1055t going down to x4

[Hagar]

Does TMPIN1 react as fast as the "Core Temperature" readings though?

Yes, in both directions, they are related.

andy.

 

[pswfps]

I have listened, considered and decided. Core Temp is the way to go - sorry you deal.

 

[Martini1991]

Ok.
If you think your CPU can run below ambient on the h50, that's your downfall.

 

[pswfps]

Yes, in both directions, they are related.

andy.

Yes they are related. Just that on mine the TMPIN figure reacts slower than the "core temp" figures. You would logically conclude that this is because the "core temp" figures are more closely related to the CPU than the TMPIN figure.

 

[pswfps]

Ok.
If you think your CPU can run below ambient on the h50, that's your downfall.

... but I don't think that. How have you come to that conclusion?

 

[Martini1991]

... but I don't think that. How have you come to that conclusion?

Because that's what'll happen....
I've had my core temp show as 9c, when my room's like.. 20c..

But if you think it's perfectly fine for my CPU to be idling at 20c at 4.25GHZ at 1.5v.. Again, your downfall.

 

[GoogalyMoogaly]

Does TMPIN1 react as fast as the "Core Temperature" readings though?

It's very close when increasing and when left to idle I think it drops down quicker.

 

[pswfps]

No, you don't understand at all Martini. I expect my CPU to run well above ambient hence the whacking great offset of 22C I apply to the "Core Temp" figures. What was that about not listening?

 

[Martini1991]

You're whacking 22c onto Core temp. How on Earth is that any accurate representation?

 

[pswfps]

Because I have found that under load, the CPU temp is around 6-8C above the socket temp (which I think is probably TMPIN0 or 1) , although that was using a 125W cpu. The offset of 22C in coretemp just seems to give the right ballpark figure.

 

[GoogalyMoogaly]

Because I have found that under load, the CPU temp is around 6-8C above the socket temp (which I think is probably TMPIN0 or 1) , although that was using a 125W cpu. The offset of 22C in coretemp just seems to give the right ballpark figure.

How did you find that? What did you use to measure the 'CPU temp'?

I will say that if I go into the BIOS and check the temps they are much closer to the TMPIN1 temps than the Core temps. I think it reads them a bit higher than the TMPIN1 values, but it's hard to compare as I have to boot the PC to check the HWMonitor values.

 

[pswfps]

Anyway, that's it for me tonight. You know, it should be possible to discuss this stuff without getting all huffy about it Martini. Huh?

 

[pswfps]

How did you find that? What did you use to measure the 'CPU temp'?

I will say that if I go into the BIOS and check the temps they are much closer to the TMPIN1 temps than the Core temps. I think it reads them a bit higher than the TMPIN1 values, but it's hard to compare as I have to boot the PC to check the HWMonitor values.

Ah well, I hope you're sitting comfortably....
Not so long ago I had a 965BE that would return sensible temps in the "Core Temp" readings in HDMonitor and Everest. Under load, this 125W CPU would give "Core Temps" around 6-8C higher than the TMPIN0 figure on HDMonitor/Everest. It was the same on this Asus board and my previous Gigabyte GA-790XT-USB3 board.

When load was removed, the CoreTemps dropped very quickly to normal values whilst the TMPIN0 figure drifted down slowly, to eventually rest about 2C less than the CoreTemp.

Given that the "Core Temp" figure reacted more quickly than the TMPIN0 figure, and that TMPIN0 was always lower, I made the following conclusions:

1) The "Core Temp" is the actual CPU temp
2) The TMPIN0 figure is a socket sensor reading under the CPU
3) That generally, a 125W CPU will reach a load temp of around socket temp + 6 or 8C.

Now I'm just applying the same methodology (admittedly as rough as it is) to this new X6 chip.

 

[Martini1991]

The core temp is a software generated estimate figure.
The fact it exists/reacts means nothing.. It's like a simulation, it's fabricated.

 

[pswfps]

All the figures you see on the screen are in fact software generated Martin. They are all the result of software algorithms applied to some base sensor data.

The fact that "CoreTemp" reacts very quickly to CPU stress is actually very significant. It means that it is the sensor most closely representative of the CPU workload and, by extension, temperature. The fact that this same sensor is used by AMDs own Overdrive Software must also surely tell you something? The problem is that the base sensor data varies form model to model which is why software has a hard time interpretting the data. It seemes to work OK for the 965BE but not this more recent 1055T.

Looking at nKata's 5 min Prime screenie, I would estimate his CPU temperature to be around 55*C. That is 33C core + 22C offset = 55C. Do you not think that sounds reasonable?

 

[Martini1991]

You're guessing and applying what you assume as fact.

 

[pswfps]

LOL, and what are you doing? I haven't seen much evidence of a well reasoned argument from you. Just a lot of shouting with nothing to back it up...

 

[pswfps]

Why don't you post up you own HDMonitor screenie - I could estimate the temperature of your own 4.25GHz overclock if you like. Go on, it will be fun!

 

[Martini1991]

I run 4GHZ at the moment with less volts and my fans slower, CPU says I'm idling 38c, Core temp reckons 18c.

 

[pswfps]

Well that seems about right to me. At idle I'd expect the CPU to be about 2-3C higher than the socket temp. If I apply the 22C offset to your "CoreTemp" reading I get 40C, ie 2C higher than your TMPIN0/1 socket reading. Seems to work.