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

[The Coolest]

Download Core Temp v0.98

 

This is a new program that lets you monitor Intel's "Core", "Core 2" and all AMD K8 and K10 chips' die temperature. The temperature readings are very accurate as the data is collected from a Digital Thermal Sensor (or DTS) which is located in each individual processing core, near the hottest part.

For more information click here

 

There is also an article posted about Core Temp and a little background on standard thermal sensors - Article

Please help out Core Temp popularity by Digging it.

 

If you find any bugs, please post it in the "Bugs" section ASAP!

Any suggestions on improvements or ideas are welcome in the "Development" section.

Note 1: Please read the ReadMe!.txt file for some useful tips.

 

What's new:

 

Version 0.98 - 23rd April, 2008

 

- Fix: Core Temp sometimes crashes when another program tries to access the log file. - Still needs further testing.

 

- Add: Show load percentage per core.

- Add: Adjustable Tj.Max value via *.INI file.

- Add: Option to start Core Temp with Windows.

- Add: Support new Intel Mobile processors

 

- Change: Maximum logging interval is now 99999ms instead of 9999ms.

- Change: Log file layout. Shows current temperature, highlow temperature, core load, CPU speed.

 

Older versions

[yfital]

ya, the prog has been running around here for a while, its nice, but i find it hard to believe its really true, cpu's cores temps can change to fast and into too much differenet numbers (e.g core 1 25c, core 2, 36)

[The Coolest]

Hello guys.

First of all I want to make something clear:

This feature was unofficial until the RevF AM2 CPUs, so I wouldn't take all readings you get from older chips as granted, as the DTS* could be off or not callibrated properly, I've seen a lot of dual core chips where the temp difference between the cores would be like 12C, and one of them would report sub-ambient temps

 

I think that in the majority of cases it is pretty much accurate, though.

 

Regarding how it calculates temperature on AMDs, Intel complicated things quite a bit with their chips, AMD kept it pretty straight forward. Just read a value from a register, substruct 49 from that value and you have the temperature in C.

The temperature can range between -49 and 206C, at least in Rev F. I don't know if in earlier chips its the same, but I'm assuming it is.

 

Another thing I noticed, sometimes, in sub-zero temperatures it'd show much lower temps than what MBM5 and the likes report, I don't know which of the two programs is more accurate, but it is possible that the DTS loses its accuracy when temps get really low.

 

*DTS = Digital Thermal Sensor.

 

This I hope answers some questions that arise, when starting to use the program

[Angry_Games]

most of us prefer MBM5 that soundx98 has put countless hundreds of hours into already, or ITE Smartguardian which is based on the ITE monitoring chip that is integrated into the motherboard.

 

Since this program only measure's cpu temps, it won't be too popular here because the two programs I listed already do much more (and this program isn't going to be so accurate that it makes MBM and ITE obsolete, and besides MBM and ITE are pretty accurate for the most part).

 

maybe if this program did more than just measure cpu temps (since there are already a plethora of programs that do that).

 

Not trying to knock ya, but just letting you know that these kinds of programs come and go and others like MBM5 have outlived their support to the point that others have picked it up like soundx98 and support it and do their best ot make sure it does the job for all newer systems.

 

Maybe you can talk the programmers of this into adding a bit more function as our users are interested in more than just temps, which is why MBM5 and ITE is so popular (and as I said, this programs is not going to be that much more accurate than ITE and MBM because no software, even reading straight from BIOS-level sensors, is going to be 100% accurate)

[mrzeld]

(and as I said, this programs is not going to be that much more accurate than ITE and MBM because no software, even reading straight from BIOS-level sensors, is going to be 100% accurate)

the thing that makes this program different from the other temp monitoring programs is that it seems to be reading the temp completely differently than the others. the creator of this program is not trying to create a program that competes with MBM5. he just created a program that seems to display the temp from 'previously hidden' locations within the CPU (more accuratly, "special register in the CPU's NB"). now we are just trying to figure out if these temps are correct (as the author said, the way he is computing temps is a newly documented feature for Rev. F chips, but it also seems to work for earlier chips too, but is it accurate??).

