Boy... it's awfully quiet in here lately...everyone else must have a life....
Time for some preliminary results/conclusions/questions.
Here are some temps from a 3 hour session last night.
Sprocket had been running stable so I ran some benchmarks to confirm the hardware/software config.
She passes Memtest and Prime, no problemo.
As I had also changed the vidcard cooling while I had her apart ( removed the Zalman Heatpipe cooler and installed the Zalman northbridge HS on the VPU) I also ran Aquamark3 a few times. I can't seem to break 33,000, but the temps were fine ( still doesn't breach 32c running the test) and I think I just need to research my settings and tweak it to go higher. Not that I care about the video that much....
So we Folded for a while...
As you can see, the CPU temp didn't break over 40c which is a perfectly acceptable result.
I think that the large reservoir that the pump now sits in really does a fine job of stabilizing the temps- there is only a 2 degree variance over the 3 hour period.
I suppose that my next step would be to increase the radiator efficiency, but that would entail a pump change also and I'm not up for that quite yet.
Last week I had also performed my motherboard wall fan mod and I'm very ambivalent about the results.
I cannot tell exactly how well it really cools the socket/CPU as I believe that I am spoofing the onboard sensors.
I have 6 hardware temp sensors and have been moving them around just to see what's what.
Two of them are just dangling in the case to give me interior ambient temps.
They consistently read 5-10c
higher than the "case temp" reported in MBM5.
Wherever that board sensor is located, the fan is obviously skewing it's function...unplugging the fan causes the reported temp to rise to within a degree or so of my hardware reading. CPU temp also rises, although not as drastically (2-3c).
So, it's obvious that the fan artificially lowers the MBM5 reports...the question is does it actually perform any useful cooling function also?
Until I can figure out how to access the on-chip diode ( which I'm beginning to fear is just not possible with the ABIT board) then this question is moot- I have no way to tell.
Sprocket has been running clad in cardboard...I took my own advice and didn't hack up the case skin yet...but I'd kinda like to finalize the layout and run her clothed properly. I'm expecting the sound level to decrease with the full steel enclosure in place and that would be nice. Not that she's loud now, but there is some air
whoosh which should be muffled by the heavier steel skin.
I also have decided that I need to provide some active cooling to the newly-mounted HDD. It is running about 5-7 over my case ambient ( the real, hardware sensor reported temp).
This could be difficult as my space is limited and I've more or less boxed myself into a corner, installation-wise. All of the other considerations are perfect as the HDD is curently mounted ( specifically the cabling/power connectors are all just the way I would like to keep them), it's only the cooling that is awkward.
I'll figure out something...I hope.
Okay, back to your breakfasts.
Edit:
Results from a similar run this morning...the difference is the CPU load, just forum browsing and e-mail...