Jump to content

How accurate is CoreTemp, the program?


27 replies to this topic

#1 Quinn Allard

    Member

  • Veteran Founder
  • 175 posts
  • LocationUSA

Posted 09 August 2012 - 01:31 PM

Heya, I downloaded CoreTemp to keep an eye on my new shiny FX8150BE. According to it, my CPU idles around 15c-19c and peaks playing Diablo 3, StarCraft2, and MWO around 44c-48c. It peaked once at 55c when I was playing Diablo 3 and installing StarCraft2 at the same time. Are these temps good? From what I've read 62c is the max safe temp, and according to CT its well below that. I have a butt load of fans in this rig, with an AMD CPU/Heatsink.



Sys. Specs:
AMD FX8150BE OC 4.4Mhz
Radeon Sapphire 6950 DiRT3 2GB
ASUS Sabertooth 990FX
Mushkin 8GB DDR3
Kingston 4GB DDR3 x 2

#2 Greyrook

    Member

  • Elite Founder
  • 1,302 posts

Posted 09 August 2012 - 01:34 PM

Do you live in a particularly cold place? 15C is really cold and unless you have a peltier heatsink it shouldn't be going below ambient temperature. Or else, I would say the program is reporting wrong. CPUz is a popular free program that most people use for this purpose
http://www.cpuid.com...#directiontouse

edit: sorry, I got that mixed up. http://www.cpuid.com.../hwmonitor.html is the monitor program, CPU-Z is the performance monitor

PS: generally 90C is the max heat before auto shutdown (I'm not sure if that's the same for AMD parts, I use Intel), 50-60 is better than average for full load on a good performance system with air cooling. I'd pay more attention to your graphics card temps, though.

Edited by Greyrook, 09 August 2012 - 01:39 PM.


#3 Odins Fist

    Member

  • Members
  • PipPipPipPipPipPipPipPipPip
  • 2,891 posts
  • LocationThe North

Posted 09 August 2012 - 01:37 PM

As always no matter what "Monitoring" software you run, make sure it is the latest, and that it supports your chipset, CPU, and all other components if possible.
Temp readings can be "plus" or Minus" a degree or so off with most software. Also your case/airflow, quality and size of power supply can affect temps, and also underpowering a system will require that your system pulls more current to supply the requirements of the system usage thus creating more heat..
.
Nice choice on your motherboard BTW, I run a Sabertooth 990FX as well since revision #1 and have had "ZERO" issues...

Edited by Odins Fist, 09 August 2012 - 01:39 PM.


#4 JuiceCaboose

    Member

  • Legendary Founder
  • Phoenix Overlord
  • Phoenix Overlord
  • 865 posts
  • LocationNorthern Michigan

Posted 09 August 2012 - 01:40 PM

I use CoreTemp all the time, seems to be pretty accurate, it's at least comparable to lmsensors in Linux which is pretty much as good as you can get. 15c isn't that crazy low, but I would definitely see if there's an ambient air temp sensor in your computer and compare against that. Mine insists that ambient temp is 0c, which obviously isn't correct. If you can find a guide as to which sensor address does what, you should be able to figure out which one is where, and what common offset values are.

#5 DV McKenna

    Member

  • Legendary Founder
  • 3,425 posts
  • LocationUnited Kingdom

Posted 09 August 2012 - 02:10 PM

As always software changes.

In the past core temp was well known for being unreliable that mostly changed over time, i still use HW monitor linked in post 2 tbh.

More accurate, shows more detail and more components than just your CPU.

#6 Quinn Allard

    Member

  • Veteran Founder
  • 175 posts
  • LocationUSA

Posted 09 August 2012 - 02:17 PM

View PostGreyrook, on 09 August 2012 - 01:34 PM, said:

Do you live in a particularly cold place? 15C is really cold and unless you have a peltier heatsink it shouldn't be going below ambient temperature. Or else, I would say the program is reporting wrong. CPUz is a popular free program that most people use for this purpose
http://www.cpuid.com...#directiontouse

edit: sorry, I got that mixed up. http://www.cpuid.com.../hwmonitor.html is the monitor program, CPU-Z is the performance monitor

PS: generally 90C is the max heat before auto shutdown (I'm not sure if that's the same for AMD parts, I use Intel), 50-60 is better than average for full load on a good performance system with air cooling. I'd pay more attention to your graphics card temps, though.


