is S12-600 enough? PSU calculator says NO! (aanet.com)

Mandor · Post by **Mandor** » Mon Jul 30, 2007 5:25 am

Hi,

I came straight here for my PSU related interrogations since you guys are the only ones not reccomending 1kw psus for no reason. I bought a nice S12-600 a couple of years ago to power my dual-opteron rig and it served me well.

Now i'm upgrading to a quad-core system (dont ask me why, I just have to!) and 4GB of RAM. I'll be adding a 5th 120mm Panaflo fan to fit on the Ultra-120 Extreme too, replacing my old 92mm one. I will hopefully overclock my new G0 chip to 400x9=3600MHz, which usually takes no more than 1.45v on those chips. Now for all you o/c'ers out there know that I don't EXPECT to attain those speeds but I still need to evaluate the power draw so I might as well list the worst (best?

) case scenario.

Now the power supply calculator reccomended here (http://web.aanet.com.au/SnooP/psucalc.php) says I need at least 546W for a reccomended brand PSU, which Seasonic is, so i'm ok there. The problem is for the 12V rail. The S12 is rated for 36A combined and the calculator says I need 39.7.

Will this be a problem?

Max Slowik · Post by **Max Slowik** » Mon Jul 30, 2007 5:46 am

Now, I have built computers that were actually capable of drawing over six hundred watts, but they all used inefficient chipsets and ridongculous video cards. The practical upshot is, we need to know what else you've got in that box to guestimate its power needs.

Mandor · Post by **Mandor** » Mon Jul 30, 2007 6:03 am

its in my sig. just change the proc for an overclocked q6600 and double the memory. the problem is not for the total wattage but the combined amp rating on the 12V rail (36A and PSU calculator says 39A is needed)

dhanson865 · Post by **dhanson865** » Mon Jul 30, 2007 7:14 am

I don't generally trust PSU calculators but here is another one to cross check. http://www.extreme.outervision.com/psuc ... orlite.jsp

Note, you should be choosing single Processor on the first drop down if your board only has one CPU socket. You choose the number of cores later.

I came up with 432 watts with full load craziness. That calculator (and pretty much all of them) are known to oversell the wattages.

Trying out the 50% overlock it went to 531 watts. I'm not sure if I trust that at all.

Do you have a Kill-a-watt or a PowerAngel to measure your current system?

There is no harm to the parts if you exceed the power draw of the S12, it'll just trigger an OCP and shut down the system, so you might as well just try it and buy a Corsair HX620W if the S12 doesn't do the trick. The corsair has 50A on the 12v rails.

The newer S12s have slightly to a bit more amps on the 12v lines.

S12 -380=25
S12II-380=27

s12 -430=29
S12II-430=30

s12 -500=33
S12II-500=34

S12 -600=36

M12-500=38
M12-600=48
M12-700=58

S12E+ 550=41
S12E+ 650=52

HX520W=40
HX620W=50

merlin · Post by **merlin** » Mon Jul 30, 2007 8:48 am

Mandor wrote:Hi,

I came straight here for my PSU related interrogations since you guys are the only ones not reccomending 1kw psus for no reason. I bought a nice S12-600 a couple of years ago to power my dual-opteron rig and it served me well.

Now i'm upgrading to a quad-core system (dont ask me why, I just have to!) and 4GB of RAM. I'll be adding a 5th 120mm Panaflo fan to fit on the Ultra-120 Extreme too, replacing my old 92mm one. I will hopefully overclock my new G0 chip to 400x9=3600MHz, which usually takes no more than 1.45v on those chips. Now for all you o/c'ers out there know that I don't EXPECT to attain those speeds but I still need to evaluate the power draw so I might as well list the worst (best? ) case scenario.

Now the power supply calculator reccomended here (http://web.aanet.com.au/SnooP/psucalc.php) says I need at least 546W for a reccomended brand PSU, which Seasonic is, so i'm ok there. The problem is for the 12V rail. The S12 is rated for 36A combined and the calculator says I need 39.7.

Will this be a problem?

Let's put it this way, my system is pretty high end and has been overclocked to very similar specs to yours (except 7900gto instead of 8800gts and non-quad) ...it never passes 200 watts in burn. The psu calculator is completely wrong if it really expects that much power. I expect under load, you won't pass 350w if that when o/ced.

