Anonymous wrote:
deimos wrote:
a 475 Watt power supply. I have room in the
Dell XPS 9000 case for a full-length, two slot card. I'm not
adverse to installing a bigger power supply if the Dell uses
standard connectors. Does anyone have any recommendations? Thanks for
any advice you may give.

Tom Lake
A GTX250 will both fit easily and doesn't consume more than a decent
400/450W can put out. Plus it has good all around performance and is
readily available.

Sorry, but I can't believe that's true. I own a 260 and I don't think
that a 250 needs that much less than a 260 and the box of my card states
that the absolute minimum required by NVidia's specs is 550W, while the
card maker (XFX in my case) recommends 630W or greater. 680W or greater
are required for SLI.

While the card may work for a while with less than that, if you start
playing games (or use other applications that make the card draw more
power than when displaying the desktop) the safety fuse of your PSU may
immediately cut the power at any time. I speak from experience.

For reference a 250 is a slightly faster rebrand of the 9800GTX+, which
was an overclocked process revision (55nm) of the previous 9800GT/GTX,
which itself was the offspring of the 8800GT (G92b)... confusing yes.

If you feel like upgrading to a decent PSU, like a Corsair 650W or so,
you could easily run a GTX260 or 275, both of which are approx the same
length as a 250 full length reference card. A good quality PSU of 550W
or more is SLI capable as well, so basically it can run any single card
config.

550 is hardly enough for a single 260 card. I recommend a good 700W PSU.

Bye!

You can get measured values for just the video card, suitable for
doing a power budget. The 55nm GTX 260 is 112W at "3D max", and this
is measured with multimeters and current shunts by Xbitlabs. That power
will be coming from 12V1, so should be included as part of the 12V1 loading.
Add 0.6A for hard drive, 1.5A for CDROM, 0.5A for fan headers, and ~9A for
the GTX 260 55nm, gives 12.6 amps.

http://www.xbitlabs.com/articles/vid...m_5.html#sect0

The total power supply capacity, doesn't have to be real high. It's just
a matter of adding the power for all the individual components. Allocate
50W for the motherboard chipset and the DIMMs, where the DIMMs can
be on the order of 2W each now. (Kingston provides numbers in their
downloadable datasheets for memory, when you need numbers.)

The end result is, you should not need a 700W power supply.

Naturally, any person buying a power supply, can pad or margin it all
they want. If the load is 300W max by calculation, and you buy a 700W supply,
there is no harm done. You've just spent more than was absolutely necessary.
A little margin is a good idea, but more than doubling it isn't necessary.

I could equally say, without doing any math at all, "get yourself a good
1200W supply", with a voice of authority. But instead, you can use math
and work out exactly what you need. Then buy the power supply that
provides a little margin and has a good reputation. That is worth more
than having paid for an extra 5 pounds of unused power supply
sitting inside your computer case.

*******

One thing you should understand, is when a manufacturer gives power number,
they assume the most power hungry processor has been installed in the
system. For example, they might assume the computer has a 130W processor.
Well, I use a 65W processor in mine, and the measured power consumption
of my processor flat out, is 36W (measured with a clamp-on DC ammeter while
Task Manager has the CPU graphs at 100%). You can see, how the manufacturer
making an assumption for my benefit, has just backfired. Now I'm buying
100W more of power supply, than I really need.

Also, there are a number of power supply estimator web pages available
on the Internet. In all the cases I've evaluated, they're giving the
wrong answer. A typical bad web site, works out a figure double what
it should be. The gullible reader of the web site, then takes that
figure and doubles it again, and goes shopping. How silly is that ?

*******

To give another great story, I can tell you about something that
happened at work. One of the staff in our lab, got a breathless
phone call from shipping and receiving. "Come quick, there's this
box down here, and you have to come and get it right away. You
can't leave it here". The guy sounds scared. The warehouse is about
a 10 minute drive from work. The person receiving the call, drives
over for a look, because the person on the other end of the phone
sounded worried, but didn't provide any details. Normally, they'd
just deliver it to us.

At the warehouse, in the middle of the floor, was a box about 3'x3'x3'.
It has a large radioactive sticker on it. The shipping and receiving
staff were so scared of it, they cleared a large area around it, removing
anything even remotely close. The scene looks like ET just landed,
in the middle of the floor.

When the box was opened, there was a smaller box inside. It had
a slightly smaller radioactive sticker on it.

When that box was opened, there was yet a smaller box inside. It
was plastered with radioactive stickers as well.

When we finally got all the way into the box, there was a
small radioisotope vial, the kind your teacher in high school
might have used for an introduction to radioactivity. It might
have had a microcurie of radioactivity, certainly nothing to strap
on a lead apron for.

What had happened, is each person in the supply chain, repackaged the
shipment, putting a larger box around the outside "just to be safe".
Until the box was huge, and the large radioactive sticker on the box,
scared the **** out of people. I'm sure the people in shipping and
receiving, who know nothing about radioactivity, thought they were
going to die.

The moral of the story is, if everyone pads the budget a little
bit, the end result is blown out of proportion. And someone eventually
has to look inside the box, for the truth. And math can help you
find that truth.

Paul