mbm diode vs socket temps

S

Spewn

2[H]4U
Joined
Jan 5, 2001
Messages
3,459
So, based on some fiddling around with mbm5, I was able to get my cpu diode sensor to "work". However, it appears that at least on my asus a7n8x with a barton 2500+@3200+, the diode isn't good for much. At idle, the socket read 32, diode read 38, I thought sure, could be true. I started up prime95, and within *seconds*(like, 2 or 3) diode temp shot up to around 48'C, and is bouncing around all over between 46 and 49. Does this sensor actually work at all? The socket temp sensor slowly but steadily rises when I run prime95, usually up to about 38'C, but the diode is jumping all over, and not staying consistent at all with the socket temp sensor. Does this make sense?

Upon exiting prime95, the "diode temp" dropped right back down to 40. The diode temp seems to be nearly 100% dependant on cpu load, and nothing else. IE, if the cpu load goes up to 20%, the diode temp will rise. If the cpu load is sitting at 0, the diode temp will sit around 38'/39'. At 100, it sits at 49. Is this normal?
 
"The diode temp seems to be nearly 100% dependant on cpu load"

Ya think? ;) That's why it moves so fast.
 
That is pretty much right. The socket thermistor, if you assume it reads temps properly in the first place, is sort of separated from the actual CPU by a buffer of sorts. All the materials between there, ceramic/organic CPU substrate, air, etc, take time to heat up as does the thermistor itself. The diode in the die is right within the heat source, so it responds much more quickly.

Of course, you still can't be sure that either of the temps they give out are valid, so you can't really compare them. They are probably off by some constant amount and some complex scaling factor as well. But theoretically then cpu diode should have a much faster repsonse time and should read higher than the socket thermistor at all times.
 
I think I should clarify my point.

As soon as MBM takes a reading with the cpu at 100%, the "diode" temp reads at 49'C. If I shut down prime95, as soon as MBM takes a reading with the cpu at 0%, or 3% or whatever, the "diode" temp reads out at 40'C or so. If I let prime95 run for a while, the socket thermistor slowly but surely climbs up to 38/39'C, but the "diode" temp doesn't budge at all. When I shut it down, the diode temp instantly drops, and the socket temp slowly goes back down.
 
Originally posted by Mo_BurLy
"The diode temp seems to be nearly 100% dependant on cpu load"

Ya think? ;) That's why it moves so fast.
Click to expand...

What I mean is, if MBM took continuous readings from the sensors, I could start/stop prime95 over and over, and the diode temp would be instantly jumping up and down through about a 10'C range, which makes no sense.
 
Originally posted by Spewn
What I mean is, if MBM took continuous readings from the sensors, I could start/stop prime95 over and over, and the diode temp would be instantly jumping up and down through about a 10'C range, which makes no sense.
Click to expand...

Why doesn't it? Heat output increases with load, so temperature increases. Heat load decreases when unloaded, temp goes back down. 10C is not that much of a change if you have any decent cooling system.
 
Originally posted by zer0signal667
Why doesn't it? Heat output increases with load, so temperature increases. Heat load decreases when unloaded, temp goes back down. 10C is not that much of a change if you have any decent cooling system.
Click to expand...

*instantaneous* temperature changes though? That's the point I'm trying to get accross, that's what doesn't make sense to me :/
 
Take your heatsink off, than turn on your computer and tell me which gets hotter first, the socket or the processor. One your bascially heating air, the other is a direct core readin. Of corse it makes perfcet sence. The heat is fairly instintaneous
 
Originally posted by Spewn
*instantaneous* temperature changes though? That's the point I'm trying to get accross, that's what doesn't make sense to me :/
Click to expand...

Nothing is instantaneous, but it could easily happen in less than a second. We're talking about a fairly large change in heat output, a pretty small die and a big heatsink. The actual amount of heat transfer that it would take to cool your core 10C is very little compared to what the heatsink can easily absorb.
 
Originally posted by zer0signal667
Nothing is instantaneous, but it could easily happen in less than a second. We're talking about a fairly large change in heat output, a pretty small die and a big heatsink. The actual amount of heat transfer that it would take to cool your core 10C is very little compared to what the heatsink can easily absorb.
Click to expand...

Well, I'm watercooling, it just seemed odd is all. Mainly because the socket temp will slowly rise/fall, but the diode temp does so instantly.
 
So what don't you understand? I thought this post explained that effect.

Originally posted by zer0signal667
That is pretty much right. The socket thermistor, if you assume it reads temps properly in the first place, is sort of separated from the actual CPU by a buffer of sorts. All the materials between there, ceramic/organic CPU substrate, air, etc, take time to heat up as does the thermistor itself. The diode in the die is right within the heat source, so it responds much more quickly.

Of course, you still can't be sure that either of the temps they give out are valid, so you can't really compare them. They are probably off by some constant amount and some complex scaling factor as well. But theoretically then cpu diode should have a much faster repsonse time and should read higher than the socket thermistor at all times.
Click to expand...
 
My temps in MBM5 on a A7N8X Delux are abit sketchy.
My cpusocket reads 41C. And my diode reads 0C. I think my diode may be broken or not recieving something correctly. ANy ideas?
 
Originally posted by swoop56
My temps in MBM5 on a A7N8X Delux are abit sketchy.
My cpusocket reads 41C. And my diode reads 0C. I think my diode may be broken or not recieving something correctly. ANy ideas?
Click to expand...
You just have it configured wrong. Look for your specific board and revision in this list, and set the sensors accordingly.
 
You must log in or register to reply here.
Back
Top