Jump to content

Wrong CPU temp. reading on Haswell platform?


ACHU

Recommended Posts

Hi, to monitor the CPU temperature on our Haswell system, we used both BIT and AIDA64. Somehow, the temperature readings are quite different. We don't know which one is correct.

 

Also, the CPU Tjmax is 100C. When CPU temperature hits 100C, it's supposed to throttle. However, we found most of the systems throttling but the max CPU temperature was only 95C or even 84C.

 

Is there any adjustment required to get the reading correct on Haswell platform?

 

We did the test on 14 systems. All of them are the same systems based on Haswell i7-4650U.

 

See our test data as posted.

post-27712-0-07426200-1408515379_thumb.jpg

Link to comment
Share on other sites

No adjustment is necessary about TJMax on Intel Nehalem and Silvermont family processors. TJMax adjustment was only necessary in the LGA775-era.

As for throttling, it may or may not kick in at TJMax. Please note that the on-die CPU temperature sensors aren't that accurate that one may expect them to be. Also, when any of the CPU cores momentarily hits TJMax, the CPU will start throttling immediately, and the reported DTS temperature will quickly drop down a few degrees Celsius. The whole thing overheating protection mechanism works very quickly, and the measured temperatures will fluctuate a lot.

Link to comment
Share on other sites

DTS can be found on the following. Which one does AIDA64 use to show throttling is triggered?

1. per IA core

2. on the System Agent (SA)/Uncore

3. on the GT core

4. L4 Cache (4+3e parts only)

 

Also, when you say when CPU start throttling, the reported DTS temperature will quickly drop down a few defreecs Celsius, how quick is that? Within  1-2seconds?

Link to comment
Share on other sites

DTS can be found on the following. Which one does AIDA64 use to show throttling is triggered?

1. per IA core

2. on the System Agent (SA)/Uncore

3. on the GT core

4. L4 Cache (4+3e parts only)

AIDA64 uses CPU MSR register to detect throttling activity. It is per core, so #1 from your list. I'm not sure if there's a register that one could use to detect throttling activity for other parts of the CPU package.

 

Also, when you say when CPU start throttling, the reported DTS temperature will quickly drop down a few defreecs Celsius, how quick is that? Within  1-2seconds?

Much quicker, typically just tens of milliseconds. AFAIK throttling kicks in when the CPU is overheating for 20 milliseconds (or more), so the whole thing is very sensitive and very sharp. It has to be, or else it wouldn't be able to prevent a critical overheating situation damaging the processor.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.



×
×
  • Create New...