Place sensor on 1st or 2nd gpu ??

Franco_cov · 27 Jun 2009 at 10:04

Guys,

I have a inline heat sensor that can be used between barbs in my water cooling kit and wanted to know which is the best gfx card to put it after to get the most accurate temp reading for my crossfire cards?

Would it be after the first card, before the second or after the second??

In the process of doing this so a quick response would be very much appreciated

twist3d0n3 · 27 Jun 2009 at 10:05

won't the temps balance out through the loop?

Franco_cov · 27 Jun 2009 at 10:22

Well I am placing a sensor at the cpu so that i can monitor that and just wanted to know where to place the sensor for the gfx cards as I have a temp display unit that will monitor cpu, gfx, case temps so I am monitoring specific parts rather than the general loop

JonJ678 · 27 Jun 2009 at 14:23

If you're following the reasoning of water in goes over hot thing and leaves hotter, you want it downstream of the second graphics card.

At reasonable flow rates, loop temperature is likely to be nearly uniform. There will be (small) temperature variations that may be worth monitoring. If the probe introduces resistance to flow (likely) I'd suggest leaving it out, as the two probes are likely to show the same temperature anyway, especially considering errors in calibration.

Are these probes good? I'd quite like a means of monitoring water temperature

Franco_cov · 27 Jun 2009 at 17:53

k a bit miffed now i have placed inline flow temp sensors one on the out going barb from cpu and one from the out barb from 2nd gfx (crossfire) and the readings are similar but compared to figures i get from gpuz and core temp its about 10c+ below?

Is there anwa I can get a true reading to feed my lcd display without relying on software as i want to monitor temps real time with the display rather than software??

JonJ678 · 28 Jun 2009 at 04:55

At least one of us is confused I fear. Your water temperatures are lower than your core temperatures, but you are surprised and angered by this? Of course your water temperatures are lower, thats the whole point. You maintain a thermal gradient down which heat moves from the cpu. The temperature difference between cpu core and water depends on water block, flow rate, coolant used. You can't hope to measure core temperatures solely by measuring water temperatures, so what are you hoping to do?

I'm unsurprised that the readings are similar.

re 'True reading'. None of your probes, built into chips or otherwise are 'accurate'. The chip ones are only designed to organise thermal throttling so will hardly be accurately calibrated. The precision of the inline probes rather depends on the quality of the probe and how much effort went into calibrating that. If you changed the wire length you'll invalidate the reading on a few of them. It'll also depend on where in your loop it's placed, i.e. near the edge of the tube, at a region of turbulence or laminar flow etc.

None of your temperatures are 'true', they're accurate to within acceptable limits.

Use the software, the computer itself absolutely dominates any fan controller system you might want to put into place. Speedfan used to do this respectably on windows, it's scripted cleanly under linux. I assume you're not running osx. LCD displays are nice I don't doubt, but you may as well feed them from software. That way you get water temp, cpu temp, and gpu temp rather than two near identical readings for water temp.
I use conky instead