CPU temps vs Core Temps. What reading matters most for OCing?

[Bleeble]

ok I've used all three monitoring programs, as well as Everest Professional edition...They all read the same thing, Core temps in the 20s-30s and a CPU temp ranging from 40s to low 60's.

 

Today when it was 77 degrees in my house, the Core temps read as low as 24 degrees C, which is around 75 Farenheit!

 

From what I've read, it is absolutely impossible to have a CPU temp lower than the ambient temp...so I'm guessing my P5B motherboard simply has bad sensors for the Cores. It must be underreporting them for the core temp to be at or below the ambient temp..

 

I can't be the only P5B owner with this problem..

Correct. With an aircooled processor, it is impossible to have sub-ambient temperatures. The temps shown in CoreTemp and TAT are reported directly from the CPU. If you have your computer on the floor, the air down there can be significantly cooler than the air higher up in the room (specifically around the thermostat). Check your temps again using CoreTemp or TAT.

[Capitalizt]

Thanks for the reply...

 

one more question:

 

For a typical system with decent air cooling, do you know how many degrees above ambient temperature the Cores usually are?

[Bleeble]

I'm not 100% certain, but I would say 5-10C above ambient for a non-OCed Core 2. It's probably closer to 15-20C above ambient with a good OC. There's really a lot of variables to consider. Some CPUs run hotter than others. Some IHS's are concave, greatly reducing contact with the heatsink. Different heatsinks have different performance. Different fans. Different thermal paste... and the list goes on. *shrug*

[Cooperstown39]

First, none of these programs provide a real temperature from the cpu core. The readings are all based on calculations. Each Intel chip is individually calibrated before leaving the factory with a temperature in which it will shut down. The only reading accessable from the cpu core is how far away the temperature is from reaching this shut down point. Unfortunately, there is no way of knowing the exact shut down point (unless you want to push your system and force a shut-down). All of these programs assume one of two Tjunction temperatures, either 85C or 100C (depending on the processer). The reading from the cpu is subtracted from either 85 or 100C to derive the temperature reading. Remember the 85 or 100 are only estimates, the actual shut down temperature in which Thermtrip# goes active is specific to each chip.

 

Rather than worrying about temperatures, many suggest that knowing the actual reading from the cpu core on how far away the system is from activating Thermtrip# is preferred. This will tell you exactly how much further you can push the system during OC'ing. There are a number of ways of getting at this reading. The easiest method is to use CoreTemp program. There is an option where you can check the box to show the delta to Tjunction.

 

A more complicated but accurate reading can be made using the following (compliments of bh626pro on Tom's Hardware forum);

 

Download crystal cpuid; http://tinyurl.com/24k6su

 

In CPUID, run Function/MSR Editor and enter 0x19c in the MSR number box - this is the register for the DTS sensor - it directly reads the DTS off of the core.

 

click the RDMSR button - In my case I get a value under EAX of 0x88370000 at idle (e6600 @ 3.0).

 

Take the last 2 chars before the 0's (37) and enter this Hex number into the Windows scientific calculator. Click the hex option, enter 37, and click the Dec option. You get 55. You subtract this value from the probable TJunction point, 85C and get a core reading of 30C.

 

 

Most of this information came from two different threads;

 

 

Toms Hardware Forum thread

 

Extreme thread

 

 

There is also a very good "temperture guide" at the following link; Temperature Guide

 

The opinions on the "acceptable" dts reading vary widely. Although I OC, it is not for gaming purposes. I am content with dts of >30C.

 

Hope this helps

[kendellrt]

Using Small FFTs will get you higher temps. TAT is higher yet. That said, the temps you see with Orthos Blend are more like what you'll find with real-world usage (games, encoding, etc.). Personally, I use Small FFTs (priority 8) for stability/temp testing.

Why do you use priority 8?

[Bleeble]

Why do you use priority 8?

It's more stressful than priority 1. I've had priority 1 run for hours on end with no errors, but priority 8 will fail within an hour. I've tried it a few times to make sure it wasn't just a coincidence and the same thing happened each time.

[kendellrt]

It's more stressful than priority 1. I've had priority 1 run for hours on end with no errors, but priority 8 will fail within an hour. I've tried it a few times to make sure it wasn't just a coincidence and the same thing happened each time.

So do you use small fft with priority 8, or just the blend? I guess it depends on whether you want to test the whole system, or just the CPU.

[Bleeble]

I guess it depends on whether you want to test the whole system, or just the CPU.

Correct. I use Small FFTs for testing the CPU and Blend for the RAM and CPU (among other things). Both with priority 8.

[kendellrt]

Correct. I use Small FFTs for testing the CPU and Blend for the RAM and CPU (among other things). Both with priority 8.

I think I will do that at stock, and then overclocked.