E6600 Temps

well i recently installed a new e6600, an upgrade from my e6400 ( i got it for free). I used AS5 on my old cpu and it would run at 35-37 idle, not to sure at load. I installed this new proc with ocz ultra stuff (same as AS5 just different name so ive heard) and the temps are 18c on core 1 and 20c on core 2. This temps have to be to cool for this proc, theres no way it can run this cool.

Also im running a cnps9500. Its kind of scaring me as i have no idea how hot this thing is running, dont even want to open up and games and such till i get get a good temp reading. Ive used Coretemp .94 and .95 and same with everest beta and they are all reporting this proc very cool.

SS added http://img157.imageshack.us/img157/2614/coretempil7.jpg

It's definately not running that cool, I have mine at 1.15v @3.1 and in the spring /summer months I idle @28/30.

What board are you running? You can also look in the bios and see what it says, if it reads the same, then I have no idea where the problem lies.

What is your room temperature?

The weird thing is that CoreTemp 0.94 and 0.95 usually report the same temperature on the revision B2 Conroe processors and that temperature is usually right.

What does SpeedFan show for your core temps and for your core voltage?

Core temperatures that approach your room temperature are possible but reported core temperatures below your room temperature are not.

vcore 1.33v via speedfan, cant tell the temp in speedfan its all messed up but i got and installed that intel core temp program TAT or w/e it was and i was still getting readings of 20c

bios says the same as everything else, took out my infared thermometer and got a rough reading of 20c as well... no idea whats up... is it possible that contacts on the cpu arent being fully made and thats causing an issue?

also room temp is fairly low... 20c or so maybe lower. in a basement of an AC cooled house and it was fairly cool outside today to.

There shouldnt be a need for me to reformat my system after plugging a new proc in should there. MB is p5w dh deluxe

In a cool basement a core temp of 20C is certainly possible.
My basement is only about 15C today.

i came from avg temps of a stock e6400 at 35-37 idle, how can this be running cooler then that? it doesnt seem possible. Is it possible that this chip is special or defective in some messed up way?

EDIT: Well i used intels little temp monitoring program and told it to put the cpus under 100% workload and the temps got up to around 40c or so which i guess isnt to cold under full workload. This just catches me as very odd and makes me question if these temps are correct.

2nd EDIT: Also looked up this chip and from what ive seen it seems to run fairly cool but it doesnt want to OC very well, makes me wish i was still using my e6400, could always go get it again if need be.

Is it possible that your previous processor was being reported incorrectly?

Do you know if it was a revision B2 or revision L2 E6400 and what software were you using?

Determining the core temperature of a Core 2 Duo can be a bit of a guessing game. Intel has no documented way to calculate the absolute temperature of a processor so programs like CoreTemp, SpeedFan and TAT are guessing.

No single program is right 100% of the time. Sometimes they come up with a reasonable number and sometimes they don't.

The basic formula is this:

Absolute Core Temp = Tjunction - DTS

The digital temperature sensor ( DTS ) is fully documented by Intel and can be read from any Core 2 Duo and is 100% accurate. The problem is that the Tjunction value is not documented. Programs guess that it is either 85C or 100C but some people believe that it might not even be an exact number.

If software thought your previous E6400 had a Tjunction of 100C but it only had a Tjunction of 85C then your reported temperatures may have been 15C greater than what they really were.

The early E6600 processors were all believed to have a Tjunction of 85C. If that's not true or if your E6600 is a new one and Intel has switched them to 100C then your temperatures might be reported 15C too low.

I'm interested to learn more about your present and previous processors so I can learn a little more about this.

So far it seems that Intel has produced the Allendale, revision L2, E4300 / E6300 / E6400 in both 85C and 100C Tjunction varieties with no documented way to tell the difference. Do you have the date code from your processor box?

The e6400 is revision b2, im looking at it right now on the computer it is currently installed on. To check the temps when it was installed in my rig i used core temp.

atm im not liking this e6600 from the poor OCing results ive heard with my week etc and the fact that its playing this weird temp game on me >.> <.<

EDIT: yes i have the date code from the box lemme run back downstairs

ok the batch is L624A769

pack date is 07/13/2006 if that helps at all,

Also im pretty sure that CoreTemp always assumed the procs had a tjunction of 85c. It was displayed above the core temps.

wasnt sure if u were asking for the e6600 code aswell so here it is

batch on the e6600 is l646g512 (bad overclocker so ive heard )
would make better sense if this proc was tjunction 100 and coretemp was thinking it was 85, it would put me near the 35c mark

All of the early revision B2 E6400 chips I've heard about have a Tjunction of 85C so your old system should have been reporting the CoreTemps correctly.

