kt3 ultra cpu probe accuracy, interesting findings

[g.l.amour]

already posted this on the amdmb.com boards to see what others are seeing.

last couple of weeks i have been fiddling with my watercooling setup alot. here's why... i see ppl reporting some pretty low temps whereas mine seems to hover more than 20°C higher than ambient.
that is in some cases not so bad, but my system consists of dual parallel mounted rads (one big momma, one double size big momma); a maze2 block ; an eheim 1250.

weird thing was that cranking up the fans from 5v to lets say 12v on the rads didn't lower temps more than 2°C. so i decided to take the senfu probe that has been living side to side with my ti500 core and give it a swim in my reservoir. the res reports 1°C over ambient!!! so those rads are doing some damn fine work. i think those temps are accurate; i got a inside/outside car thermometer in my room that reports in a 0.1 tolerance the same as the sensor in the res, after res water has been alowed to cool for 4 hrs with nothing running, i thk that after such a time interval those temps should be the same (water , air).

at that point i was thinking some pretty nasty stuff about the maze2, it didn't seem to be absorbing alot of heat from the cpu since cpu temps were always at +/-40°C under 100%load (for the record 19.5°C ambient, +/- 1670MHz , 1.775 V)

aaight time to stick that probe as close to the cpu as possible; after changing position of probe + bending + dipping it in heatpaste; i got the highest result : +/- 5°C less than reported by mbm5. the best trick seemed to be to drench it in heatpaste and bend the tip 90°C so good contact is made.

now, everything seems to fall in place. the block might do an even better job, considering i got a system optimized for high flow, but i got a low flow WB; but the temps seem to be what can be expected of such a system. the new swiftech block would be ideal i guess.

what am i getting at? learn from my mistake, don't be a l4m3r, before investing in new material, gather all the info u can, watertemps, flowrates, cpu temps ,... otherwise u will be throwing away a considerable amount of money. point nr.2 [H]'s stance on the accuracy of the motherboard probes isn't helping ppl who believe that alot; c for yourself.


what goes for casemodding also works for H2O cooling. measure twice, buy once; lol.

ohh, and for the ones that might think i can't mount the WB properly; think in all the fiddling i have done, i mounted it +/- 5 times for some unlocking, same readings over and over.

the msi kt3ultra seems to be reporting +/- 5°C more than actual core temps

if others have some experience with the sensor on this board, pls share them with me and others. interesting for sure.

thx for having taken the time reading all this ranting.