TechSpot - PC Technology News and Analysis

Right well I OC'd my e6300 from the stock 1.86 to 2.45 and I was just looking at temperatures. I got speedfan and HWMonitor they seem to disagree on the temp of my CPU. Speedfan will say its 15º lower than what HWMonitor says it is.

Can anybody tell me which program is accurate or if neither what program is?

But they agree on all the other temperatures, GPU, mobo, HDD, etc

What does PC Wizard say? http://www.cpuid.com/pcwizard.php

Just dl'd PC Wizard. It says 30º when Speedfan says 32, so pretty close there. And at that same point in time HWMonitor is saying its 47. I mean seeming as PC Wizard and HWMonitor are both programs made by CPUID you would think they would report the same CPU temp right?

Good point

We need others to reply.

Maybe go to Google and download them all (free ones)

I'm just going to stay subscribed to this thread, because I'd like to know the best one too. (other than putting a temp gauge on the CPU)

Haha, I do think that it is quite strange seeming as they would all be using the same thermomitors on the actual CPU right?

OK I decided to start Googling all over the place.
After visiting a few different boards (over clocking and alike)

I found that CoreTemp and SpeedFan seem to be preferred (didn't get the links)

Also instead of these programs actually checking the motherboard sensor, it seems they check the bios information, and then just copy that to Windows gui.

Also one user stated that you can never get a true reading from a sensor, due to many enviromental factors, and tolerances.
And that the only true reading was to use a thermometer.

So there you go, no one knows their CPU true temperature !

Cheers for your googling efforts kimsland. I downloaded Core temp. that agrees with HWMonitor, so Core Temp and SpeedFan which you say are prefered have a 15º difference, which is a pretty big difference I think.

Just hoping that if someone can clarify it, that SpeedFan is correct and that its the lower of the two temperatures

I've always heard Intel TAT is the most accurate, makes sense, its made by intel to run on their processors to read temps and stress the CPU.

If you are cooling by air, then 47C makes a lot more sense for an overclocked C2D than 30C does.

Intel Thermal Analysis Tool

Quote Intel(R) Software Network Support


Quote http://www.overclock.net/air-cooling...ysis-tool.html
Hmm OK

Yeah Well I am just using the stock HSF that came with my cpu (C2D E6300), aaah I would of prefered if the 30º was true.

Hmm I decided to give Intel TAT a try. I downloaded it and it gives me an error:
Error enumerating On Demand Clock Modulation support
Terminating Tool.

Hmmm I have a C2D I saw someone had an error in another forum because they werent using a C2, but I am Ohwell.

Yeah I got the same error, and promptly removed the program.

So, still the answer is none are truly accurate

well have u guys tried to use everest i mean that is the most accurate for me i have used like another 10 but nothing is close as everest its like 2 degrees off the bios reading. speedfan was 10 degrees of soo it makes alot of difference.

Amen to that! Especially with a stock Intel HSF! "Speedfan" seems to be more prone to telling users more of what they want to hear, (IMHO). Certain mobos have BIOS updates that actually add up to 15 C to their (proprietary) monitoring software. personally I'd buy into the higher reading, even if it's not what you want to hear. Your CPU will love you for it.
Note, this is far from the first thread raising questions about Speedfan's measurement validity.

lol captain i guess ur right it will keep u on the safe side if speedfan shows higher temps but it can also make u worry if they r toooooooo high

Well under load (with hyper pi) my CPU temp is 57-60 or 42-45 depending on what I go by. Say its for the sake of my CPU is that too high? I mean cos I was thinking I might just put it back to stock speeds because I don't really need the extra GHz provided by my OCing.

Core temp is close to TAT in temperature reporting. Like others have said, nothing heats up a core like TAT. Sandra burn-in loops may raise temps to 48C but TAT can drive temps up to 59C. 60C is about the upper edge of temp you'd want to see. A nice after-market HSF with Arctic Silver 5 can make a big difference.

I Have also been monitoring my cpu temp and have used sandra, Pc wizard (that was recommened further up the post) and speed fan. My processor temp seems to stay, between the 3 aps, the same (26 c). however, the core temps are the ones changing, but with onlt 1 or 2 degrees different. should i be looking at the core temp or just the cpu temp?

I have a 3deg difference between Speedfan and a program called MobileMeter. It's free too

I've just started to use PC Probe II that came with my motherboard. A lot of other programs I've tried read my temperatures horribly, ranging from the NEGATIVES all the way to 110 Celsius! Others seemed alright and read about 47 Celsius or above but I didn't think they were that high so I eventually gave in to what came on my motherboard CD. Right now PC Probe and NVIDIA control panel is telling me these temperatures / fan speed;

CPU: 26 Celsius
CPU fan speed: 3068 RPM (1600 RPM when quiet fan is enabled, CPU temp goes to about 30 Celsius)
MB: 33 Celsius
GPU: 48 Celsius
GPU fan speed: 675 RPM

Those seem about right, nothing is being to stressed right now.

Every culture has a belief system in place based on that which they'd like to believe. You know, if you're good you'll go to heaven, like that. Where do Klingons go? "Stovalcor", or something like that. since they're both heavens will they meet the Christians there? Stay tuned.
A philosophy based on logic such as the aforementioned dictates that you pick the CPU monitoring program that gives you the answer that you want to hear. That is, after all, what most people do.

What I'm still grappling with is this; if you can make a CPU with millions of transistors on a tiny chip that can decipher billions of bits of information in less that a second, why can't you make a sensor probes and software that will work together and give a consensus answer as to what the temperature actually is. Because, IMHO, most of the software is just giving the answers that everyone wants to hear.

If the machine blows up, then the highest reading was correct. If it throttles back, then the next highest reading was correct. If it does neither, then who cares? Believe what you like!