A date code of 624 means the 24th week of 2006 or about June when it was manufactured and then shipped in July which makes sense..

At the moment I'd have to assume that your newer E6600 has a Tjunction of 100C so CoreTemp and others are reporting it 15C too low.

I came up with an indirect test to try and determine what your Tjunction is.

If you go into the bios and lock it at 6 X 266 and drop the voltage down to 1.1 volts if your board allows or drop it as low as you can then let your computer idle, preferably with the case open. You are trying to drop your idle temps as low as humanly possible. Crank the CPU fan up if it's adjustable too.

If you get a reported temperature that is less than your ambient temperature then your software is wrong. You are close to that already. You need an accurate reading of your room temperature to 101% prove it.

With my early B2 E6400 I was able to get my CoreTemp to exactly equal my ambient temperature so that proved to me that the assumed Tjunction of 85C for my processor was correct.

Well I guess ill need a good old mercury thermometer to get a good ambient temp. I already expect the tjunction to be 100c because just running mozilla im down around 17c on my lowest core 0.o

when i get home ill take the case off and drop the clock as low as it will go.

I call this my Limbo test.



You run your processor as slow as you can with the least amount of voltage you can and see what your reported idle temperatures are. At idle they should decrease and approach your room temperature and due to integer values and rounding errors, it's even possible for a reported core temperature to be 1 degree lower than your ambient temperature.

With a room temperature this morning of 21C, I got my core temperature down to a reported 21C which seems pretty reasonable to me.
I set it in the bios to 200 MHz x 6 and set the bios voltage to 1.125 which got me an actual idle voltage of 1.104 volts. At this level a C2D puts out virtually no heat. What little heat it does put out is easily handled by even my OEM heatsink and fan at max.

If you're getting 5C or 10C or 15C lower than ambient then obviously the software is using the wrong Tjunction value in its calculation. A reading 15C higher than ambient would also look pretty suspicious to me. This is common with the E4300 when CoreTemp 0.95 assumes a Tjunction of 100C even though many of them only have a Tjunction of 85C.

This is a great way to calibrate your temps readings and make sure it's right. Depending on the results, get the software who provide the most accurate reading then use it to watch the actual temps.

Xilikon said:

This is a great way to calibrate your temps

Click to expand...

Thanks. It has worked for me and a few other users that have tried it.

I know the E4300 is available with both a 85C and a 100C Tjunction and now it looks like the E6600 might be the same. There is no Intel documented way to tell the difference so users that need to know an absolute temperature will have to use some sort of indirect method like the above.

It's a lot safer than trying to take your processor up to the boiling point of 85C or 100C which might not tell you anything anyhow.

Room temp seems to be about 17c and im down to 11 -13c on this proc now running at 1.2v and i turned the multiplier down to 6. i guess this one really does have a tjunction of 100c

still find it odd that even the bios temp is wrong, should i have reset teh bios when i get the new proc?

Is this only on newer e6600's that run the 100c Tjunction? Does this mean anything performance wise on the chip? Ive already heard this chip = the suck for overclocking and im debating putting my e6400 back in. Any opinions?

SavageThrash said:

Is this only on newer e6600's that run the 100c Tjunction?

Click to expand...

I've done a pile of research on this topic and so far yours is the first revision B2 processor that seems to have a Tjunction of 100C.

Too many users read on a forum that their processor is no good so they stop trying to overclock it without putting in much effort or even knowing what the hell they're doing.

The common myth has been that the Allendale E4300 doesn't overclock very well. Some people like the guys at Nordic Hardware try not to let the myths get in the way of their overclocking.

It will be interesting to see what a new E6600 is capable of.

I googled the same week proc as mine i have now and such and i found people saying they could only reach about 3.2ghz with this one... doesn't sound to promising.

How would you recommend i go about overclocking this chip, set fbs to 333 x9 on auto voltage then try to lower the voltage and find a stable area then move the clock slowly back up?

I don't have any direct experience with the 975 chipset but 333 X 9 or even 366 X 9 sounds like a good place to start.

