I checked the Abit forums, that is indeed their official response to the people asking about these high readings. This is the original thread here I saw it in:
http://forums.anandtech.com/messageview.cfm?catid=37&threadid=1059901&FTVAR_MSGDBTABLE=
Here's Abit's:
http://forum.abit-usa.com/showthread.php?s=4895d5830ca059f5333027d6f647cbbc&threadid=11496
It seems to apply to all Intel P4 chipsets on Abit boards, and seems to be due simply to the way Abit calculates the temperature reading. I haven't read that whole thread. It may be that all the other brands simply do calculations that are valid only on their boards, and can't be compared to others. We all know that temperature readings can vary widely by brand or model. In a way the readings can't be considered absolute temperature readings, but more like numbers relative to themselves -- the best idle temperature on an Abit Intel board is 50 or so, while the usual load temp is 75, and that's normal for their boards, while Asus may calculate it so that the temps are 35 idle, 50 load.
They may have realized people are used to seeing temps below 50 or even under 40 with their P3's and XP's, even after such extreme frequencies that we see now, so they just made the numbers come up in a way that people would recognize. Like, the chip diode may actually be sending a signal indicating the internal diode is at 75 degrees, but on an older board that didn't access an internal diode, the external measurement would have come up with something around 50 degrees, the best measurement possible externally -- the actual core was still 75 degrees, and that's why we worried when it reached 60 degrees because it might actually have been around 95 degrees in the actual core.
Abit therefore is reading the diode and outputting the actual core temperature, and you have to use the Intel specs to be sure it's within valid ranges, while other brands are outputting an "interpolated" or compensated reading comparable to previous motherboard sensor readings.
My roommate's board is indeed completely stable at 75C at 3GHz overclock (and the overclocking doesn't make the temperature much higher than the default speed of 2.4). We even got it running at 3.6GHz with the memory at 3:2, with a bit of voltage increase, but it wasn't stable in 3DMark still so we need to tweak more. When we first installed it I used the Intel black thermal pad, and decided to take it off to put Arctic Silver on (before I had found these are common temps). The thermal pad didn't seem to be making great contact; in spots it had been rubbed completely off the aluminum foil while in most spots it had just barely had the ridgy compound pressed flat, and yes the heatsink was well secured.
Since putting Arctic Silver 3 on, with the case closed up it only gets to 63C while running 3DMark2k1, and is entirely stable. It idles at about 50C. Yet another justification for Arctic Silver, it does make a difference in some cases.
It's very odd how Abit designed the BIOS settings for the thermal warnings and stuff. The thermal shutdown can only be set as high as 75C. But you can set the thermal alarm as high as 120C. So it's impossible to have the alarm go off above 75 and then have it shut off if it continues getting hotter, like when you're not on site.
Oh yeah, he has a 9800 Pro installed which is far hotter than what's coming off the CPU. It's kind of scary how small the heatsink on the 9800 is given how hot it really gets; it's too hot to touch for more than a second.
We also discovered that the BIOS version makes a big difference in overclocking ability. With the 1.0 version, we can overclock to whatever we want. With I think 1.1 and 1.3, the newest we could get, we can hardly get more than like a 210MHz bus. Any higher and the system was highly unstable and rebooted randomly.
) then you should be fine.
Facebook
Twitter