**Official e8400/e8500 Retail OC Thread

[ThatGuy16]

so do you think these temp readings are from the chip itself, poor bios support, or the monitoring software isn't up to date with it? would that not be a possibility?

[Merc14]

so do you think these temp readings are from the chip itself, poor bios support, or the monitoring software isn't up to date with it? would that not be a possibility?

It could be any of those but so many folks are having the same problem on so many different boards and chipsets that one has to assume that the only common denominator, the chip, is at fault. Unfortunately, Tankguys doesn't warranty CPUs and I have never returned a chip to Intel. At $200 it isn't a huge deal but this is pretty disappointing. Great chip but WTF, you can't read the temperature? Makes it worthless on the resale market. Looks like the wife may be going Wolfdale as she is still using my old 4400X2 in her rig.

[unclewebb]

Intel never documented the on chip digital thermal sensor (DTS) to be used for reporting idle temperatures and they've never documented these sensors to be used for reporting absolute core temperatures.

Using the DTS to report idle temperatures has always resulted in readings that are too low at idle. This has only become more obvious with the E8x00 series that uses very little wattage at idle and creates very little heat.

The new E8x00 chips still have a huge amount of temperature headroom when CoreTemp reports that they are running at 60C. At this point you are still 45C away from the Intel documented maximum safe operating temperature so what's there to worry about. Even at 4500 MHz and beyond I've yet to see a screen shot where anyone is even close to the throttling point.

This thread proves that Wolfdale is working great so Intel is not going to change anything. CoreTemp reads directly from the cpu so no bios update is going to change anything either. Everyone better get used to absolute temperature readings that don't have as much meaning as they used to.

Use CoreTemp and set it so it displays how far you are away from TjMax. That's the only fully Intel documented temperature information from these chips that you need to be aware of.

[mrcape]

It could be any of those but so many folks are having the same problem on so many different boards and chipsets that one has to assume that the only common denominator, the chip, is at fault. Unfortunately, Tankguys doesn't warranty CPUs and I have never returned a chip to Intel. At $200 it isn't a huge deal but this is pretty disappointing. Great chip but WTF, you can't read the temperature? Makes it worthless on the resale market. Looks like the wife may be going Wolfdale as she is still using my old 4400X2 in her rig.

Dude these chips are sold out, and hard to get in many places. ebay it if you don't want it.

[dinos22]

There seems to be the usual amount of confusion about temperatures so here's another way to directly access the data from within the Intel CPU.

Start up CPU-Z and go to the About tab and click on the Registers Dump (.txt) button.
Save this file, open it and scroll down to the model specific registers for each of your cores.



MSR 0x0000019C contains the only Intel documented temperature data for these processors. By looking in here you are reading the digital thermal sensors (DTS) directly. Bits [22..16] are located in the third and fourth digit. In my example that is 0x3F hexadecimal which equals 63 decimal.

If you are mathematically challenged then fire up Calc or head to Google and type in:
0x3F in decimal
and it will do the conversion for you.

This shows that my Core0, at idle, is 63 degrees away from TjMax or its safe maximum operating temperature.

I am using an older E6400 which has a TjMax = 85C so CoreTemp reads the DTS data and then calculates that my absolute temperature is 63C degrees away from 85C so it reports an idle temperature of 22C (85C - 63C). CoreTemp is using TjMax = 105C for the new 8x00 series processors.

Intel neither publicly documents that these on chip sensors are accurate for low idle temperatures or provides any public documentation stating what TjMax actually is. It only documents that these sensors are used for thermal throttling and ultimately for thermal shut down if needed.

When your processor heats up the DTS will decrease. When the DTS approaches zero you will begin to slow down due to thermal throttling kicking in but most people are miles away from that ever happening. As long as you always have some headroom as reported by the DTS and your computer is stable at full Prime load then there is no reason to be too worried about your absolute temperature. It is just a number and you are operating within the Intel temperature spec for these chips. At 4+ GHz you might be a little over on that spec but that's a secret that Intel doesn't need to know.