Run your memory with loose timings of CL5-5-5-15 and lock the voltage at 2.1 to 2.2 volts so it doesn't hold you back. Use the smallest memory multiplier so your memory runs as slow as possible. This helps take your memory out of the equation so it doesn't hold you back. Once you find a good MHz that your cpu likes to run at, you can go back and adjust the memory speed and timings upward to maximize your overall performance.

Even with a Tjunction of 100C, your cpu is running very cool and has lots of head room. A bios setting of 1.45 volts like I'm using seems like a reasonable place to begin. Intel rates the C2D to 1.55 volts so as long as your temps are cool, you can experiment with the core voltage without hurting anything.

If it boots up, run Orthos. If it's stable, give it some more MHz. If it's not then give it some more voltage. Your northbridge and southbridge should also be set to a fixed voltage rather than AUTO. It makes it easier to troubleshoot the fewer variables you have to keep track of.

I also use SuperPI mod as a good initial testing tool. When it starts to report errors you know you need to add some more voltage without having to do a 4 hour Orthos run.

Here's one thing that X-Bit Labs ran into when overclocking with their Asus board:
http://www.xbitlabs.com/articles/cpu/display/core2duo-e6300_8.html

Your 9X multi will help you out and won't hold you back like their E6300 with a 7X multi did.

For me a practical 24/7 kind of overclock is about 3400 MHz to 3600 MHz for the early Conroe core B2 revision C2D processors. You should be able to get up into that range when you're all done if you put some time and testing into this project. You'll know you're reaching the limit of your processor when the voltage has to go up by a large amount to maintain Orthos stability even though your overclock in MHz has only increased slightly .

Ok well im currently running some OC's on this machine, using this http://forumz.tomshardware.com/hardware/Core2Duo-Overclocking-Guide-v1-ftopict197995.html as well for voltages and such and what to disable etc.

If i can get the system to boot what exactly should i do before i reboot go back to bios and up the mhz. Such as calculate pie to 8 million digits and if its ok go to the next level etc because i don't want to run 4 hours of orthos just for every 5 mhz i increase. i can understand once it gets to where i want to keep it ill run tons of stress tests but not when im just testing the waters etc

EDIT: one other thing, i shouldnt be passing about 60c so that means 45c since this things got 100c tjunction. im at 3ghz and im already hitting 46c+ in TAT when i tell it to stress both cores to 100% Im running a 9500 and to my knowledge all fans in my case are at 100%

running orthos atm and it seems so far that even though teh cpu is maxed out its not as hot as when i ran TAT

I must have a 85c Tjunction E4300, at least I think I do. I'm running water cooling (see sig) and with my CPU down clocked like you suggested, I'm slightly under (1c) room temp for the 85C Tjunction. When I check 0.95 CoreTemp with the 100C Tjunction, my idle is 13c or so higher than room temp, which really can't be close for the type of water cooling I'm running.
So going by this test, I asume that my E4300 is a 85c Tjunction and I should go by the temps from CoreTemp 0.94???????

Bio-Hazard: So far the early E4300 with a date code of L640 seem to have a Tjunction of 100C and ones with a date code of L643 or L644 have an 85C Tjunction. I'm not sure about the other ones. My test seems to work for air cooled systems. For water cooling I think you'd be comparing your core temperature to your water temperature.

I'm not a big fan of TAT. It's a laptop testing tool and runs a bunch of instructions in a tight loop that heats up one part of the C2D. There are multiple temperature sensors in a C2D so Intel designed their program to heat up one part of the processor. There doesn't seem to be any real world applications that heat up your cpu like TAT does so I can't figure out a reason to use it.

Even Orthos is a little overkill. There are some encoding apps that heat up both cores but I haven't seen any results where they create more heat than Orthos. For me, the [H]ard standard of 4 hours Orthos stable is good enough.

I use ClockGen when testing. You can bump the FSB up 10 MHz at a time and do a quick 1M or 2M SuperPi run and you will know real quick how your stability is coming along. Orthos stability usually requires another bump up in voltage beyond what SuperPI stability requires but it gets you in the neighborhood.

A processor with a Tjunction of 100C has a lot of headroom so your arbitrary limit of 60C seems pretty conservative to me. Intel guarantees your processor to be able to run reliably up to 100C so why stop at 60C?

CoreTemp 0.95 can be set to directly read the digital temperature sensor ( DTS ) of the Core 2 Duo and this number is 100% correct for all C2Ds regardless of what Tjunction is right. DTS decreases as your processor gets hotter and it tells you how many degrees until your processor will reach its maximum allowed temperature.