Thanks to my wife its a chilly 70-72 in here. I have CPU Z but I dont see where it reports temps. Playing StarCraft2 on Ultra for about 45min-1hr the CPU was at 53c, the GPU at 51c. Just want to make sure I wasnt getting hot. I have a monster full tower, with alot of fans (even one we rigged, angled, to blow directly onto the backside of the GPU and onto the bottom of the CPU Heatsink. Its chilly in the case, the air coming out of the vents is never more than ambient.

#7 Quinn Allard

    Member

  • Veteran Founder
  • 175 posts
  • LocationUSA

Posted 09 August 2012 - 02:21 PM

View PostJuiceCaboose, on 09 August 2012 - 01:40 PM, said:

I use CoreTemp all the time, seems to be pretty accurate, it's at least comparable to lmsensors in Linux which is pretty much as good as you can get. 15c isn't that crazy low, but I would definitely see if there's an ambient air temp sensor in your computer and compare against that. Mine insists that ambient temp is 0c, which obviously isn't correct. If you can find a guide as to which sensor address does what, you should be able to figure out which one is where, and what common offset values are.



I remember the guy I built it with (Im a noob) say that there was a temp sensor under the MB, how do I access it?

#8 SakuranoSenshi

    Member

  • Legendary Founder
  • 1,255 posts
  • LocationSan Antonio, Texas

Posted 09 August 2012 - 02:29 PM

I hate to break it to you but it mostly comes down to the hardware. In this case, the sensors that the software is taking its data from. They could be very accurate, through placement and being a quality part or they could be well off, due to being not especially sensitive and placed in a less than ideal place.

In the case of your CPU the sensor readings are normally pretty close to 'reality', though. As noted above, if you're seeing 15 Celsius, I can only assume you've recently powered the system on and live somewhere with outdoor temperatures that regularly approach zero. Room temperature is between 20 and 25 typically, for comfort (it depends a little on the person and tends to be a bit lower in Europe a bit higher in the USA, for example), so unless you're spending a lot of power cooling it ten degrees below room temperature, that's a (relatively) cold room it's sat in.

Edit:

P.S. Even 70 Fahrenheit (I presume that's what you meant there) is 21 Celsius, it's unlikely you're really cooling six degrees (Celsius, about 10 Fahrenheit) below room temperature.

Edited by SakuranoSenshi, 09 August 2012 - 02:33 PM.


#9 Greyrook

    Member

  • Elite Founder
  • 1,302 posts

Posted 09 August 2012 - 02:29 PM

View PostQuinn Allard, on 09 August 2012 - 02:17 PM, said:


Thanks to my wife its a chilly 70-72 in here. I have CPU Z but I dont see where it reports temps. Playing StarCraft2 on Ultra for about 45min-1hr the CPU was at 53c, the GPU at 51c. Just want to make sure I wasnt getting hot. I have a monster full tower, with alot of fans (even one we rigged, angled, to blow directly onto the backside of the GPU and onto the bottom of the CPU Heatsink. Its chilly in the case, the air coming out of the vents is never more than ambient.


Honestly, you're probably fine, but if the program is saying it's at 15C when idle, that's 10F below your ambient temperature. I'm not familiar with CoreTemp, so unless it's automatically subtracting the ambient temperature and giving you a delta temp (this would be unusual and not very useful) I'd say it's not reading correctly. Unless you are using a heatsink that has a peltier or conditioning unit, the temperature of the core cannot possibly be below ambient

Also, I was mistaken when I linked to CPU-z, the same company makes another program called HWmonitor that automatically grabs the sensors from your system here's the link again: http://www.cpuid.com.../hwmonitor.html

Edited by Greyrook, 09 August 2012 - 02:32 PM.


#10 Quinn Allard

    Member

  • Veteran Founder
  • 175 posts
  • LocationUSA

Posted 09 August 2012 - 03:49 PM

Ok, downloaded HW. It says temps, under the MB tab right now are 33c-34c, under the CPU tab each core is about 15c-18c....does that sound right? Is that ok? CoreTemp is still reporting 15c-24c. GPU, according to HW, is around 38c.



Edit: Now I am playing Battlefield 3 and keeping an eye on temps, and I got a message saying "Windows has noticed your performance is slow" and that I should put the theme on Basic to improve performance.....thoughts??

EditEdit: And how can my performance be low? It didnt say what was low, or how it knew that. FX8150 @ 4.4, 16GB ram, and a 2GB GPU......impossible to be low.

Edited by Quinn Allard, 09 August 2012 - 06:01 PM.


