is why I mentioned the SQ's... not clear that it'll ever be cost effective on a wattage basis, as you note.