Mandor · Post by **Mandor** » Mon Jul 30, 2007 9:13 am

dhanson = I tried this one and it came 600w flush. I checked 20% capacitor aging since the S12 is nearly 3 years old and has been powered on close to 24/7 for downloads and folding. yet, this doesn't rate the 12V rail requirements, which seems to be te problem the other PSU calculator pointed out.

as for the OCP kicking in, I have no worries, I completely trust seasonic psus and I know I won't fry anything with it. I just don't want me to work my ass off on o/c'ing that processor if I end up being limited by the psu not giving me enough juice. Then, i'll be pissed.

I just want to know if I should trust the calculators or guys like merlin who tell me there is no problem.

EsaT · Post by **EsaT** » Mon Jul 30, 2007 9:33 am

Mandor wrote:...says I need at least 546W for a reccomended brand PSU...

You're looking wrong place, "recommended" already contains lot of safety margin!
(Didn't notice "actual power consumption" part?)

And instead of brute force overclocking to certain speed I would prefer overclocking only to point before CPU starts needing bigger voltage increase. (heat output probably starts to rise much faster than clock speed)

dhanson865 · Post by **dhanson865** » Mon Jul 30, 2007 9:38 am

I'd trust guys like Merlin that tell you there is no problem.

If you ever get a Kill-A-Watt or similar device (they only cost about $25 or so) you'll see how little your PSU is actually pulling.

If you can borrow a Gigabyte ODIN PSU you could even tell how much 12V juice you were actually pulling, though if you aren't a reviewer I sure wouldn't pay $100+ more than what a Kill-a-watt costs just to have the detailed numbers.

Besides the lower efficiency would cost you more every day you used that PC. See http://www.jonnyguru.com/review_details.php?id=123 for the review of that PSU.

If I remember correctly when the OCP kicks in you have to use the power switch on the back of the PSU instead of the power button on the case to turn the PSU completely off and then back on. If so it should be easy to tell the difference between overclocking instability (where a reboot occurs automatically or the reset switch is used) and OCP triggers due to the PSU being insufficient.

dhanson865 · Post by **dhanson865** » Mon Jul 30, 2007 9:43 am

oh, and if you are on 220v you already have insurance as most of the PSU calculators are based on 120v efficiency which is worse than 220v.

according to Wikipedia Quebec is on 220v, is that true for you?

Mandor · Post by **Mandor** » Mon Jul 30, 2007 9:49 am

wikipedia is wrong. everything here is 120V. 240V is only for big appliances like washing machines and stuff. pretty much like everywhere in the states, as far as i know.

continuum · Post by **continuum** » Mon Jul 30, 2007 9:52 am

I've got a Q6600 @ 3.2ghz/1.325v, five 10K and 15K harddisks, 4GB... only thing I'm lacking is the high-end video card, and my box doesn't break 350W from the wall at full load. (I forget the actual number...)

With another 100W from your video card that's still under 450W.

I'd say your S12-600 will be fine.

In fact I'm running my box on a S12-600 right now and it doesn't even break a sweat.

Mandor · Post by **Mandor** » Mon Jul 30, 2007 10:14 am

thats great news !

thanx m8!

walle · Post by **walle** » Mon Jul 30, 2007 10:33 am

In my experience, hence my opinion; pc calculators are based on providing potentional buyers with misinformation. Their single purpose (feel free to disagree) is to "assure" that potentional buyers are revealed of their hard earned money, more so, than providing them with accurate information, and that ticks me off. rant mode...oFF

Your Seasonic should find your setup (o/ced or not) a breeze. Also, must say that your unit has aged well, suppose itâ€™s a testimony to a product focused on quality instead of bling.

cheers

merlin · Post by **merlin** » Mon Jul 30, 2007 12:43 pm

walle wrote:In my experience, hence my opinion; pc calculators are based on providing potentional buyers with misinformation. Their single purpose (feel free to disagree) is to "assure" that potentional buyers are revealed of their hard earned money, more so, than providing them with accurate information, and that ticks me off. rant mode...oFF

Your Seasonic should find your setup (o/ced or not) a breeze. Also, must say that your unit has aged well, suppose itâ€™s a testimony to a product focused on quality instead of bling.

cheers

I don't believe they're trying to be false...but they're conservative to the point of giving people like 50% more power than they ever need. It's basically to the point of being 100% useless for REAL power measurements. The only power measurements I trust are those from review sites such as here, where a full dc/ac check are done and we know exactly what the amount of dc power used is. I'd really use mike's system power measuring article as a guideline to your psu needs. Those calculators are junk in comparison most of the time.

walle · Post by **walle** » Mon Jul 30, 2007 2:12 pm

merlin wrote: I don't believe they're trying to be false...but they're conservative to the point of giving people like 50% more power than they ever need.