#11 SakuranoSenshi

    Member

  • Legendary Founder
  • 1,255 posts
  • LocationSan Antonio, Texas

Posted 09 August 2012 - 04:32 PM

The thing with temps is... if you had a problem, you'd know it, long before any software told you that it was hot. You're fine. I'm not necessarily convinced your temperatures are accurate but they're not utterly unbelievable and unless you're getting thermal shutdowns there is nothing to worry about really. Semiconductors for civilian use can get pretty hot, military grade stuff can get very hot.

#12 BLOODREDSINGLE

    Member

  • Members
  • PipPipPipPipPipPip
  • 289 posts
  • Google+: Link
  • Facebook: Link
  • LocationOklahoma, USA

Posted 09 August 2012 - 07:19 PM

I use a 8120 on the STOCK cooling gear and it sits between 25c and 30c when Idle in a house thats a constant 75-80F due to 105+ temps where i live. It hits around 48-55c under load in games like BF3 and D3. Ive overclocked it aswell to 4.2Ghz and the temp might have risen a few degrees C but that was it.....

Not sure why people bag on the AMD stock coolers because the BE solutions are actually fairly beefy and do a great job asfar as i have seen.

#13 SakuranoSenshi

    Member

  • Legendary Founder
  • 1,255 posts
  • LocationSan Antonio, Texas

Posted 09 August 2012 - 07:32 PM

Stock cooling solutions are almost always perfectly fine, actually. The only reason there is press about them being inadequate is a combination of ignorance/buyer's remorse and the fact that without that press the manufacturers of supposedly superior stuff (which isn't always) would have no market. There's a world of difference between a properly engineered liquid cooling solution and a standard air-cooled heatsink with fan, of course but until that it's pretty much a case of 'a fan is a fan, a chunk of metal is a chunk of metal'. Some stock solutions are better than others, mostly because they evolve as the TDP of the current generation chips changes, otherwise you probably should just leave it alone.

(Edit: silly misspelling corrected :-P)

Edited by SakuranoSenshi, 09 August 2012 - 07:33 PM.


#14 Ouster

    Member

  • Legendary Founder
  • 47 posts
  • LocationSilicon Valley

Posted 09 August 2012 - 07:36 PM

I always used amd overdrive to check cpu temp on AMD parts. It should be able to read the most accurate temp sensor data available. Last time I have seen it give the wrong values was when I used a phase change cooler to try to overclock my computer. And that was only because the temp sensors are not really always design to be accurate at below 0. One of the nice things about AMD overdrive it shows you your core frequency as they turbo core up and down as well. CPU-z probably has this capability now a days but I have not used it recently.

Also remember with technology like turbo core the cpu will run about as hot as it can as long as there work to be done so high temp reading are probably a bit more normal then they used to be.

I not sure about actual max temp limit for current cpu’s but usually the newer the cpu process tech the low is max rated tempters are. But I wouldn't worry to much about heat damaging the parts. The safety measures in modern computers are usually quite good and your far more likely to crash out from the heat make the computer unstable then for it to get to the point where it damages the cpu from heat. Now with that said to much current can be an issue but that generally not going to happen unless you overclock your system and you really push the voltage. Strangely enough I think modern gpu max tempters limits seem to go up with each process generation I think I had one with a max temp of 105 C in the last couple of GPU cycles but I don’t know what current max temp ratings are.

If the software has access to the info most modern parts have on die or on socket temp sensor now a days so there quite accurate as long as the software knows how to read the right sensor.


Additional info: GPU-Z is grate for GPU temp monitoring you can also set it up to log GPU temps which is grate for verify if a crash may be heat related or due to something else. You can also get all kinds of extra info out of it like current and voltage as well.

Edited by Ouster, 10 August 2012 - 01:22 AM.


#15 Sephlock

    Member

  • Legendary Founder
  • The Ironclad
  • The Ironclad
  • 10,802 posts

Posted 09 August 2012 - 07:45 PM

View PostQuinn Allard, on 09 August 2012 - 02:17 PM, said:


Thanks to my wife its a chilly 70-72 in here


Have you considered counseling :)?

Alternatively, share your supercool computer building secrets with us!

#16 Greyrook

    Member

  • Elite Founder
  • 1,302 posts

Posted 09 August 2012 - 07:52 PM

View PostQuinn Allard, on 09 August 2012 - 03:49 PM, said:

Ok, downloaded HW. It says temps, under the MB tab right now are 33c-34c, under the CPU tab each core is about 15c-18c....does that sound right? Is that ok? CoreTemp is still reporting 15c-24c. GPU, according to HW, is around 38c.



