Slashdot Mirror
New 1 Kilowatt PSU - Too Much Power?
Steve from Hexus writes "While at the GC 2005 gaming convention in Leipzig, Germany, Hexus.net encountered a new 1kW PSU from Enermax, called the 'Galaxy'. At peak output it will use 1.4kW of mains power to provide a total of 66 amps across its various power rails. Who will actually have a need for this PSU, and when this amount of power is being consumed, shouldn't we be thinking about redundant power systems (or perhaps energy efficiency) instead?"
550W will handle a dual core Athlon, dual 7800 GTXs, 4 SATA drives, 2 optical drives, and a decent number of case fans, at peak load.
I guess, since you specified Intel, you might need an additional 100W, but thats still just 2/3s what this thing outputs.
Hunt your preferred prey at Aliens vs Predator MUD. Join the war at avpmud.com port 4000
It's no where near as exciting as it sounds. Right now I have a regular 3-axis mill with a rotary table as a fourth axis. The steppers draw up to 3A per coil, making 24A total. My fifth axis would be a second rotary table mounted 90 degrees on the first. It would make really nice chamfers without changing tools.
Do we need 1KW PSUs? no. I don't think so.
Generally, when I buy an overpowered PS, it's because I need a particular amp capacity on one of the rails. So I need a 550W PS for an Athlon 1.4Ghz box that probably draws 200W. This was because only the 550W model had the proper rating for +5DC.
"We returned the General to El Salvador, or maybe Guatemala, it's difficult to tell from 10,000 feet"
Err, if it's a polynomial of degree 5, and there are 5 numbers listed that are supposed to be roots, then of course there are no nonreal roots -- the sum of the number of real roots and the number of nonreal roots equals the degree of the polynomial.
So REAL nerds recognized that you don't need to manually check it!
not as much as you might think.
HDD Power consumption
IDE/SATA drives only draw about 7-13W Idle/read&write, 15K SCSI drives a bit over 20W read&write.
Spin up might be a problem, but I'd assume you'd want to use cards that supported staggered sinpup on a setup that large.
So, yes 16 HDDs can pull quite a bit of power, about 300W for top end SCSI solutions. Though you wouldn't be thought of as particularly bright if you entrust a setup like that to a basic quality desktop PSU. And the quality of supplies you'd be using with a high end storage array like that (ie something in the N+1 redundant Zippy line) have been availible at well over 1000W for a while.
I think a 1000W PSU in a standard EAXT setup is massive overkill. I really have a hard time thinking of a workstation / stand alone server setup that would be too much for quality 500-600W PSUs to handle right now.
Anandtech reviewed a 4 CPU dual core Opteron setup from SUN while back, it only drew about 600W.
Just because a device is rated for a load does not mean it will use that load. Because you have a 200A pannel does that mean that you use 200A all the time? No, you doen't even come close unless your elecric range, water heater, hot tub, and resistive heat are all going at the same time. The real reason to have a 1000W power supply is to get clean and stable power if you use 200W. For our purposes a power supply has a part (rectifier) that chops off the negative voltage leaving you with a bunch of pulses of voltage. There is a second component (capacitor) that stores energy at the peaks and delivers it during the valleys. There is a third component (added to the rectifier makes it a bridge rectifier) that will turn the negitive voltage into positive voltage that fits nicely in the spaces inbetween the existing positive pulses. This doubles you efficiency by giving you twice the power at the output from the same input. Everything is great untill you put a load on the power supply, then you start to actually use the power out of the capacitor, this leads to a dip in the voltage called ripple. The higher the max wattage for a power supply the more power you can use before the ripple becomes a problem. Ripple in a processor is bad, this is why you will notice capacitors all over your motherboard and on some chip packages. These capacitors help smooth out the ripple.
If you run a 200W load on a 250W power supply then you will have a great deal of ripple. If you run a 200W load on a 1kW power supply then you will have much less ripple. ripple == fluctuating voltage == unstable pc
I have a Pentium D 8200, 2 GB of DDR2, Dual HDs (one SATA one IDE) and dual DVD burners, and a Radeon X800. Also, add a couple of firewire and USB devices. All of this runs off of a good 350W power supply with room to spare. If a 6800 (which uses more power than the 7800) uses about 135W, SLI would use 270W. That means you really only need a 500W supply. I think the idea above is right. By having an obsenely overkill flagship model, you not only make a couple of bucks off rich, poorly endowed kids who want a 1000W supply, but you get to brag that your company makes the most powerful power supply comercially available.
While a power supply with a higher rating may have a cleaner output than a one with a lower rating, this is not necessarily the case. Without other measurements there simply isn't enough information to say.
Computer power supplies use a switching circuit to generate the output voltages. Not the simply, rectifier and filter system you described. They do use a recifier and filter, but this is only at the initial stage and ripple at that point doesn't significantly affect the output.
In a rectifier and filter supply, using a full wave rectifier does not double efficiency. It does allow you to get by with less filtering and a reduced peak input current.
Creating a full wave bridge rectifier doesn't involve adding 1 component to a half wave rectifier. In fact, you need 3 more components. A half wave rectifier is just a single diode. A full wave rectifier consists of 4 diodes.
Switching supplies do have ripple, and it is effected by the load on the supply. Some of the other factors affecting the amount of ripple are switching frequency, inductance of the switched coil, capacitance and resistance of the output capacitor, and input voltage.
Capacitors placed near ICs on the motherboard are for filtering out high frequency noise than can be induced on the circuit board traces. These capacitors are not normally the right values for filtering out 60 or 120 hz noise from a rectifier. If it weren't for induced noise on the traces, you could simply place one capacitor at (or in, as there already is) the power supply instead of one at each IC.
Your computer should run perfectly stable on any supply up to its rated output power and current. If a supply outputs so much noise that your computer is unstable before you reach the rated output, then it is almost certainly faulty or rated in such a way as to scam consumers.
With the same load, a higher rated supply might run cooler or with a cleaner output; but it depends on many different factors. You need to know things like output noise and efficiency. Output power alone is not enough information.
----
All of whose base are belong to the what-now?
All power supplies have a "sweet spot" from an efficiency standpoint which is normally 75% of rated power. That would imply that this power supply is most efficient around 750kW. At 350kWs I would guess this supply is only 65-75% efficent. Perhaps it is designed to be efficient across a broad range of output.
There appears to be a number of homebrew designs here. One other thing that I noticed from my hour of looking at the subject yesterday is that you need a dedicated computer to control the system. There appears to be Penguin CNC, which seems to do the job, using Realtime Linux.
Jumpstart the tartan drive.