You can also start CoreTemp and go into the Options->Settings menu item and turn on Show Delta to Tjunction max temp.
This setting also reads the DTS data directly and should exactly correspond with the above DTS data that CPU-Z prints out.

nice explanation there

check out what i also found in that CPUz dump
(btw 0x34 was the corresponding value as per your explanation which is also mentioned in my screen under Core0)

[UnSeenSVK]

i dont understand this my temperature..where is problem?...

http://img218.imageshack.us/my.php?image=hhhhko1.jpg

my temperature is 11C i don t understand this...why...?..please help me..where is problem?..cant be 11 C


Mb has 38C
NB has 50C
SB has 47C


and processor has still 11C

showing in bios...core temp..asus probe..everest...

[UnSeenSVK]

i used same everest..shows 11C...

core temp 0.96.1 showing core 0 12C core 2 16 ....
when i running 3dmark06 core 0 shows... 17C and core 2... 16C

look

http://img85.imageshack.us/my.php?image=ddddui7.jpg

[unclewebb]

Intel has been using bits [22..16] of MSR 0x19C to store the DTS temperature data for all Core based processors including the mobile chips and even the early single Core chips.

As far as I know they started using the DTS sensor and 0x19C in the late P4 series but I don't have one of those ovens handy to confirm this. All my info comes from the publicly available Intel documentation:

13.5.5 On Die Digital Thermal Sensors
http://www.intel.com/design/processo...als/253668.pdf
(7.5MB download)

Now that we know how to get this information out of a Core processor and we know that CPU-Z is capable of reading the model specific registers (MSR), maybe someone could convince Franck Delattree, programmer of CPU-Z, to include a real time DTS reading on the main CPU page of his excellent program.

UnSeenSVK: Read post #730:
"Using the DTS to report idle temperatures has always resulted in readings that are too low at idle."

[aoch88]

45nm dies and degrades quick. 1.45V is the limit air without definitive damage and you have no way to verify temps, but they will be very high with >1.4 on air as stock is around 50C loaded at 3GHz 1.1V. I've already returned the first batch we had at work since they all read completely wrong temps, now waiting for the 2nd one.

You won't kill the chip for running 1.5V for a short while. The degrading and dying part only comes about if it's prolong usage.

[UnSeenSVK]

So can i do overclocking?...how i see my real temperature?...

i m really unhappy .....

how many C i have to plus to my teperature?..

thanls mate

Intel has been using bits [22..16] of MSR 0x19C to store the DTS temperature data for all Core based processors including the mobile chips and even the early single Core chips.

As far as I know they started using the DTS sensor and 0x19C in the late P4 series but I don't have one of those ovens handy to confirm this. All my info comes from the publicly available Intel documentation:

13.5.5 On Die Digital Thermal Sensors
http://www.intel.com/design/processo...als/253668.pdf
(7.5MB download)

Now that we know how to get this information out of a Core processor and we know that CPU-Z is capable of reading the model specific registers (MSR), maybe someone could convince Franck Delattree, programmer of CPU-Z, to include a real time DTS reading on the main CPU page of his excellent program.

UnSeenSVK: Read post #730:
"Using the DTS to report idle temperatures has always resulted in readings that are too low at idle."

[Blackwhite]

I found interesting article:
http://www.anandtech.com/cpuchipsets...spx?i=3184&p=5

It is about Q9650, but I it is 45 nm too.