Using a separate program called CrystalCPUID I was able to directly read the DTS temperature register within the C2D and I can tell you that CoreTemp 0.95 is reporting this data correctly with no time lag. I cannot say the same thing for TAT. The temperatures it reports can vary by a couple of degrees from day to day even when the core DTS value is exactly the same. It also suffers some time lag where the real temperature has changed before it gets around to reporting it.

My C2D starts to throttle when the DTS=2 which is equivalent to an absolute temperature of 83C. Your processor shouldn't start to throttle until 98C. You don't have to use all of your available head room but don't be afraid to use some of what Intel has given you.

A processor with a Tjunction of 100C is more desirable than one with a Tjunction of 85C. It shows you that it is of better quality and has been tested and proven to run at a higher temperature.

unclewebb said:

A processor with a Tjunction of 100C has a lot of headroom so your arbitrary limit of 60C seems pretty conservative to me. Intel guarantees your processor to be able to run reliably up to 100C so why stop at 60C?

Click to expand...

I googled it and i found the max tempature for the e6600 to be 60c, thats why i didnt want to go above that.

SavageThrash said:

I googled it and i found the max tempature for the e6600 to be 60c, thats why i didnt want to go above that.

Click to expand...

The 60C is a Tcase temperature and not a core temperature. Nothing the processor does is based on Tcase temperature so It's irrelevant for the C2D.

Thermal throttling and processor shut down are based solely on your core temperature readings. A 100C Tjunction processor can get up into the high 90C range before you have any real worries.

Any worries before that are just your own personal fears.



And to think that you've got another 15C of headroom over and above my 85C Tjunction processor! I'd be able to really turn up the voltage with your E6600!

ok so ill use core temp and just remember that my temps are 15c above that... so i can reach 75c and be fairly safe

There is a lot of confusion about what Tjunction a processor has and what temperature reading to believe and what's a "safe" temperature, etc, etc.

The most important thing is running your processor within the Intel guidelines when it comes to heat. Intel fully documents the digital thermal sensor built into all C2D processors and CoreTemp 0.95 is the only program that lets you see exactly what this number is. CoreTemp reads the DTS in real time directly from the processor and displays the results without any further manipulation. This is the only temperature number anyone needs to know.

DTS is the number of degrees until your processor reaches its maximum, Intel guaranteed, safe operating temperature. As your processor warms up, this number decreases until it reaches zero. Soon afterwards if your C2D isn't able to cool itself down, it will automatically shut itself down. Intel designed it this way so they wouldn't have to worry about warranty claims. The C2D is extremely well designed and will take care of itself in the event of your cpu fan failing or even if you're a crazy ass overclocker with not a speck of common sense! .

Before it reaches the shutdown point at approximately DTS=2, the cpu will start to throttle itself and slow down. No one wants their processor running at part throttle so you definitely don't want to be running it this hot. If you leave yourself 15C of headroom you'll never have any stability issues due to heat and you'll never damage your processor. Even though your absolute temperature may be a big number, Intel designed their processors to be able to operate reliably all the way up until they start to throttle at DTS=2.

Woah, thanks for the wealth of information Unclewebb. Cleared up a lot of the questions i had

Just to clear up a few points.

In my personal testing when mildly overclocked you can run reliably right up to the throttling point at DTS=2.

When you are fully overclocked the amount of head room you have decreases. I have excellent long term Orthos stability at DTS=15 when fully overclocked but when that drops down to DTS=10, Orthos will soon crash and my computer will reboot.

This is almost a good thing. If you are overclocking to the max then you'll never have to worry about damaging your cpu due to heat. It will shut down before it even gets close to its maximum rated temperature.

Go for Orthos stability and don't worry too much about temps. Four hours of Orthos stability when overclocked automatically means that your processor is not overheating. You wouldn't be able to complete a good Orthos run if it was.

unclewebb said:

Just to clear up a few points.

In my personal testing when mildly overclocked you can run reliably right up to the throttling point at DTS=2.

When you are fully overclocked the amount of head room you have decreases. I have excellent long term Orthos stability at DTS=15 when fully overclocked but when that drops down to DTS=10, Orthos will soon crash and my computer will reboot.

This is almost a good thing. If you are overclocking to the max then you'll never have to worry about damaging your cpu due to heat. It will shut down before it even gets close to its maximum rated temperature.