I would not go as far as to state that they are conservative per say, when giving potentional buyers an additional ~50% in power, power that they will never need. I canâ€™t help but view that as providing potentional buyers with misinformation. As a consequence, those that are in the market for a new power supply will most often end up spending a large amount of money on a product that they donâ€™t need. I for one, find it painful to see people (all to many) purchasing power supplies using these severely flawed calculators as a measuring stick, ending up with a bling bling product which at times doubles their expenses. Of course, it could be argued that they (potentional buyers) should be able to make a sound decision. I would agree to a point, then again; thereâ€™s such a large amount of noise out there making it difficult for anyone, let alone for someone whom is green, to be able to find the correct information. This site however, is a great place, with people that share a common interest, that are knowledgeable and most often eager to help. I can only speak for myself, but it took me quit some time to find this beacon of tranquillity, so I suppose that all one can hope for is that more people will find this site whilst cruising in the minefields out there, most are bound to hit a mine or two, and those ankered ones, are those that really ticks me off!

rant mode 2...oFF

cheers

SnooP · Post by **SnooP** » Thu Aug 02, 2007 11:46 pm

dhanson865 wrote:oh, and if you are on 220v you already have insurance as most of the PSU calculators are based on 120v efficiency which is worse than 220v.

according to Wikipedia Quebec is on 220v, is that true for you?

its irrelevent, any decent psu does its full wattage rating at 110VAC as well as 240VAC, they are rated by what they can output (12V, 3.3V, 5V etc), not the what they draw from the wall outlet (110VAC or whatever).

In my experience, hence my opinion; pc calculators are based on providing potentional buyers with misinformation. Their single purpose (feel free to disagree) is to "assure" that potentional buyers are revealed of their hard earned money, more so, than providing them with accurate information, and that ticks me off. rant mode...oFF

Not my intention at all! The psu calc recommends a high 12V amps because the overclocking feature is excessive (based on the wrong rule of o/c power = stock power usage* (o/c freq / orig freq)* (o/c volts / orig volts)^2

The stock power of a Q6600 is ~100w, so once introducing 50% frequency increase + 16% volts increase (so another 34%) it ends up close to 200w just for cpu let alone 8800gts which is again has 30% freq increase. I know the formula aren't that accurate needs changing, but i don't have any data to go off to get it more accurate. Anyone with a killawatt or even better dc clamp meter care to do some tests?

Joe Public · Post by **Joe Public** » Fri Aug 03, 2007 4:49 am

Xbits labs did a test. A Q6700 @ 3.5 GHz, 8800GTX in SLI + some extras draw a max of 442W in burn mode. Meaning it draws a bit more than it would in ordinary use.

First page of this article.

dhanson865 · Post by **dhanson865** » Fri Aug 03, 2007 1:34 pm

SnooP wrote:
dhanson865 wrote:oh, and if you are on 220v you already have insurance as most of the PSU calculators are based on 120v efficiency which is worse than 220v.
its irrelevant, any decent psu does its full wattage rating at 110VAC as well as 240VAC, they are rated by what they can output (12V, 3.3V, 5V etc), not the what they draw from the wall outlet (110VAC or whatever).

Many good PSUs will do more than their full rated wattage before shutting down. That isn't the point.

It's nitpicking but in a borderline extreme environment the waste heat from that extra 2% of draw from the wall is radiated in the PSU. You are right that it really doesn't matter for the average situation but for an overclocker every little bit of waste or stress could effect power regulation (ripple, noise, droop)...

But if he is only going to draw 300W on a 600W PSU I doubt he is going to be too worried about the thermals or the power regulation.

