Personal computing discussed

Moderators: renee, SpotTheCat, Nemesis

 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

PSU/Wattage Needs, Explanation & Examples

Sun Feb 05, 2006 1:13 am

I see so often people talking about wattages and amps and PSUs and needing 5404230420 watt PSUs to supply their computer with enough power. This isn't meant to be comprehensive (including PSU efficiency, design factors, brand-names, etc, would take way too long, but I thought I could shed a little light to debunking some of the common perceptions.

Let's start with something simple.

To convert Amps to Watts:

Watts = Amps x Volts


This is why--when assessing which PSU to get-- it's important to look at the 3.3v, 5.5v, and 12v "Amps". The 12v rail is arguably the most important since your CPU, Video card, hard drives, optical drives, etc, draw power from it.

Here are some max-power draws expressed as Watts:

AMD X2 4200+: 89 watts MAX
Dual 7800GT in SLI: 112 watts MAX
Two Raptor 74GB in RAID 0: 16 watts max
Extra Seagate 7200.9: 12 watts max
TOTAL: 229 watts


Obviously, this doesn't include some other minor things, but it is more than ample to give you an idea of the actual wattages involved here. Also, let's keep in mind this is about maximum PEAK LOAD on every component. This system would draw significantly less during normal non-gaming use.

Based on the formula I stated above, that means an AMD X2 4200+ expressed as "MAX AMPS REQUIRED" on the 12v rail would be:

AMD X2 IN AMPS|
89 watts / 12 volts = 7.41 amps

Single Geforce 7800GT IN AMPS|
56 watts / 12 volts = 4.66 amps


You can factor in 2 amps for the three hard drives. 1 amp for cooling. These are 12v rail items as well.

This puts our test system at ~20 amps (rounding up).

Now let's go ahead and assume 10% overhead. Just to build in a ton of extra breating room. That puts us at ~22 Amps or ~260 Watts for our main hardware. Need another hard drive? Add an amp or 10 watts. Going to add another CPU? 90 watts or 8 AMPS.

So, let's back track a bit and think about this. If our test system has nice overhead @ 22 amps, with 3 hard drives, an AMD X2 and 7800GT SLI setup.... what PSU can we get?

http://www.newegg.com/Product/Product.a ... 6817104154
OCZ Powerstream 450 watt: 12v RAIL? 26 AMPS (that's enough)
Cost? $90

http://www.newegg.com/Product/Product.a ... 6817192002
Atrix 500 watt: 12v RAIL? 15 AMPS (no way... nowhere near enough).
Cost? $29.99

So, yeah, Atrix has a "500 watt" power supply, and it's cheap... but it's completely insufficient for our system. There is no way the configuration would run properly on that "500 watt" PSU. The "450 watt" OCZ powerstream, however--SOMEWHAT against conventional wisdom that you need 500 watt or 600 watt PSU to do SLI--would readily run our test configuration.

==============================
Here are some other common wattages:

Radeon X850XT = 45 watts @ idle / 68 watts @ peak load
Geforce 7800GT = 39 watts @ idle / 56 watts @ peak load
Radeon X1900XTX = 49 watts @ idle / 120 watts @ peak load
Geforce 7800GTX = 53 watts @ idle / 94 watts @ peak load.

And AMD X2 Manchester (3800+, 4200+) run @ MAX 90 watts. The Toledo cores (4800+, 4400+) run @ MAX 110 watts.

A Raptor 74GB draws 8 watts.
A Seagate 7200.9 draws 12 watts.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Sun Feb 05, 2006 1:31 am

I'd toss in another 10-20W for the motherboard and RAM... but yeah, I'd say you're pretty much on target here. Your 10% "fudge factor" should cover any inefficiencies in the switching regulators on the mobo.

I think the myth that you need a monster PSU resulted from a couple of things. Firstly, many cheap PSUs have specs which are optimistic (to put it kindly). Secondly, older PSUs put more of an emphasis on the +5V and +3.3V rails; but modern mobos draw CPU power from the +12V rail. So when mixing and matching mobos and PSUs from different generations, it was quite easy to run into problems even if you had plenty of total wattage.
Nostalgia isn't what it used to be.
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Sun Feb 05, 2006 1:39 am

Agreed, take a look at that "500 watt" Atrix.

