Jump to content
Sign in to follow this  
The Coolest

"Core Temp" - monitor temps on K8s and Core/2 CPUs (updated

Recommended Posts

Thinking was that you need to add +6 to SG or MBM5, or speedfan. Like I said, I am not sure if that is valid.

It' hard to know for sure. I haven't been able to find any info one way or the other. If the offset feature is being used correctly the value should be read from the register and added to the value the onboard sensor is displaying.

 

Here is a thermal picture of a P4 processor from SilentPCReview. To a lesser extent it also applies to 939 processors. It's easy to see why programs such as CoreTemp read higher then programs using the thermal diode.

 

p4_heat.jpg

 

Simulated temperature plot of a P4 processor. The arrow on the upper left marks the location of the P4's thermal diode, while the arrow on the lower right is the hottest portion of the die. The image, many times larger than the actual size of a P4 core, is from Differentiating PCs in a 'Toaster World' by Robin Getz of Analog Devices, published in the April 2002 edition of the Intel Developer Update Magazine.

 

ProCooling has a graph of temperatures running various stress programs using a Maxim 6655EVSYS diode evaluation system. The graph contains a lot of information.

 

ph-dt-deltats.jpg

Share this post


Link to post
Share on other sites

Dual priming and artifact scanner:

 

dualprimeandartifactscanner.JPG

 

 

Idle:

 

idle.JPG

 

 

It seems that the first time I ran artifact scanner I forgot to load the o/c profile, so my total delta has increased this time to 7 C on SG. Deltas on the new program however are huge.

Share this post


Link to post
Share on other sites

on my optie 180 (updating sig as soon as i am done) coretemp was showing 5- 15 degrees difference between core1 and core2... i believed it and stripped the 180 and bam primable at 3012mhz and now only a few degrees (3 at tops) difference if any

 

wish i took screenies :mad:

Share this post


Link to post
Share on other sites

VVJ has released a new version of CBID - For those of you that are interested in it's OFFSET Reading.

It's reporting all the same info on my 146 and still does not work in Vista.

Central Brain Identifier 8.2.0.0 Build 0722

Central Brain Identifier is designed to provide detailed recognition and obtaining the most complete extended information for all AMD processors. It recognizes more than 40 various models of the AMD processors. It shows not only the general information about the processor, but also allows to determine its important technical features, such as processor core, core revision, a nominal processor frequency, the date of announce, an OPN number, the components of the PowerNow! Technology and also calculates an XP-Rating for the processors having a Model Number.

The benefits and unique advantages of CBId are the following:

 

- Complete and precise identification of the AMD processors.

- P-rating calculation and OPN determination.

- The real-time CPU frequency, voltage and temperature monitoring.

- An entering the processor the low power state of Stop Grant.

- On the fly FID transition for mobile AMD processors.

- Performance state tables searching and decoding.

- DRAM timing configuration of a huge number of chipsets.

- An ability to change the feature bits of the processor.

- Quick CPU name string modification.

FSB/DRAM frequency control for the nForce2 chipset on the fly.

- An advanced tuning of VIA V-Link bus speed parameters.

- An enhanced set of tweaks to optimize the performance of DRAM with the nForce2 chipset.

 

http://cbid.amdclub.ru/files/cbid82b.zip

 

@ Neo, big thanks. Now I understand fully what you are saying.

 

@ Praz, nice post :) - I Just lost many hours reading all the links.

Wound up at www.amd.com readings the pdfs for ages (yes, I know, I need to get a life)

 

The "Pro Cooling" article and screenshot explained why Super Pi and other programs seemed to heat up the CPU faster. I was only running them for a couple of minutes.

I'd like to try duplicating the tests with our sensor and see if there is any difference.

 

I found the pic of the P4 CPU Diode location and Max heat from SPCR very interesting.

I found it to be one of the Best articles I have read in a long time.

And the links and discussion in their forums on the article provided even more extremely informative information. Got the juices flowing so to speak even if it wasn't apples-to-apples with the processors we are discussing here.

I'd really like to find the time to try and duplicate this procedure as well.

 

p4_heat.jpg

 

The diode is located in the "coldest" part not the hottest.

But it makes sense as it's probably the easiest place to have it and allow for pin outs etc.

It also shows why placing a thermal probe against the side of the IHS "lid" is not going to work.

Even if it the CPU was "lidless", thermal probe contact method and placement would make a huge difference.

 

It lead me to the AMD Functional Data Sheet for the 939 Package (Pub 31411, Rev 3.03, dated May 2005) and the AMD Athlon 64 Processor Power and Thermal Data Sheet (Pub 30430, Rev 3.51, from March 2006) among a host of others.

 

While I still have no idea of where the diode is located in a 939 package (and whether that is near or far away from the actual heat source) it does seem to seem to make sense that it would be located close to the edge/outside of the chip as well.

 

Looking at the pin out of a 939 CPU I came across the following locations for CPU Temerature Dode "Anode, Cathode, and Thermal Trip"

 

thermalpinoutxz7.th.jpg

 

While I am not sure the actual thermal diode is near the pin out I would see the logic in having it located close to them.

