"Core Temp" - A program for accurate K8 temperature monitoring

[pcdoc1]

Nice app….. I’ve never been much for running “Topless”. I usually only pick up a few degrees anyway, nothing that could really help an OC….

Testing with this app I found I was running:

Core #1 @ 41 idle and 58 load
Core #2 @ 36 idle and 48 load

Finally a tool that can detect a problem…. Needless to say I popped the top.

Results:

Core #1 33 idle and 45 load
Core #2 33 idle and 45 load

Pretty damn cool! Guess it’s time to bump the clock!

Thanks again - Great Tool.......

[The Coolest]

So basicly, all the other programs to date have been relying on motherboard sensors? I haven't really researched into that. I was just assuming support for reading the internal sensor was implemented shortly after the CPUs started to come equiped with them.

But yea, if thats the case, then obviously the internal sensor will be much more accurate, probably as close as close to a truthful reading that we can get, depending on the quality of the sensors themselves (which I assume are within 1-2 degrees maxium given todays technology).

You should definately contact the creators of the other popular monitor programs and share your findings so they can implement the same into their own creations. Not that your program isn't adequate, but it would be a shame not to spead the information. I assume that commercial profit from this isn't your goal anyway =)

-Stigma

Not quite.
The mobos read the temp using external sensors connected to the internal diode inside the CPU. So the accuracy of the reading will really depend on the quality of the sensor and on the quality of the signal which that little diode is putting out, going outside the CPU, into the mobo's circuitry and into the external sensor located on the board.

hey The Coolest...is there anyway that a naked core vs an IHS covered one would result in a false reading. i have a 3/8" waterchill loop setup on one test rig and on my 3200+ AM2 I would load around 45-50C @ 2.8GHz depending on ambient temp.

a few minutes ago i popped the top on the little sucker and put it back under water...almost 20C drop in load temps was the result. i am not one to trust such a wide swing and was curious to know if you think a naked core would effect the temp readings that much?

Are you speaking of a 20C drop via Core Temp? If so its quite possible I guess, if the IHS had really poor contact with the core. Also IHS represents another layer of material interfering the HSF/Waterblock to take the heat away from the CPU core as efficiently as possible.

Nice app….. I’ve never been much for running “Topless”. I usually only pick up a few degrees anyway, nothing that could really help an OC….

Testing with this app I found I was running:

Core #1 @ 41 idle and 58 load
Core #2 @ 36 idle and 48 load

Finally a tool that can detect a problem…. Needless to say I popped the top.

Results:

Core #1 33 idle and 45 load
Core #2 33 idle and 45 load

Pretty damn cool! Guess it’s time to bump the clock!

Thanks again - Great Tool.......

That is some interesting info. I wonder if this kind of result will be consistent or not.

[3oh6]

Are you speaking of a 20C drop via Core Temp? If so its quite possible I guess, if the IHS had really poor contact with the core. Also IHS represents another layer of material interfering the HSF/Waterblock to take the heat away from the CPU core as efficiently as possible.

yeah, 20C gap between the Core Temp readings. the contact looked good, that is what stumped me. i thought perhaps there was something else going on because 20C seems waaaay to much for just the IHS coming off unless of course it was dry under there but it wasn't. the thermal paste looked in really good shape. thanx for the response.

[NickS]

My old Venice looked to have great contact but I still got 3-5*C drop!

Nick

[lross78550]

I noticed a big difference in the temps on my a8n32-sli and fx-57 the asus temp program and speedfan both report my cpu temp as being -12c at idle in windows. your program reports the core temp as being -28c (im running a prometeia mach II) I hope you program is right as being i like your numbers better I really think you should have the title of your thread changed now that you have it working on processors other than the Rev F chips. and i think you should have it stickied on one of the forum topics, finally there is a way for people with fairly current hardware to get accurate cpu temps. As soon as my jinuv2000 ss phase shows up and i build my fx-62 system i will be downloading all your software and sending you a paypal. Very nice work, Thanks a ton. Please keep working on it. Give us a speedfan type app but with your accurate measurements, logging and gui enhancements are all thats lacking.
Thanks again and my paypal is on the way
Thanks
Lee