The 5v has 50 ams, 3.3v has 30 amps... that's 350 watts, 70% of its rated power. It's really a classic PSU scam. Newegg shouldn't sell them. I guess it does have awesome colored lights though!!!111

I think the other thing is that people associate higher numbers (in computer-stuff) to be better / newer. Automatically, 500Watts is better than 450Watts. Just like... a P4 2.8Ghz must be better than an AMD 64 2.2Ghz, right? hehe
Last edited by computron9000 on Sun Feb 05, 2006 1:42 am, edited 1 time in total.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Sun Feb 05, 2006 1:40 am

It's all about headroom, not just for expansion, but for device startup as well. The HDs and optical drives are mechanical devices and will take more wattage to power up when you turn it on. Not sure how many more watts you need to tag on in order to make the formula work, but it is true that we are talking quality not quantity here. Also, USB powered devices can also draw a couple of amps too.

Let's hope staggered spinup will be implemented soon on more HDs so we don't have that huge power draw.
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Sun Feb 05, 2006 1:43 am

AFAIK hard drive spin-up can't require more than the PEAK LOAD values I supplied. A drive can't work harder than it works at its hardest? hehe All values supplied are PEAK / MAX LOAD values. If they were idle values, assume nearly 50% of what I stated overall. I guess it depends on whether the watts to maintain spin and move the head are greater than power needs to get up to final speed?
Last edited by computron9000 on Sun Feb 05, 2006 1:46 am, edited 1 time in total.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Sun Feb 05, 2006 1:46 am

Also needs to be considered are the latest fashion trend, the bling bling LED lights, flouresent tubes and such. A few more watts here and there, but it shouldn't detract from the main idea that you don't need 1000W.
 
Usacomp2k3
Gerbil God
Posts: 23043
Joined: Thu Apr 01, 2004 4:53 pm
Location: Orlando, FL
Contact:

Sun Feb 05, 2006 1:48 am

What about my watercooling?
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Sun Feb 05, 2006 1:50 am

Usacomp2k3 wrote:
What about my watercooling?

It depends on where are you powering that from of course. Your minigate is off the main PSU, so it is something else to be considered. But you already know that, right? :wink:
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Sun Feb 05, 2006 1:53 am

Flying Fox wrote:
Also needs to be considered are the latest fashion trend, the bling bling LED lights, flouresent tubes and such. A few more watts here and there, but it shouldn't detract from the main idea that you don't need 1000W.


http://www.dansdata.com/bluelights.htm

This guy has a whole bunch. Seems like they aren't too bad. Even the big fancy tube blue one is only pulling 4 watts. <2 watts seems commong for the bigger LEDs.
 
Usacomp2k3
Gerbil God
Posts: 23043
Joined: Thu Apr 01, 2004 4:53 pm
Location: Orlando, FL
Contact:

Sun Feb 05, 2006 1:53 am

Flying Fox wrote:
Usacomp2k3 wrote:
What about my watercooling?

It depends on where are you powering that from of course. Your minigate is off the main PSU, so it is something else to be considered. But you already know that, right? :wink:

Of course.
Lets see what other toys I have in here.
2xCCFL lights, 6x80mm LED fans, 2x120mm fans, two optical drives ;-)
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Sun Feb 05, 2006 2:04 am

Flying Fox wrote:
It's all about headroom, not just for expansion, but for device startup as well. The HDs and optical drives are mechanical devices and will take more wattage to power up when you turn it on. Not sure how many more watts you need to tag on in order to make the formula work, but it is true that we are talking quality not quantity here. Also, USB powered devices can also draw a couple of amps too.

Let's hope staggered spinup will be implemented soon on more HDs so we don't have that huge power draw.


The other thing to think about here is that at startup the video card(s) won't be at full-tilt, which comprise about 50% of the overall Amps / Wattage requirements in most SLI gaming rigs DURING PEAK LOADS (which is what we're calculating by). In a single GPU rig, this would easily account for any additional (however negligible) spin-up for fans / opticals / etc. I've found spin-up wattages for a few notebook drives and they were all <9 watts. Can't find any for modern drives like a Raptor, but still this has to be less than, say, 1-2 AMPS total for 3-4 drives. In a single-drive system, the effect would be moot.

The thing about peak-load estimations is that it's somewhat unusual for a system to have EVERY SINGLE COMPONENT at peak load. Even then, I don't know how far beyond spec / estimates you'd actually have to go to see damaging voltage fluctations on your 12v / other rails. That's another reason people can get away sometimes with crappy PSUs (although I'm sure alot of them have unexplainable problems / crashes).

