Jump to content
Sign in to follow this  
palehorse

What does your DFI board report for CPU temps?

  

347 members have voted

  1. 1.

    • I believe that the reported CPU temperatures are accurate
      119
    • I believe that the reported CPU temps are off by less than -10C
      100
    • I believe that the reported CPU temps are off by -10C to -20C
      63
    • I believe that my CPU temps are all over the place and have no idea how far they are off!
      24
    • I have no idea.
      41


Recommended Posts

one can hope, cant he? there was also a very lengthy discussion and even more data in this thread:

http://www.dfi-street.com/forum/showthread...light=cpu+temps

 

here are some excerpts from Bulldog in those threads:

 

* "Just so everyone involved in this discussion can sleep at night I have had some reports of low CPU temp readings from various users here in Europe as well, which lends credibility to this discussion. I couldn't reproduce these funky temp readings myself with my San Diego 4000+ or Venice 3000+. I have however taken the claims seriously and am currently discussing the matter with my contacts at DFI HQ to see if we can pin the problem. It's difficult to get a concensus on this as it seems random. However if you guys can can report the "ADA..." codes imprinted on the surface of your CPUs it would help alot. I will check back in a few days and gather the reported codes and see if we can get a pattern out of this."

* "Ok thanxs for those codes. I will discuss this matter with my fellow techs in Taiwan and see if we can get this nailed down."

* "In any event the matter with the inaccurate CPU temp readings on the Expert board is currently being looked into. The problem is that it is the Chinese New Year vacation in Taiwan so the headquarters is closed this week. When the engineers return next week I will chase them on this matter further."

 

 

I made this thread and poll in response to his and the engineers' need for more accurate data; so that one day, god willing, we may see this fixed.

 

So like I said, they ARE aware of the issue and have been looking into it with the intent of fixing it. There hasnt been anything "official" stated or announced, but that would be nice.

 

Thus, "FUD" or not, I'm doing my best to ensure the issue is not brushed aside. So what was your point again?

 

 

What bulldog is saying is pretty much the same what I've said in a previous post - there is enough evidence to form a suspicion, and they will look into it. But no word of actual acknowledement. For the sake of splitting hairs I have to agree with LDog69 here. Don't make the mistake of overinterpreting, palehorse. As I said before, you can't have a true academic approach at something if you are biased to begin with aka trying to prove your point.

 

On a side note, is the Opteron officially supported on the Expert? I'm pretty sure it isn't on the Ultra/SLI models.

Share this post


Link to post
Share on other sites
What bulldog is saying is pretty much the same what I've said in a previous post - there is enough evidence to form a suspicion, and they will look into it. But no word of actual acknowledement. For the sake of splitting hairs I have to agree with LDog69 here. Don't make the mistake of overinterpreting, palehorse. As I said before, you can't have a true academic approach at something if you are biased to begin with aka trying to prove your point.

 

On a side note, is the Opteron officially supported on the Expert? I'm pretty sure it isn't on the Ultra/SLI models.

DFI does not support the Opterons at all, but unfortunately for them, this issue does not seem to depend upon any particular type of cpu.

 

The other thing I'd like to state is that they have never been too clear is exactly how/where they measure the cpu temp. the answer to this question could go a long way in settling the grievances with this issue.

Share this post


Link to post
Share on other sites

There are only a handful of boards that officially support Opterons (primarily made by Tyan). With that said, this problem is not limited to that chip.

 

Here are my thermal resistance numbers:

 

Proc Speed: 2700/1800 = 1.5

Voltage: 1.485/1.32 = 1.125

 

1.125 x 1.5 x 110 = 185.625 (Total Output)

 

38-20 = 18 (CPU Delta)

 

18 / 185.625 = 0.09697 Thermal Resistance

 

Best recorded thermal resistance for the SI-120 that I've found is .14 C/W. So either I've installed water cooling without realizing it or my reported temps are off. Given that I have no idea what my actual core temps are like, I've decided to back off on my OC.

 

I'd still stand by my statement that the enthusiast should steer clear of the DFI boards until something is done about the temperature issue.

 

[edit]I'd just like to add that I doubt that DFI would ever come out and officially state that there is something wrong with the temperature sensor in their boards on this semi-official support forum. In fact, I'm surprised that Bulldog has offered up the information that he has. [/edit]

 

Chum

Share this post


Link to post
Share on other sites
edit]I'd just like to add that I doubt that DFI would ever come out and officially state that there is something wrong with the temperature sensor in their boards on this semi-official support forum. In fact, I'm surprised that Bulldog has offered up the information that he has. [/edit]

 

Chum

 

We have to get clear on the definition of "wrong" first, to determine an issue. By the definition, all sensors from all manufacturers are 'wrong' to a degree, depending on the exact location of the sensor, the type of sensor used and the way the reporting is implented in the BIOS.

 

Hey, I hope that you all still consider this to be fruitful debate - I don't want to see this thread turning into a fruitless argument and hope that my input (ie constructive criticism) helps to achieve what palehorse is trying to achieve. If not just tell me to shut up. :D I won't be offended.

Share this post


Link to post
Share on other sites
We have to get clear on the definition of "wrong" first, to determine an issue. By the definition, all sensors from all manufacturers are 'wrong' to a degree, depending on the exact location of the sensor, the type of sensor used and the way the reporting is implented in the BIOS.