 

previously people have complained that the Expert MB gives inaccurate (usually low by 10C) temps. but using this program, it seems to be giving the correct temps. and also this program gives the temp of EACH core, not just one temp.

[ollietwinam]

Doesn't Everest already measure Core temps?

[soundx98]

yes, as does Nextsensor.

If you can give me a couple of minutes I trying to get some screenshots and thoughts together.

[mrzeld]

here, i will repost what i wrote in another thread...

hmmm. ok. very odd. here is my experience... using MBM5 and every other temp reading program out there, my idle temps are around 26C (ambient being around 19-20C). and my load temps being around 36-38C. when i first got my DFI expert MB, there was a thread where people swore up and down that the temps readings were off by about 10C (some people said more, some said the temps were right). so.... in MBM5 i just add +10C to my CPU temp reading. therefore i assume that my system runs under load at about 46-48C (it is always running at load since i run SETI@home except for one hour a day).

 

now i start up this Core Temp program. nice thing about it, it seems to display the core temps of BOTH of my cores. nice. but the temps it is displaying are about the same as the MBM5 temps with the +10C added. Core Temp is reporting my Core #1 at 46C and Core #2 at 44C. hmmmmmmmmm.... interesting. is it correct, who knows. but it seems about right to me. why would this program be reading correct temps and every other program out there (MBM5, smartguardian, SpeedFan, Everest) be giving me temps that are much cooler (but are all the same, which means they are all reading the temp from the same location)??????

 

idle temps using the Core Temp program... 31C for both cores roughly. whereas MBM5 is reporting 26C (or 36C if you do the +10C).

 

here are some screenshots of temp readings from Core Temp versus MBM5 (when viewing the MBM5 screenshots, please remember that i specifically have MBM5 add 10C to my CPU temps)...

- Idle screenshot

- Load screenshot

[soundx98]

I invited TheCoolest to start his own thread here in the appropriate section and post his program "Core Temp" and thoughts he had in it. It's been mentioned in 3 or 4 posts here at DFI-street and it just made sense to have it in the right section so that members could ask questions and be able to find it using the search function.

 

I find it interesting because it is reporting the same thing as the latest Everest (3.01) and Nextsensor (RC4).

There is an internal temperature Diode on K8 processors and it is being measured by Core Temp, Everest, and Next Sensor and reporting the same values.

It is NOT the same as the CPU Sensor that is read by ITE SG and MBM5.

CPU Diode responds (polls) much faster than SG or MBM5 which check their CPU sensor roughly every 5 seconds.

What I find unique about the Core Temp program is that it seems to work with board sensors and processors that SG and MBM5 do not. It works on a lot of boards and Intel Processors. I am not sure if Tamas (AIDA 32 dude) at Everest or VVJ (CBID) at NextSensor use the same method and can read Intel Processors (I am sure we will find out soon enough).

 

So I'v done the 1st screenshot at Idle with my rig in XP Pro 32 bit comparing;

Core Temp by TheCoolest (link to latest version in 1st post in this thread)

ITE Smart Guardian (latest version from www.dfi.com.tw

MBM5 by Alex Van Kaam (with DFI config file - link to thread below my sig)

NextSensor RC4 by Vitaliy Jungle http://cbid.amdclub.ru/ (make sure you select DFI LanParty under "Control" - I supplied this info to VVJ via a "dump" from this board.

Everest Ultimate 3.01 (Trial available at www.lavalys.com - link below my sig)

 

You will notice that the temperatures and voltages are virtually identical when you compare apples-to-apples. I've tried to color code each.

 

 

The second screenshot is the same but while running an 8M SuperPi (Mod 1.5) to "heat up the processor (test had run for about 2.5 minutes)

 

Again when you compare apples-to-apples you will find the temps and voltages to be the same (or at least close enouigh for government work as they say).