Image I imagine wattage usage looking like that, from boot-up, through game-playing, and back to Windows (VERY ROUGH EXAMPLE). But, when do they all three hit 100% exactly, and how close to the come? Keep in mind, just calculating for PEAK MAX LOAD is already building in a safety margin. 10% extra is a huge overhead. Anything beyond that is for expansion purposes.

Like, say, you're right on your AMP/Watt limits for your hardware with your PSU. And once every 2 months your computer crashes without any explainable reason. Could it be your PSU if the overhead for MAX PEAK LOAD is very tight? Of course. Does it otherwise work flawlessly otherwise? Yes... obviously, what happens is EVERY PIECE OF YOUR HARDWARE isn't all working @ 100% for sufficient durations to freak out your PSU.

For example, let's assume 2 of 3 hard drives are idle and only drawing 3 watts. Then let's assume the processor is only drawing 60 watts. Then let's assume the optical drive is idle. Then let's assume some fans aren't spinning @ 100%.

To top it off, if you have an ATX2 certified PSU (dual-12v Rail), "spike load" from mechanical drives goes on the 12v(2) rail while the motherboard is powered by the 12v(1) rail. Keeps the power cleaner where it matters.
---

120mm fans can run at a lot of different wattages. I'm guessing anywhere from 0 watts to 15 watts (possibly higher for completely insane RPM ones). I'd need some numbers to back that up though. I did find this crazy guy that built box-fans out of tons of 120mm fans, who lists wattages for his stuff. And here's a quote:

MFN: DF128025SL-4 23.12 CFM, 27.07 dBA, sleeve bearing 20K hour MTBF@25°C, 18 inch power leads, 4pin power connector, includes mounting screws
Unit Price: $0.85
Uses 1.68 watts each. I normally use 80mm fans because I have them on hand for computers, also they are cheaper each and use less power each. I think the 120mm I looked at was around 5 watts.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Sun Feb 05, 2006 12:52 pm

computron9000 wrote:
AFAIK hard drive spin-up can't require more than the PEAK LOAD values I supplied.

I believe the values you quoted are actually for peak draw during normal operation (not including spinup).

I'm having a hard time finding peak current draw info for the Raptor. But the Seagate 7200.9 is speced for a peak draw of 2.8 amps from the +12V rail during spinup, which is nearly 34W!

A drive can't work harder than it works at its hardest? hehe All values supplied are PEAK / MAX LOAD values. If they were idle values, assume nearly 50% of what I stated overall. I guess it depends on whether the watts to maintain spin and move the head are greater than power needs to get up to final speed?

At least for the Seagate, getting up to speed apparently requires nearly 3x the power of normal operation.
Nostalgia isn't what it used to be.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Sun Feb 05, 2006 1:35 pm

just brew it! wrote:
I'm having a hard time finding peak current draw info for the Raptor. But the Seagate 7200.9 is speced for a peak draw of 2.8 amps from the +12V rail during spinup, which is nearly 34W!

:o:o:o

We sorely need staggered spinups, like yesterday.
 
Captain Ned
Global Moderator
Posts: 28704
Joined: Wed Jan 16, 2002 7:00 pm
Location: Vermont, USA

Sun Feb 05, 2006 1:37 pm

Flying Fox wrote:
We sorely need staggered spinups, like yesterday.

Has the SATA specification allowed for this like SCSI has done for eons?
What we have today is way too much pluribus and not enough unum.
 
JustAnEngineer
Gerbil God
Posts: 19673
Joined: Sat Jan 26, 2002 7:00 pm
Location: The Heart of Dixie

Sun Feb 05, 2006 1:40 pm

Yes, staggered startup is supported with modern drives.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Sun Feb 05, 2006 1:41 pm

Captain Ned wrote:
Flying Fox wrote:
We sorely need staggered spinups, like yesterday.

Has the SATA specification allowed for this like SCSI has done for eons?

It's been optional in the SATA spec IIRC. Which drive implements it is kind of hard to know, since their specs are often unclear. They only want to show STR and access time these days. :(
 
CampinCarl
Graphmaster Gerbil
Posts: 1363
Joined: Mon Jul 04, 2005 9:53 pm

Sun Feb 05, 2006 2:11 pm

So, pardon the newb question, but:

What exactly are the 3.3v and 5v rails for these days?
Gigabyte AB350M Gaming-3 | R7 1700X | 2x8 GB Corsair Vengeance DDR4-3200 (@DDR4-2933)| Samsung 960 Evo 1TB SSD | Gigabyte GTX1080 | Win 10 Pro x86-64
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Sun Feb 05, 2006 2:14 pm

CampinCarl wrote:
So, pardon the newb question, but:

What exactly are the 3.3v and 5v rails for these days?

