Using an old dual CPU server as a gaming PC?

John Doe wrote:
If you know offhand, Paul, should the faceplate the grounded to the
case?

Or are all the other ground connections good enough?

Thanks.

The faceplace is already grounded, and the ground
connection would extend to GND on the edge card.

If you look at the mounting holes for the faceplate
screws, you'll see surface copper, as well as
via stitching to an internal layer. And that
should be tied into circuit ground. If you
ohm from faceplate to chassis ground right now,
there should be a zero ohm path.

*******

Why make the faceplate of conductive metal ?

1) ESD - you want the faceplate metal to touch the
chassis metal. The idea behind this, is so an ESD
discharge flows into chassis metal, instead of through
any other conductors. There are two strategies for faceplates,
and the alternative strategy is make the entire outside of the
machine out of plastic, so ESD can never discharge through
the plastic. If the faceplate doesn't touch, the GND
connection still provides a "Plan B" discharge path.

Why ground the faceplate ?

2) RFI - the faceplate could be part of a radio frequency
interference containment strategy. But computer cases
are not normally considered "RF tight". When we made
stuff at work which is "RF tight", we used beryllium copper
spring fingers, where faceplates touch chassis metal,
and we might use honeycomb air filters over the fans
(which form a microwave cutoff filter). A computer case
on the other hand, simply isn't "tight" enough to be
considered as a real RFI containment solution. However,
having "shield ground" around cabling, may help, and the
connector ground joined to faceplate ground is intended to
keep the faceplate as part of a metal shielding container.

If the faceplate doesn't touch the chassis, the circuit
could end up more ESD sensitive. Try not to discharge
yourself through that faceplate metal, and touch the
chassis first.

*******

And before you game on it, what are you going to
do with the cooling ? The thing is cooled by a
high velocity, passive heatsink strategy. The
fans are 40mm and deep, to give high static
pressure. When they use a wall of fans like that,
the wall has to be consistent all the way across,
so air doesn't go through a gap in the wall
and back out the front of the rack. Like, if
you removed one of the wall of fans, you can't leave
a gap, and have to put a "filler plate" in its place.

On our systems at work, we used "louvers" in the
fan bay. If you removed a fan while the equipment
was in service, the louvers prevented backward
airflow in the wall-of-fans. That was our solution.

Paul

John Doe wrote:
Posted the Passmark results. Probably not a great test, but it's
something. I might do some 3DMark or PCMark too.

Maximum short-term wattage (for the whole system, measured at the plug)
was 230 W during that test. Peak was about 240 W. Go figure.

You should be testing for Throttling.

It's possible your CPU is hot to begin with,
and when you start the test, it throttles.

You could at least try the BIOS screen,
see if it has a hardware monitor, and see
what kind of temps the CPU hits in the BIOS.

Paul

Paul wrote:

John Doe wrote:

Posted the Passmark results. Probably not a great test, but it's
something. I might do some 3DMark or PCMark too.

Maximum short-term wattage (for the whole system, measured at the
plug) was 230 W during that test. Peak was about 240 W. Go figure.

You should be testing for Throttling.

It's possible your CPU is hot to begin with, and when you start the
test, it throttles.

You could at least try the BIOS screen, see if it has a hardware
monitor, and see what kind of temps the CPU hits in the BIOS.

All I'm saying is that the whole system uses a maximum of 230 W even
when doing a benchmark, even with the GTX 960 installed.

The CPUs are 40 W package chips (I think that means the CPUs can
dissipate a maximum of 40 W each), but even with CPU upgrades the whole
system is probably never going over 350 W total.

That's a good thing.

The BIOS Northbridge PCI express lane setting was changed from x8 to x16.
That increased the passmark 3-D video score by 23%.

Hopefully I am not missing any other important non-overclocking related
settings.

This brings up 27 relevant comparison results.

("Display Baseline" "Dual CPU" -wiktionary -wikipedia)

John Doe wrote:
The BIOS Northbridge PCI express lane setting was changed from x8 to x16.
That increased the passmark 3-D video score by 23%.

Hopefully I am not missing any other important non-overclocking related
settings.

This brings up 27 relevant comparison results.

("Display Baseline" "Dual CPU" -wiktionary -wikipedia)

So how does it compare to your single socket
desktop, running the same video card ?

Paul

Paul wrote in :

John Doe wrote:

The BIOS Northbridge PCI express lane setting was changed from x8 to
x16. That increased the passmark 3-D video score by 23%.

Hopefully I am not missing any other important non-overclocking
related settings.

This brings up 27 relevant comparison results.

("Display Baseline" "Dual CPU" -wiktionary -wikipedia)

So how does it compare to your single socket desktop, running the same
video card ?

Sorry, but for some strange reason, the 3-D test wouldn't/didn't run on
my LGA 775 Q9550 system. I should have put some effort into making it
work before moving the GTX 960 to the server.

The CPU score is twice as high, going from single 2.83 GHz to duel 2.13
GHz.

CPU-Z says triple channel memory, to me that suggests something like six
channels total.

At the moment, I'm moving the SSD to the server.

Will separate the counter rotating fan pairs into single fans. Then use
hot melt glue to secure them into place and help dampen the vibration.
Preliminarily speaking... Seems one fan is much quieter than two. I
don't need massive air flow, so that should work. And that doubles the
number of fans.

Speedfan appears unable to control my system fans. Some way (like the
above) to quiet them more than the lowest BIOS setting was needed.
Connecting them to the 5 V supply or making my own voltage regulator
would have been much more work.

Got a USB Wi-Fi antenna/adapter today, TL-WN722N V1, works in Windows
8.1 (not sure about 10).

And that just about makes it ready to use. A little short on expansion
slots, but I hardly use expansion slots. And it has eight USB
connections.

I haven't really looked, but apparently the GTX 960 can output audio, so
that will be the audio source (there is no built-in audio). I suppose
that means using the HDMI connector. Need to grab a cable off of eBay.

The drives plug into the backplane. Even SSDs slide neatly ino the power
and data connector sockets.





http://www.ebay.com/itm/371543511693?

http://www.ebay.com/itm/301325754598?