 

What you will notice is that the delta/change/difference between the CPU Core reading and CPU Diode Temp reading is much higher. Makes sense it should be be higher since it's an internal diode.

 

 

Now in talking to some other nerds (fugu very much Praz, Pyr0, Vitaliy, and ExRoadie) what we did notice is that with Dual Cores Core Temp was reporting a pretty substantial difference tween the two cores (Sometimes a very large difference).

Praz has reported that it is reporting temps below ambient (WTF) on one of his cores.

The VID reported by Core Temp is the default not the Vcore + Special VID%.

 

TheCoolest has made it very clear that this program is in the Beta Stage, that it was designed for "F" chips, that he is as perplexed as the rest of us on the core differences, and that there may be errors at sub-zero temps.

He's been very helpful and responded quickly to requests for a x64 version and the program is working just fine in Vista for me as well.

 

let's look at memory usage at Idle

(ignore the reading for NextSensor as the high usage is caused by the advertising gif in RC4, it is usually about the same usage as the other programs 3,000K-5,000K)

 

 

Each of the programs can offer you useful information.

Saying that one is better than the others is being naive.

But there are some features that some may like and others consider unnecessary.

ITE SG couldn't be easier to setup LOL - it comes with your board - easy to calibrate with a Digital Voltmemter. Joop's got some excellent info in his thread's here.

MBM5 was just way ahead of it's time (still the best OSD) and with the work done by W1zzard (ATITool) and _xhp_ (NV Plug-ins) it's like a familiar old friend to many of us (and wearing a sexy new thong).

Everest 3.01 can give you about any information you want including benchmarks - yes, you pay for it but it's just excellent in so many areas.

NextSensor is very appealing and delivers what VVJ promised and more.

CoreTemp is a great program that looks into areas that others are not able to.

Hat's off to TheCoolest for all his efforts.

 

All of these guys bust their butt to help the enthusiast community and if they are lucky they may get a donation for their efforts. (Pretty friggin rare).

They get NO help from AMD or Intel or any motherboard or GPU supplier.

I got a ton of respect for these dudes

 

(Lord it's tough as you get old, I ramble on like my Grandpa, )

[soundx98]

Links to some screenshots of Core Temp using Vista & x64 Edition OSes and comparing software utilities

 

http://www.dfi-street.com/forum/showpost.p...&postcount=1505

 

http://www.dfi-street.com/forum/showpost.p...&postcount=1506

 

Some pics from Praz with his Dual Core, comparing programs

 

http://www.dfi-street.com/forum/showpost.p...&postcount=1507

 

http://www.dfi-street.com/forum/showpost.p...&postcount=1509

 

Feel free to to provide screenshot comparisons of your Dual Cores at Idle and at load

Forum Rules - No pics larger than 800x600

use a link or Image Shack

 

All input is appreciated

[zarathrustra]

My experience with this utility and the w/c temps with my expert, in direct comparison to ITE:

 

Both programs read within a dergree or two at idle, with 'Core Temp' being the higher. (21/22 at the moment in a cold room)

 

Under Prime load, ITE increses only a few degrees (23-24), while Core Temp jumps up almost instantly to 31-32 degrees.

 

The very quick jump in temps (and decrease when stopping Prime) is similar to the way GPU temps can jump/drop 10 degrees in a second or two, which makes it seem plausable to me.

 

I know the 'Expert temp bug' is no big deal, but if this new utility is accurate with the older A64 chips on this board, then it is certainly a welcome addition. Many Thanks.

[yfital]

well, im currently on stock, w.c... low level one...

mbm gives me 36-7 c idle, 42 load (2 prime)...

core temps are 27/30 (1,2 core) idle and 40-41 load...

i have seem that the temps of each core (usually not in full idle) can really very... but it does seem to be pretty accurate...

i just means my expert is doing -8c more or less (im on +10c now...) so the temps kinda fit together...might be intereseting if the prog can be incoporated into something else... or the opposite way...

maybe a more "relaxed" timer for the temp update will make it a bit better and will give a better idle average...

best of luck for the author..