[Fpat]

Hi,
Here is my best fsb with my DFI P35 T2R and an Intel E8500.

http://valid.x86-secret.com/show_oc.php?id=306349

[Supershanks]

I'm showing a difference between the cores CPU temp(Tcase) v Cores Temp (DTS based)

[UnSeenSVK]

Hallo mate i ve gor problem with my E8500 still showing 11C temp...

which program on temp you using?

tnahks

check this please...

what do you think about that?


http://img85.imageshack.us/my.php?image=ddddui7.jpg

Hi,
Here is my best fsb with my DFI P35 T2R and an Intel E8500.

http://valid.x86-secret.com/show_oc.php?id=306349

[kiwi]

Yep, 0x0000019C shows same value as coretemp, I am @ 41C (full load) till TjMax

You won't kill the chip for running 1.5V for a short while. The degrading and dying part only comes about if it's prolong usage.

Well, someone reported here a huge and fast degradation running @ 1.5V on water. Lost like 100Mhz orthos in 3 days or something. And that's 8 similar cases.

[Blackwhite]

I'm using coretemp 0.96.1, and my idle temp is still the same, doesn't matter do I set 1.225v or 1.5v

[UnSeenSVK]

I thing we have to wait on proper version core temp for new processors

[Blackwhite]

I thing we have to wait on proper version core temp for new processors

And for a new BIOS as well - my Voffset after OC is 0.117v

[stasio]

Usefull Core 2 Quad and Duo Temperature Guide
HERE
Updated January 30th, 2008

[UnSeenSVK]

i m thinking about update chipset...can be in there?...

[zakelwe]

In similar fashion to Zeus' chip mine requires 1.48v to run 3dmark at 4500MHz but then will not SuperPi at 4600MHz which is a lot smaller difference than between the 65nm Core2's which allowed a lot higher 2D benching speed than 3D.

Mine also hits a wall at 4.6GHz where more voltage does nothing at all. Either it is a temperature thing or more likely some side effect of the new hafnium 45nm process that shows, perhaps unsurpisingly, a different way of scaling.

4500MHz on air equivalent of 4700MHz Conroe is nothing to be sniffed at though, especially if it is just leass than 1.5v.

Regards

Andy

[Viss]

Hi,
Here is my best fsb with my DFI P35 T2R and an Intel E8500.

http://valid.x86-secret.com/show_oc.php?id=306349

Nice, what cooling do you use?

[Praz]

Usefull Core 2 Quad and Duo Temperature Guide
HERE
Updated January 30th, 2008

Although useful for someone looking for the an explanation of terminology used in processor measurement that article really doesn't apply to issues being seen with the 45nm family of processors. Reliable temperature reporting is broke and no amount of calibration or adjustment of software is gong to fix that.

[ArCElM]

I personally, would like to see people being more specific with what votage they are using, instead of just saying, "my chip needs 1.45" to do such n such speed etc, I think they need to specify if it needs that REAL or BIOS, because what I for example set in my BIOS is NOTHING close to what my chip gets, according to PC Health Status in BIOS, CPU-z, and Everest etc, I could say my chip requires 1.465v to ORTHOS 4.1Ghz - BUT..



As you can see, it requires only 1.32v load, idle is 1.352v




So I am just sugesting as several people are having similar problems to me, specially Abit IP35 Series users, it would be HELPFUL to be more specific about what our chip is USING, and what is being SET, just my thought.

Because at the end of the day, I was really panicing when I realised I needed to set 1.465 to get 4.1 Stable, I thought I had a REALLY BUM chip, but its actually average-decent, I think..

[Blackwhite]

So I am just sugesting as several people are having similar problems to me, specially Abit IP35 Series users, it would be HELPFUL to be more specific about what our chip is USING, and what is being SET, just my thought.

Because at the end of the day, I was really panicing when I realised I needed to set 1.465 to get 4.1 Stable, I thought I had a REALLY BUM chip, but its actually average-decent, I think..

Did you measure voltage by voltometer, or in BIOS?
I need set in BIOS 1.485V , MCH1.45V, VTT 1.31V.
My Voffset is 0.085V + Vdrop 0.032V- so to Run at 4GHz I need 1.368V
After 1.485Vcore and restarting comp - PC Health Status showing me 1.39Vcore - I didn't mesure voltometer.