 

Hey, I hope that you all still consider this to be fruitful debate - I don't want to see this thread turning into a fruitless argument and hope that my input (ie constructive criticism) helps to achieve what palehorse is trying to achieve. If not just tell me to shut up. :D I won't be offended.

 

I accept that there will be some variance when it comes to interpreting the internal thermal diode on these chips. My board as well as many others are reporting temps that are beyond what I would say is acceptable. If you look around at some of the other enthusiast forums, you can generally find threads on this subject (DFI Temp issues) when other manufacturers don't appear to be having problems.

 

With that said, I think your input has been very valuable to the thread and by no means want you to "shut up". :) Your thermal resistance posts motivated me to calculate the above. I guess I just find myself frustrated when someone tries to tell me that there is no problem, or that "it hasn't been proven" - when I can clearly see first hand that there is. Ultimately, I'm hoping there's a simple fix that doesn't require replacing the board.

 

For those that are curious, if I use the value of .15 C/W for my SI-120 (I'm not using a super fast, high CFM fan), then my Temp Delta should be more close to 27.8 (.15 * 185.625), giving me a reported temp of ~48*C at load; a full 10*C above what's currently reported.

 

Chum

Share this post


Link to post
Share on other sites

In an attempt to bring this conversation full circle and gather more data...

 

FOR ALL: Please continue to provide information based upon the following template:

 

What choice did you make in the poll above? I believe mine is 10-20C low.

Motherboard: SLI-DR Expert

Mobo Revision: ???

Mobo BIOS version: 12/7 Official

CPU: Opteron 170: OSA170DAA6CD (CCBWE 0546XPMW)

CPU ID: 1367284K50103

Idle Temp: 22C

Load Temp: 31C

Cooling: Air (SI-120)

 

also, if you would, Add this to it as well:

 

Avg. Ambient Temp: 21C

Share this post


Link to post
Share on other sites
With that said, I think your input has been very valuable to the thread and by no means want you to "shut up". :) Your thermal resistance posts motivated me to calculate the above. I guess I just find myself frustrated when someone tries to tell me that there is no problem, or that "it hasn't been proven" - when I can clearly see first hand that there is. Ultimately, I'm hoping there's a simple fix that doesn't require replacing the board.

 

 

Chum

 

No no! I don't deny there is an issue! All I'm saying is that I'm staying objective at this point on the subject and I try to point out that there is reasonable doubt, nothing more. I am curious to see what the outcome will be, but I would like to see some more hard facts (like probe measurements). The theory on thermal resistance is good and solid but leaves a few loopholes. I am inclined to believe that there is an issue, simply because this thread exists (and apparently others elsewhere).

 

I neither deny nor acknowledge the issue, but I am very curious. Also this is one of the more intelligent threads in this forum and I enjoy the debate.

 

Yes, and keep the data and the input coming.

Share this post


Link to post
Share on other sites

What choice did you make in the poll above? I believe mine is accurate.

Motherboard: SLI-DR Expert

Mobo Revision: ???

Mobo BIOS version: 12/7 Official

CPU: Athlon 64 x2 4400

CPU ID: ??? I've got a pic of the CPU somewhere, not sure where...

Idle Temp: 30C

Load Temp: 40C

Cooling: Zalman Fatal1ty Thermal resistance - 0.17C/W

 

I'm at 2250MHz currently and 1.42V with a delta of 20C (ambient room = 20.5C, Load = 40-41C). Using this to calculate the thermal resistance gives a little better than 0.16C/W. This is pretty accurate because on Zalman's site it rates this cooler @ 0.17C/W, plus AS5 is good and burned in, and I have excellent air flow in a big case.

 

This is a good thing... I was worried about this issue being a problem for me, but not anymore. I'll recalculate when my new PSU gets here and I start doing some major overclocking.

Share this post


Link to post
Share on other sites
What choice did you make in the poll above? I believe mine is accurate.

Motherboard: SLI-DR Expert

Mobo Revision: ???

Mobo BIOS version: 12/7 Official

CPU: Athlon 64 x2 4400

CPU ID: ??? I've got a pic of the CPU somewhere, not sure where...

Idle Temp: 30C

Load Temp: 40C

Cooling: Zalman Fatal1ty Thermal resistance - 0.17C/W

 

I'm at 2250MHz currently and 1.42V with a delta of 20C (ambient room = 20.5C, Load = 40-41C). Using this to calculate the thermal resistance gives a little better than 0.16C/W. This is pretty accurate because on Zalman's site it rates this cooler @ 0.17C/W, plus AS5 is good and burned in, and I have excellent air flow in a big case.

 

This is a good thing... I was worried about this issue being a problem for me, but not anymore. I'll recalculate when my new PSU gets here and I start doing some major overclocking.

well done and congrats on the accurate readings!

Share this post


Link to post
Share on other sites

Just a quick question - we've been using ambient room temperature when calcuating the delta. Shouldn't we really be using ambient case temp? I realize the DFI boards don't have an ambient temp sensor, but it's just a thought.

 

Anyone have a 3rd party sensor that they use inside their closed case?

 

Chum

Share this post


Link to post
Share on other sites
Just a quick question - we've been using ambient room temperature when calcuating the delta. Shouldn't we really be using ambient case temp? I realize the DFI boards don't have an ambient temp sensor, but it's just a thought.

 

Anyone have a 3rd party sensor that they use inside their closed case?

 

Chum

no, the actual formula uses the ambient room temp.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×