Chipset, DRAM, PCI cards... hard drives and DVD drives still use the +5V to run the electronics on the drive (it is just the spindle motor that uses +12V).
Nostalgia isn't what it used to be.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Sun Feb 05, 2006 4:48 pm

To further the discussion along, let's take the ThermalTake TR2 420W for example. The OCZ example in the FP has only 1 12V rail of 26A. Dual 12V rails seem to be the rage these days, and the TT is one such unit. However, if we look closer we find 12V1 is doing 10A while 12V2 is doing 15A. If the X2 is on rail 1 coupled with a hard drive (or 2), will that be too close for comfort?

Of course, this TT is not to be used for SLI.
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Mon Feb 06, 2006 1:19 am

However, if we look closer we find 12V1 is doing 10A while 12V2 is doing 15A. If the X2 is on rail 1 coupled with a hard drive (or 2), will that be too close for comfort?


Nope. Basically, all that matters is the total amperage, not the specified amps for individual 12v rails. A quote from Silent PC Review (http://www.silentpcreview.com):

What the above means is that you don't need to worry about imbalances in power draw on the 12V lines �as long as no single line is asked to deliver more than 20A. PSU makers seem to mark each line for max current on a purely arbitrary basis, probably more for marketing reasons than any other. A PSU rated for 32A max on the 12V lines can be labelled many different ways:

* 12V1: 18A, 12V2: 14A
* 12V1: 17A, 12V2: 15A
* 12V1: 16A, 12V2: 16A
* 12V1: 15A, 12V2: 17A
* 12V1: 14A, 12V2: 18A

It could be marked 20A + 12A, but being a cautious bunch, the engineers will probably not specify more than 18A on any one line. This gives 2A headroom to allow some room for error for the current limiting circuit.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Mon Feb 06, 2006 1:22 am

What's the reason of going 2 rails instead of 1 these days? Too much amperage will burn the wires?
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Mon Feb 06, 2006 1:37 am

Safety requirements. Intel's ATX12V PSU Design Guide v2 reccomends no more than 18A @ 12v. They specify that PSU designers should keep the current under 20A / 12v rail.

Intel ATX12V v2.2 guide:

-- In cases where expected current requirements is greater than 18A a second 12 V rail should be made available. (18x12=216VA, to allow peak headroom to 240VA)
-- The 12V rail on the 2 x 2 power connector should be a separate current limited output to meet the requirements of UL and EN 60950.
-- 12V1DC and 12V2DC should have separate current limit circuits to meet 240VA safety requirements.
-- 12V2DC supports processor power requirements and must have a separate current limit and provide 16.5A peak current for 10 ms; minimum voltage during peak is > 11.0 VDC


I've also read that Intel has since revoked this bit and some power supplies made to the specification either weren't tested or didn't pass for this when tested. There may be exotic cases where a single-rail works where a split-rail setup doesn't (where both have the same total amps). But I haven't heard of anyone having that problem that has bought a good-quality dual-rail PSU.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Mon Feb 06, 2006 11:42 am

Found an interesting tidbit on SPCR. It recommended a 30% headroom for peaks. So we have to adjust our calculations accordingly.

---
In other news...

[Edit: removed stupid math that does not make any sense]
Last edited by Flying Fox on Mon Feb 06, 2006 11:48 pm, edited 1 time in total.
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Mon Feb 06, 2006 7:29 pm

I'm not sure efficiency should be calculated into there... I'm heading out to the store, but when I get back I'll take a closer look. AFAIK the DC Output of a PSU is going to be its "rated output" -- the efficiency is in the conversion of the AC power of the wall. For example, if you tested it and your your power supply was getting 259 AC watts from the wall, and it was 80% efficient, then your computer is using 207 DC watts. What this means is efficiency is in the conversion process, not involved in peak load. This also means it contributes to heat and, usually, noise (as well as wasted power). For example, to max-out a 600 watt PSU, you would be drawing 750 watts from the wall.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Mon Feb 06, 2006 11:47 pm

Oops, I think you are right. I think I may have gotten the rated wattage (on the label) and the power draw at the wall mixed up. :oops: Still, the Seasonic is clearly a better power supply, with the combined 12V rails supplying the same amps.
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Tue Feb 07, 2006 12:04 am

Yes, I agree. Let's look at the money saved by having a more efficient PSU. To summarize the below: Going with an 85% PSU vs 70% PSU on a system that draws 200 watts, 24x7, results in a savings of $120 over 3 years (ignoring extra cost for the better PSU).