SnooP · Post by **SnooP** » Fri Aug 03, 2007 5:21 pm

ahh ok, i was guessing you were one of those people that thinks a more efficient psu = can provide more power compared to equivalent wattage but not as efficient psu. my bad. From reviews i've read though it seems most quality psus will output their full wattage at 110VAC, whether its 25deg amb or 50deg. check out some of Jonnyguru reviews, rarely is there any noticable deviation between 'hot box' (ie higher ambient temps) and 'cold box' tests, occassionally a drop of 1% in efficiency or a .01V voltage change, nothing serious.

http://www.jonnyguru.com/review_details ... page_num=2
http://www.jonnyguru.com/review_details ... page_num=3

also looking at hardocp reviews, they compare how a psu does at 100V compared to 120V (interesting dropping efficiency as much as 240V vs 120V at high loads), again doesn't see to make a big difference to voltages etc.
http://enthusiast.hardocp.com/article.h ... VzaWFzdA==
http://www.hardocp.com/article.html?art ... VzaWFzdA==

but on the cheaper end you do have a point, this xclio passes at 120V but not 100V @ full load (overheated and shut off). I'm just not convinced yet the voltage regulation / ripple are a problem at 120VAC vs 240VAC, more just the extra heat from efficiency drop on a 'borderline' psu causing it to shut down.
http://www.hardocp.com/article.html?art ... h1c2lhc3Q=

mcoleg · Post by **mcoleg** » Fri Aug 03, 2007 5:44 pm

the cpu will draw 212W at full load (overclocking part of the calculator here is actually pretty good:

http://www.extreme.outervision.com/psuc ... orlite.jsp );

the card should be around 120W; it's overclocked but not overvolted, right?

(212+120)/12=27.7A for both cpu and vid. card. in theory should be enough amperage left for whatever else is needed.

one thing though. that quad cpu and the vid. card will dump an enormous amount of heat in the case when loaded. you really should measure the exhaust temperature on the psu. the s-12's are rated at 40C. if the exhaust is hotter, start de-rating; i think it's 2% per each 1C.

dhanson865 · Post by **dhanson865** » Fri Aug 03, 2007 6:57 pm

SnooP wrote:ahh ok, i was guessing you were one of those people that thinks a more efficient psu = can provide more power compared to equivalent wattage but not as efficient psu. my bad. From reviews i've read though it seems most quality psus will output their full wattage at 110VAC, whether its 25deg amb or 50deg. check out some of Jonnyguru reviews, rarely is there any noticable deviation between 'hot box' (ie higher ambient temps) and 'cold box' tests, occassionally a drop of 1% in efficiency or a .01V voltage change, nothing serious.

http://www.jonnyguru.com/review_details ... page_num=2
http://www.jonnyguru.com/review_details ... page_num=3

I looked at jonnyguru's site several times today before you mentioned it. He has not reviewed the old S12-500 or old S12-600. While he has reviewed the m12-700 it is a totally different design.

In addition his ripple tests are either hotbox or not, he doesn't do both. Going one step further his ripple tests are only done at 120V since he is in the US. So how do we know that ripple isn't affected by environmental factors?

SnooP · Post by **SnooP** » Fri Aug 03, 2007 11:09 pm

from hardocp review of s12 2 500w (yeah i know not the s12 model)

torture test @ 412w (80%)
http://www.hardocp.com/image.html?image ... 9sLmdpZg==

120V@ 384w (75%)
http://www.hardocp.com/image.html?image ... 83X2wuZ2lm

100V@ 384w (75%)
http://www.hardocp.com/image.html?image ... 83X2wuZ2lm

They all look similar values, all well under 120mV allowable.

for silencer 750w (again a seasonic build)
http://www.hardocp.com/image.html?image ... 83X2wuZ2lm
http://enthusiast.hardocp.com/image.htm ... 82X2wuZ2lm
http://enthusiast.hardocp.com/image.htm ... 9sLmdpZg==

Again I'm not seeing any appreciable drop in output quality (increased ripple) when environment conditions change, for any quality design psu (of which the s12 600w). s12 500w review:

http://www.xbitlabs.com/articles/other/ ... html#sect0

review mentions how the circuit is identical to corsair hx620. hardocp ripple tests of hx620:
http://www.hardocp.com/image.html?image ... 82X2wuZ2lm
http://www.hardocp.com/image.html?image ... 83X2wuZ2lm
http://www.hardocp.com/image.html?image ... 9sLmdpZg==

again all looking about the same amplitude.

mcoleg · Post by **mcoleg** » Fri Aug 03, 2007 11:19 pm

the silencer is a completely different animal... can't really compare it to regular seasonic models.

is S12-600 enough? PSU calculator says NO! (aanet.com)

is S12-600 enough? PSU calculator says NO! (aanet.com)

Re: is S12-600 enough? PSU calculator says NO! (aanet.com)