Go for Orthos stability and don't worry too much about temps. Four hours of Orthos stability when overclocked automatically means that your processor is not overheating. You wouldn't be able to complete a good Orthos run if it was.

Click to expand...

I agree... I also pushed the E4300 to the thermal limits in the past as a way to research about temps accuracy and it shut down before damage is done. Now, after getting thru, the cpu is still running strong with a run at 9x400 and 1.6v yesterday to test the fsb limit

At 3400 MHz as soon as DTS hit 9 it would usually reboot within 15 seconds.

I decided to leave everything the same but using ClockGen I dropped down to 3200 MHz.

This increased my temperature head room and allowed me to run Orthos for half an hour at full temperature with no problems.

The Thermal Monitor occasionally kicked in to keep the temps from exceeding what Intel considers to be safe.
TAT reports this in bright red, "Thermal Monitor Active."



Today's Lesson: As long as your Orthos stable, it's good!

Ok so from what you just said i got that : the lower the OC the higher the temps can go (or closer to tjunction) but for higher overclocks the temps need to be lower to be stable?

SavageThrash said:

Ok so from what you just said i got that : the lower the OC the higher the temps can go (or closer to tjunction) but for higher overclocks the temps need to be lower to be stable?

Click to expand...

Exactly.

For me 3200 MHz is still a pretty decent overclock but if you're pushing your cpu to the absolute limit then you won't be able to do that if it's red hot near the DTS=2 throttling point.

Sorry to dig up an older thread, but I do thank you unclewebb for all of your research. I'll try the low clock testing today and see what happens, but I never thought of it in the sense that having a 100C tJunction might be a good thing in the sense that 70C isn't that hot in comparison to what it is capable of handling...kinda makes you feel GOOD about having a 100C tJunction where I was always thinking the opposite.

HELP... speed fan tells me e6600 core 1 temp is 3 degrees higher than my core 2 temp...

has anyone experience this situation?

should i try and reinstall my fan?

chino182 said:

HELP... speed fan tells me e6600 core 1 temp is 3 degrees higher than my core 2 temp...

has anyone experience this situation?

should i try and reinstall my fan?

Click to expand...

Its normal for the core temps to vary. Mine are normally between 2 and 4 degrees apart.

chino182 said:

HELP... speed fan tells me e6600 core 1 temp is 3 degrees higher than my core 2 temp...

has anyone experience this situation?

should i try and reinstall my fan?

Click to expand...

Is that under load ? Normally, under load, the temp difference should be almost nonexistent but 1-2 degrees is normal. If you get over 3 degrees, start to worry and be ready to RMA since it may indicate a badly assembled CPU where the IHS may not be sitting equally on the core. This happened to me with a E6420.

Xilikon said:

If you get over 3 degrees, start to worry and be ready to RMA

Click to expand...

WTF!!!???

my E6600 is currently idle at 25/27 and under full load hits about 45c.

thats with it oc`d to 3.2GHz on 1.240v

Xilikon said:

Is that under load ? Normally, under load, the temp difference should be almost nonexistent but 1-2 degrees is normal. If you get over 3 degrees, start to worry and be ready to RMA since it may indicate a badly assembled CPU where the IHS may not be sitting equally on the core. This happened to me with a E6420.

Click to expand...

I beg to differ...it has been well documented on XS and other forums that a few degrees difference is nothing to worry about, and should actually be expected, especially once you start talking about quad core, as one core takes on the brunt of the labor and offload work to the other. I'm no expert in how these things work, but 3C difference in core temps is perfectly fine.

cornelious0_0 said:

I beg to differ...it has been well documented on XS and other forums that a few degrees difference is nothing to worry about, and should actually be expected, especially once you start talking about quad core, as one core takes on the brunt of the labor and offload work to the other. I'm no expert in how these things work, but 3C difference in core temps is perfectly fine.

Click to expand...

In my case, it actually reached up to 6C difference

Xilikon said:

In my case, it actually reached up to 6C difference

Click to expand...

I've seen it up to 5C, but only briefly...typically it's a 1-3C difference...much tighter under load though.

cornelious0_0 said:

I've seen it up to 5C, but only briefly...typically it's a 1-3C difference...much tighter under load though.

Click to expand...

It's just one of the reasons. The other reason is my inability to run it for 24 hours stable at stock speed and stock voltage, hence my observation.