Say your system draws 200 watts.

85% efficient PSU = 235.3 watts from the wall (200 / .85)
70% efficient PSU = 285.7 watts from the wall (200 / .70)

This means the 85% is wasting 35.3 watts as heat, vs 85.7 watts. (235.3 - 200 = 35.3 | 285.7 - 200 = 85.7)

In other words, the 85% PSU is almost 2.5x better on electricity savings / heat output. (85.7 watts / 35.3 watts = ~2.5)


That's ~1.2kWh / day more for the 70% efficiency vs 85% efficiency. Or 438kWh more / year.

[ (85 watts - 35 watts) / 1000 ] * 24 hours * 365 days = 438kWh

Now let's run that out over a year. DOE energy estimates the national average in the USA to be $0.09 / kWh.

So, by having the 85% efficient PSU in this scenario (vs the 70%) you should see savings of ~$40 / year, provided you leave your system on 24x7. Adjusted even by 20-30% to account for times when you might shut it off, that's still ~$30 US / year saved.

So, here's the chart:

In a 200-watt system, here's the savings per year when using an 85% PSU vs a 70% PSU (assuming 24 hours / day use, less 10% for when computer is off). Let's also assume that the 85% efficient PSU is $20 more to purchase:

1st Year Savings: $16
2nd Year Savings: $36
3rd Year Savings: $36
Total savings over 3 years: $88

Assuming that the 85% PSU was $60 and the 70% was $40, that means over 3 years the 85% PSU as an alternatively actually paid completely for itself and MADE $20 for the person that bought it.

So in a sense, you buy an 85% efficient PSU vs a 70% and, through your electricty bill, the PSU is paid for in full, plus they put an extra $20 in your pocket, over 3 years.
Last edited by computron9000 on Tue Feb 07, 2006 12:30 am, edited 1 time in total.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Tue Feb 07, 2006 12:29 am

Funny thing about applying this knowledge. I went to my friendly neighbourhood store to get some quotes from the guy and to check out cases, and he was showing me one of his working builds for a customer. It was an AOpen H600B, with a 400W PSU. The customer wanted him to change that with a 500W unit. I checked out the label and it only had one 12V rail of 17A! :o The included AOpen 400W featured 12V1@10A and 12V2@17A(may not be correct, but the 2 rails combined to be over 20A)! I was laughing so hard and just shaking my head. The guy told me that the customer was putting a P4 in it. I just said "good luck" (apparently it may work it's supposed to be a low GHz P4).
 
computron9000
Minister of Gerbil Affairs
Topic Author
Posts: 2420
Joined: Sun Oct 16, 2005 7:21 pm

Tue Feb 07, 2006 12:36 am

Hah, yeah, that's too bad.

I was at a Best Buy (for those not in the US near one, a giant computer/electronics/camcorder/digitalcamera/hdtv store) one time in the computer isle.

A salesman was explaining to an older couple something about UPS backups. They had like a $50 one, a $100 one, and a $200 one. The couple was asking which one they needed. He said, "Well, if you get the cheaper 1.5Ghz processor in your computer, you should stick with the $50 one, but if you upgrade to the 2.0Ghz processor like we talked about, you're definitely going to want the $200 version." They nodded in agreement, and understanding, and put a $200 UPS in the cart for their $600 computer... I guess... so it can stay on for 2-5 extra minutes when the power goes out while they're trying to get their internet explorer to "Connect to AOL" or something goofy. I'm sure the difference in wattages between the two processors were no more than 20-30 watts, tops. Just classic "You need this if you get that" computer-hardware sales tactics.
 
Flying Fox
Gerbil God
Posts: 25690
Joined: Mon May 24, 2004 2:19 am
Contact:

Tue Feb 07, 2006 12:44 am

Well, I guess the couple did the right thing for the wrong reason. Who in their right mind get a UPS for $50? It's not going to be a real UPS, is it? :wink:

The bigger question is, did they really want an UPS, or just a better surge protector?
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Tue Feb 07, 2006 12:45 am

The really sad thing is, the BB salesperson was probably working from a memorized script, and had absolutely no clue about what he was selling.

IMO, BB is good for only one thing: They tend to be relatively hassle-free when it comes to returns. Their prices are mediocre, and their salespeople are clueless; I generally don't buy computer stuff there any more.
Nostalgia isn't what it used to be.

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On