x264 encoding SD content at 1.88 Frames per second ?!

[ghostonline]

BIOS temperature reading is always less accurate than coretemp/realtemp. Motherboard's cpu temperature reading sensor is located somewhere around cpu, usually right below cpu socket. there's no way for such a sensor to read more accurately than dts sensor on cpu die... basically, regarding cpu temperature, intel/amd know better than any motherboard maker.........

[LoRd_MuldeR]

Well, from what I understand there are two types of sensors for CPU temperature: The mainboard's "CPU sensor", which gives you exactly one value an isn't very accurate. And the sensors on the CPU die, which give you one value per CPU core (I'd expect these to be more accurate, actually as accurate as we can get). So overall there are three CPU sensors for a Dualcore and five sensors for a Quadcore. There aren't any other CPU sensors and there is no other way (for a software program) to obtain CPU temperature information - expect for mounting your own sensor on the CPU, but that really would be a special case.

So any application that shows the "CPU Temperature" has to read out those sensors. And since the sensors are the same and consequently the data from the sensors is identical (and also the "accuracy" is identical), the one and only difference between all those applications showing you the "CPU Temperature" is how the sensor data is interpreted by the software! I'm looking only(!) at the temperature for individual CPU cores. And that data obviously must come directly from the sensors on the CPU die. Still HW Monitor shows different values than CoreTemp. The latter shows ~10°C colder!

Consequently the two applications must be interpreting the sensor data differently! And since the difference is quite big and it's impossible that both are right at the same time (unless doublethink is enabled ^^), one of them has to be wrong. I tend to believe that HW Monitor is right, because roughly 60°C for a Q6600 (B3 stepping) under full load with crappy "boxed" cooler sounds too unrealistic to me...

http://img35.imageshack.us/img35/9971/cputemp2.th.png (http://img35.imageshack.us/img35/9971/cputemp2.png)

[ghostonline]

yes, there is a formula from intel that will convert those dts sensor data to a so-called meaningful temperature number. i am pretty sure you can google it. afaik, that formula needs the TjMax #. In the past, intel did not publish that # for core 2 desktop series. everyone just guessed a # and plug it in and calc temperature. HWmonitor probably used TjMax# as 95 C. On the other hand, TjMax # used by coretemp is 85 C. as far as i remember, 85 c should be more accurate for Q66 B3.you will find some of those figures in that pdf file on my previous post. Not everyone agreed with intel. if you really want to know the max on your cpu, just burn it down once to see.

if you got free time, you can read the development of realtemp http://www.xtremesystems.org/forums/showthread.php?t=179044

a good thing about realtemp is that you get the distance to TjMax, which is more useful than those temperature numbers.

last word on this topic, each cpu is calibrated different. if you really wish to get more accruate reading, you will have to do idle calibration by use infrared thermometer.



----------------------------
http://www.xtremesystems.org/forums/showpost.php?p=3928312&postcount=3630
http://www.xtremesystems.org/forums/showpost.php?p=3928312&postcount=3632

probably give you a better idea about role dts sensor data played here.

[LoRd_MuldeR]

Well, according to this (http://www.tomshardware.co.uk/intel-dts-specs,news-29460.html) article, the correct TjMax for Q6600 would be 80°C, neither 95°C nor 85°C :confused:

Also it says that 65nm CPU's give "reasonablel" temperature information even without calibration. Only 45nm CPU's need calibration to get "meaningful" info.

It seems Intel won't allow us to read out the correct temperature that easily :D

Anyway, I know that my CPU can encode at 100% load on all four cores for many hours. So I don't worry too much about the temperatures right now...

[ghostonline]

that article was not updated as intel made "error" when they first published that info. the corrected version was issued in nov'08. the pdf reflects the change.

again, not every temperature reading software author agreed with intel. they can choose to use the "real" # if it ever existed.....

[popper]

LOL, in all the time iv been reading D9, i dont think iv ever seen the lads collectively work out a hardware problem like this thread, well done.

to answer your question can you get it lower on your stock cooler, yes you can, its ALL about the way you fit your cooler, always pop into the bios and watch your temp for a good 30 minutes while you wiggle the block clockwise and anticloclwise to have it sit flush far better as it warms up the small amount of new thermal paste, but ONLY IF ITS WELL BELOW 60C...or you may loose it

my mildly overclocked core2 E4400 to get it to 2.1gig is running on its generic cooler fan running at 2000R no more ,95% CPU useage with AVIdemux AVC,ref3,AAC=2 1080P mp4 is running consistantly day in day out at 42C and NO higher than 50C in this UK summer weather according to the "real temp 3" you used. no higher than 102F if you prefer according to my prefered old but tiny "MB probe" always running in the tray from startup, right now its 98F.

so you can get it a lot lower than you have now with a little patence and several trys, cleaning off Both the CPU and the base of the cooler every time to make them match as flush as is possible.

a lower temp means you can mildly over clock it perfectly safely from your bios settings, getting it easily to 2.5+ at least and getting the Encodes done that little bit quicker, with patence and some OC reading, and you already know its able to take some extra heat without falling over anyway, running it so hot as you did for so long without realising it, and not having it drop speed until you ran the encoder.... and pushed it over the edge.

[bob0r]

And the CPU is still working fine?
Got to love those Intel CPUs!

I had some pentium 1 200MHz PC once, it ran like 4 years with a broken cpu fan, it ran untill the day i trew it away. :)
Maybe the dust in it had special cooling features! :D

[TheImperial2004]

LOL, in all the time iv been reading D9, i dont think iv ever seen the lads collectively work out a hardware problem like this thread, well done.

to answer your question can you get it lower on your stock cooler, yes you can, its ALL about the way you fit your cooler, always pop into the bios and watch your temp for a good 30 minutes while you wiggle the block clockwise and anticloclwise to have it sit flush far better as it warms up the small amount of new thermal paste, but ONLY IF ITS WELL BELOW 60C...or you may loose it

my mildly overclocked core2 E4400 to get it to 2.1gig is running on its generic cooler fan running at 2000R no more ,95% CPU useage with AVIdemux AVC,ref3,AAC=2 1080P mp4 is running consistantly day in day out at 42C and NO higher than 50C in this UK summer weather according to the "real temp 3" you used. no higher than 102F if you prefer according to my prefered old but tiny "MB probe" always running in the tray from startup, right now its 98F.

so you can get it a lot lower than you have now with a little patence and several trys, cleaning off Both the CPU and the base of the cooler every time to make them match as flush as is possible.

a lower temp means you can mildly over clock it perfectly safely from your bios settings, getting it easily to 2.5+ at least and getting the Encodes done that little bit quicker, with patence and some OC reading, and you already know its able to take some extra heat without falling over anyway, running it so hot as you did for so long without realising it, and not having it drop speed until you ran the encoder.... and pushed it over the edge.

Thanks ! I also think that hardware must be discussed extensively just like those software articles , cuz if you couldn't get a stable hardware enivronment , there is no way that you'll get any benifit from softwares .. ( A harsh truth I learned as this thread was in progress ) ... By the way , I like your idea about trial and error method , but I was struggling to get that old cooler to fit and I don't think that I'll mess it up again unless I change it :)

And the CPU is still working fine?
Got to love those Intel CPUs!

Yup ! It works just fine ! I too share the love of Intel CPUs :)