Or are these just pin outs? Is the temp diode off core (ie; a seperate component) or something built into the "die"?

If off core, can anyone that is "lidless" identify it with a pic?

I could find no info on a Dual Core but there are plenty on N/C pins on the 939 in the vicinity. Baffled>

 

Made me start thinking about heat sinks, heat transfer, types of cooling, thermal compression, thermal material, application of thermal compounds, and all the other variables that enter the equation.

 

amdthermalgreasevo8.th.jpg

 

Since most of us are generally using a little less than a single grain of rice and allow the HSF to compress the material (Lord only knows how well the IHS transfers the heat, much less the application of AMD's thermal compound below the IHS and between it and the actual chip) it would seem that attaching a thermal probe to the IHS is "luck-of-the-draw" at best.

 

Combining this with the accuracy of the thermal diode itself I wonder if there is a way to "accurately" measure the temp.

 

Couple of other quotes from the SPCR article in Praz's post.

 

1. Most PC users do not need to worry about precise CPU temps. It's an odd disclaimer for an article about how to correct CPU temps, but the oft-quoted axiom of "If your PC is stable, your temps are fine" is true.

 

2. What this article describes as "accurate" is a matter of debate. This methodology attempts to remove the motherboard-induced variations to the CPU on-die thermal diode readings. However, the accuracy of the temperature output from the CPU thermal diode is a completely different issue. The P4 is known to report temperatures well below the hottest portions of the core, and to a lesser extant this is true of the AMD processors as well. The nice aspect of reading directly from the CPU diode is that at least you know that everyone else's readings are equally inaccurate. (On a philosophical level, if everyone is equally wrong, does that make everyone right?)

 

3. The calibration described here only works for a specific CPU in a specific motherboard. Swap the CPU after you calibrate the thermal reporting system, and it won't be accurate anymore. This is because there is enough variance between different samples of the same model processor that their heat output will not be the same. Change the motherboard, and it won't be accurate, again because of sample variances. Changing to a different model virtually guarantees a different temp result. Even removing and replacing the CPU from the socket may effect the results, due to the potential for altered resistance between the thermal diode output pins and the motherboard socket.

 

On a last note, I'm still amazed that at Idle the three OSes are reading slightly different temps on my rig.

32 in XP32, 34 in x64 Edition, 36 in Vista x64.

More processes running?

Different programs or the order they load at startup?

Effects of the OS on the software monitoring programs I am using?

Just the normal margin or error?

 

I need to do some more testing and I'll have to stop bt the store for supplies :D

 

shitcreeksd6.th.png

Share this post


Link to post
Share on other sites

Isn't this fun? The further we dig instead of answers there's more questions and confusion. There are numerous white papers that either state outright or can be inferred that the accuracy of the thermal diode is ±10C-14C. I haven't been able to find any accuracy information pertaining to the thermal sensors. Even if they were extremely accurate there would still have to be a small margin of error. The only time I ever see them mentioned is in reference to Thermtrip and its related registers.

Share this post


Link to post
Share on other sites
...On a last note, I'm still amazed that at Idle the three OSes are reading slightly different temps on my rig.

32 in XP32, 34 in x64 Edition, 36 in Vista x64.

 

More processes running?...

 

That makes the most sense to me.

Share this post


Link to post
Share on other sites
Isn't this fun? The further we dig instead of answers there's more questions and confusion.

 

ROFLMAO - hehehe, More like "well that's funny" (as in scratching of head or pulling out of hair).

 

Certainly explains why threads like "My Expert mobo temps are wrong" never have a resolultion.

 

P.S. I got a "quantity discount" at the store so I picked up Paddles for everyone. :D

Share this post


Link to post
Share on other sites
I need to do some more testing and I'll have to stop bt the store for supplies :D

 

shitcreeksd6.th.png

 

That's awesome!!!

 

P.S. I got a "quantity discount" at the store so I picked up Paddles for everyone. :D

 

Can you next day air mine? :rolleyes:

Share this post


Link to post
Share on other sites

I was reading this thread and I just wanted to make a comment.

As it is with any real digital reading, the program here can not be reporting inaccurate temps, as it simply reports the temperature read from the DTS. The DTS on the other hand may be inaccurateor not calibrated properly.

 

I've also thought about adding offset to the program, but it didn't go too well, some chips reported offsets in the 10-20C range, and then the temperature reported by the program would be simply silly, so I'm guessing that the offset is only meant for the external CPU sensor.

Share this post


Link to post
Share on other sites

Looking at the various white papers it appears that the offset is for the external sensor only. If the bios/external sensor is configured according to AMD specs the temperature reported by the bios or programs such as MBM should already include the offset. But like almost everything else related to 939 temperature readings this is next to impossible to verify.

 

It appears that programs such as yours , using the DTS (not to be confused with the thermal diode), report the correct temperature for SC processors and core 2 of DC processors. I'm still looking for documentation from AMD that will indicate the accuracy of the DTS and why the core 1 DTS of DC processors seems to not have been calibrated.

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  

×