[dinos22]

yeah, 20C gap between the Core Temp readings. the contact looked good, that is what stumped me. i thought perhaps there was something else going on because 20C seems waaaay to much for just the IHS coming off unless of course it was dry under there but it wasn't. the thermal paste looked in really good shape. thanx for the response.

i had a 20C dropped on one of my Opti CPUs after popping the hood

[sephiroth8748]

I have a 3 degrees difference

[3oh6]

i finally have some screenshots of Core Temp in action, catching the AMD IHS napping. notice how the idle temperature at a lower clock with less volts is higher than the load temp with more volts and clock. if i didn't do the testing myself i wouldn't believe it. no IHS makes the waterchill unit look king

290 x 10 / 1.60v set in nTune / IDLE - w/IHS max CPU on waterchill


300 x 10 / 1.65v set in nTune / LOAD - wo/IHS max CPU on waterchill


bloody hell i hate not having working programs for voltages and whatnot that are ss friendly. ntune works with the C51 for clocking and is the only way i can get past HTT290 but to take a ss with any amount of info...forget it.

[MrRevhead]

Firstly, nice proggy Coolest Many people have been waiting for something like this for a long while.

I thought I might make you aware, when running OCCT & Core Temp simultaneously, the Core Temp reading jumps straight to -49 until OCCT is exited.

Not too much of a biggie as I can just run Prime, Stress CPU etc but thought I may as well make you aware while the proggy is still in beta mode.

Best of luck

[turbotoaster]

this is the best program i have ever used

for people who are a having a weaker core, or what they belive to be weaker isnt so

i was getting 60c load on core 1 then 40c on core 2

if u run prime watch the core temps, slowly adjust the screws on you heatsink until the temps are close or the same, then u will notice your weak core will run alot better

[dinos22]

this is the best program i have ever used

for people who are a having a weaker core, or what they belive to be weaker isnt so

i was getting 60c load on core 1 then 40c on core 2

if u run prime watch the core temps, slowly adjust the screws on you heatsink until the temps are close or the same, then u will notice your weak core will run alot better

nice idea

[Okda]

[alpha0ne]

I cant wait to try this great proggy on my Conroe

[Jaco]

I had 20°C dif6ference between core0 & core1 !
My waterblock barely made contact with the core (it's a naked chip)
I tried to correct this while the system was runnning , very stupid of me.
I think I killed my cpu

But anyway , cool program

[tzitzos]

for my opty 170 this program shows only -1C

i use smartguardian for the temp!!

thx

[The Coolest]

for my opty 170 this program shows only -1C

i use smartguardian for the temp!!

thx

Can you post a screenshot and tell us what kind of cooling you use?

[Pyr0]

I would also like to see pics of idle, load core0/idle core1, idle core0/load core1 and load both cores.

Core0 loaded:



Core1 loaded:

[Gak]

Still no one answered if this program readings are to be trustworthy...?

[Chicco85]

for my opty 170 this program shows only -1C

i use smartguardian for the temp!!

thx

Lol Giorgio....

My core 2 in 10 °C hotter than core 1 in idle... Next week i will set up more tests.

[The Coolest]

Still no one answered if this program readings are to be trustworthy...?

The accuracy of the readings are as accurate as the Digital Thermal Sensor (or DTS) is.

[Gak]

So why haven't mb manufacturers used this fuction? Temp readings would be a lot more accurate....

[MikalCarbine]

Pyr0: YGPM
This feature was undocumented in any documentation prior to the official implementation in RevF chips. A user reported that this could also work on his RevE chip, and after some digging around and testing it appears to work on all the chips since the early Rev C0.

Does this mean there will be Rev E support soon?

[ozzimark]

Does this mean there will be Rev E support soon?

if i'm not mistaken, there already is

[ineedaname]

I tried it out on my opty 165 seems to work great altho its a bit awkward cuz my core 0 shows up as core 2 and my core 1 shows up as core 1. Or at least i think it is. When i load core 0, core 2 goes up more but when i load up both cores core 2 has a lot lower temps but fails earlier, weird.
Core 1 would go up to like 60C but keep going while core 2 fails at about 50C. Also i noticed the program sometimes jumps to some crazy number then goes back to normal.

[Stigma]

Question:

How low temps would this sensor be able to read from a conroe? I will be phase-cooling one soon to very low (far sub zero) temps.

-Stigma