Edit: Now I am playing Battlefield 3 and keeping an eye on temps, and I got a message saying "Windows has noticed your performance is slow" and that I should put the theme on Basic to improve performance.....thoughts??

EditEdit: And how can my performance be low? It didnt say what was low, or how it knew that. FX8150 @ 4.4, 16GB ram, and a 2GB GPU......impossible to be low.


It's probably the sensors in your heatsink/motherboard then. Not sure what could be done about that beyond asking for a replacement. Faulty low performance warnings ought to be ignored, I believe. You can try making sure all your drivers are up to date, don't know what else to do beyond that. As I said, you'll be fine as long as you aren't manually increasing the voltages to overclock on anything.

#17 Shivus

    Member

  • Elite Founder
  • 394 posts

Posted 09 August 2012 - 08:13 PM

View PostSakuranoSenshi, on 09 August 2012 - 07:32 PM, said:

Stock cooling solutions are almost always perfectly fine, actually. The only reason there is press about them being inadequate is a combination of ignorance/buyer's remorse and the fact that without that press the manufacturers of supposedly superior stuff (which isn't always) would have no market. There's a world of difference between a properly engineered liquid cooling solution and a standard air-cooled heatsink with fan, of course but until that it's pretty much a case of 'a fan is a fan, a chunk of metal is a chunk of metal'. Some stock solutions are better than others, mostly because they evolve as the TDP of the current generation chips changes, otherwise you probably should just leave it alone.

(Edit: silly misspelling corrected :-P)


Stock coolers are great for limited overclocking, even the FX stock cooler will let the good chips hit 5ghz without a problem. However the main benefit to aftermarket coolers is improved airflow as they pump air towards one of the output fans rather than down towards the mobo, dispersing heat throughout the case.

#18 SakuranoSenshi

    Member

  • Legendary Founder
  • 1,255 posts
  • LocationSan Antonio, Texas

Posted 09 August 2012 - 09:21 PM

Aye, well aware. I was overclocking before there were forums dedicated to it and these days I mostly roll my eyes at those when I come across them. Water-cooled rigs were interesting when they were a new idea and engineered by the person who owned the kit, nowadays bragging about that stuff is just like bragging that you bought an expensive computer in the first place; who cares?

Edit:

P.S. The maximum temperatures don't alter with generation, they're fixed by the technology, basically. Above 90 celsius is usually a bad idea. Semiconductors for commercial use are typically expected to operate between 0 and 70 C, military specs can require -40 to 105 C. They'll probably survive up to perhaps about 115 C or so but don't, for the love of Eris, try to operate something at that temperature regularly. So, if it's less than 70 when hot, all is great, if it tops out below 90, don't worry but make sure your cooling works, if you keep spiking above 90 C, get better cooling and/or stop doing whatever it is that makes it that hot.

Edited by SakuranoSenshi, 09 August 2012 - 09:27 PM.


#19 Quinn Allard

    Member

  • Veteran Founder
  • 175 posts
  • LocationUSA

Posted 09 August 2012 - 09:30 PM

View PostSephlock, on 09 August 2012 - 07:45 PM, said:


Have you considered counseling :)?

Alternatively, share your supercool computer building secrets with us!



Well, considering she let me spend $900 on a rig and another $100 or so in software/games AND the fact I am playing said games......yeah she can put the A/C on whatever she wants lol.



Anyways I talked to my friend who built the bottom end (MB,CPU,PSU) and what he told me I shall share the with you guys....you know, cause we are MechWarriors. He said he used the same thermal paste that diesel mechanics use, that in the past 10 years of him having a computer business it has proven to be the best. That and cable mgmt, the size of the tower, Air flow, and push/pull techniques. It certainly doesnt look very pretty, but it works. I counted 13 (!!) fans of all different sizes in/on the case, which includes GPU, CPU, PSU, and HDD fans. All channeling Air from top to bottom, with different rpms. I looked in BIOS and it said core temp was same as HM, 30c-31c with each core around 15c-17c. Now, Ive had the machine on all day. Played Diablo 3, Battlefield 3, and burned 3 DVDs. Plus played 10-15 rounds of MWO:Beta. Right now at this moment HM is reporting MB CPU at 30c, and indiviual cores at 13c.

Edited by Quinn Allard, 09 August 2012 - 09:31 PM.


#20 wint0nic

    Member

  • Veteran Founder
  • 32 posts
  • LocationChicago, Illinois

Posted 09 August 2012 - 09:33 PM

It's accurate bro, is something I use to monitor temps.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users