• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Ghetto Water Works.

Q

Qinsp

2[H]4U
Joined
Jan 7, 2011
Messages
2,154
OK, second effort. First did not have adequate radiator area. This is about 47 sq in of radiator per CPU. SWTX Tyan S8812 mobo (4P G34) is in a Rosewill Blackhawk case with a subplate to raise mobo .125". Chinese water blocks are held down with homemade spring clamps. Red vinyl is hot, blue is cold. Radiator flows 700gph, CPU's flow 140gph each.

http://youtu.be/xtSGRYmH2Bs

http://youtu.be/Uu1OnOTJwwU

Going to let it run a few hours and check flow again before booting. Crossing fingers.
 
Last edited:
So why have you moved it to a case? That box is probably worth more than those El Cheapo blocks combined. :D
 
IIRC, that case is $130 right now. And Yes, that's more than I spent on the blocks. :D
 
BTW usually those peak flow specs on pumps are at zero restriction, just as the max pump head spec is measured at zero flow. El Cheapo waterblocks are more than likely anything but low-restriction....so unless you have a flow meter, you probably are not getting 140GPH through any of those blocks. ;)
 
TPC is not reading temps correctly.

It's claiming at idle, the highest is 6°C, the lowest °4.

Air temp is 20°C, water sump is 19°C.
 
So if the water was running 19°C, average was 5°C at idle, it's folding at 2.7ghzx64 cores running 19°C average, the true core temp avg is perhaps 33°C.

But it's all meaningless unless the room temp is summer = 40-45°C. Pretty sure it will stay under 60°C. Guess we will see.
 
Holy Christ you were not kidding lol.... (golf clap)...
 
LOL. I love it!

Why not go block to block on the water loop?
 
Let the CPUs run idle in the lowest p-state (if you can drop the voltage even more, do so). Then measure the water temperature, and read the CPU temps.They should be very close to the water temperature when they transfer nearly no heat to the loop at all. Then you've got your delta, assuming the sensors don't "freeze" at some temperature, and we're only talking about an offset, and not a difference that's not constant as the temperature rises.

Here's a pic of my WC setup - not sure it's been posted here already:

http://imgur.com/pBYaJUJ - your mounting solution is alot easier than mine :p
 
Block to block would mean CPU's get progressively hotter down the chain.

Need to do more measurements, but TPC isn't reporting right.
 
Qinsp said:
Block to block would mean CPU's get progressively hotter down the chain.

Need to do more measurements, but TPC isn't reporting right.
Click to expand...

That would only be true if the system does not equalize in temperature. But in fact if this is completely passively air cooled, the system will equalize in temperature.
 
Qinsp said:
Block to block would mean CPU's get progressively hotter down the chain.

Need to do more measurements, but TPC isn't reporting right.
Click to expand...

Temperature delta is usually pretty small (assuming adequate flow), on the order of a few degrees C.

TPC does not measure (as AMD chips do not report) actual temperature.
 
jmilcher said:
That would only be true if the system does not equalize in temperature. But in fact if this is completely passively air cooled, the system will equalize in temperature.
Click to expand...

It will reach an equilibrium, where the temperature will no longer change at a given point in the loop, but that does not mean the temperatures will be equal at all points in the loop.
 
OK, inlet air temp to radiator: 26.2°C (this is room temp)
Outlet air temp from radiator fan: 33.1°C
Water temp at sump: 34.3°C
Water temp at CPU1 outlet: 35.5°C
TPC reported CPU temp: 37-41°C (one CPU is high).
TPF for WU P8103 = 8:25
Watts for system, 855w + water pumps and AC fans.
 
So in theory, if restriction isn't an issue, the 4th CPU in a string would be 4°C hotter. Not much.
 
I'm going to oneunit it and let it set for the weekend.

HAHAHA!! This is cool (sic).

I halted it, within seconds it dropped to 21°C. You could feel the water temp falling by putting your hand in it.
Put -oneunit on it, and as soon as it starts folding, it jumps to 40°C and you can feel the heat coming out of the radiator.

I'm amused easily. :D
 
Last edited:
Temps reported by AMD CPUs aren't technically deg C (though they are deg-C-like).

From AMD documentation:
Tctl is the processor temperature control value, used by the platform to control cooling systems. Tctl
is accessible through SB-TSI and F3xA4[CurTmp]. Tctl is a non-physical temperature on an arbitrary
scale measured in degrees. It does not represent an actual physical temperature like die or case
emperature. Instead, it specifies the processor temperature relative to the point at which the system
must supply the maximum cooling for the processor’s specified maximum case temperature
and maximum thermal power dissipation.
Click to expand...

Despite this shortcoming, Tctl is still of significant use. It _does_ tell you how well you're cooling
your system. Seeing single-digit number tells you you're doing helluva job :)
 
Thanks for explanation Tear!

I kind of figured it used a different zero point. The degree deltas look right, just not the zero.

So each chip design can have a different zero, depending on it's limit temp.

(sidebar - I said the wrong speed in the second video, the "big" machine has 2.6 chips, the "little" one has 2.2's)
 
It's a bit more than just zero-point. To me it looks more like full operational-temp-range calibration.

Once can see it pretty clearly in AMD power and thermal datasheets (e.g. pub 43374 http://support.amd.com/us/Processor_TechDocs/43374.pdf).
All CPUs have Tctl Max set at 70 deg Tctl (erroneously labeled deg C!) but their Tcase Max temperatures vary significantly.
 
just goes to show how overly priced everything really is...
marketing n plans for company growth is just a terrible ways to look at engineering solutions.
imo
awesome work tho, i love seeing people realizing they have brains too lol
 
You must log in or register to reply here.
Back
Top