Heat probs reported for Intel E8400s

[Plaicd]

I think I figured out what is going on here. I am not 100% sure and I am by no means the brightest bulb in the light shop, but I think our motherboards (or in my case the ASUS P5E3) might be sending high voltage for FSB Termination.
I had a read of this QX9650 Datasheet

Now, I don't quite understand the whole electrical technical data, but it looks to me like the FSB Termination Voltage should be ~1.15V Max?
Now the minimum I can set FSB Termination Voltage on my P5E3 is 1.20V
CPU PLL Voltage it reads like the maximum Intel suggest should be no more than 1.55V?
The minimum my ASUS board will let me select is 1.50V
I was naive thinking that FSB Termination of 1.38 and SPU PLL of 1.56 were safe, but not anymore.
John

I am using a e8400 with an asus P5K deluxe mb bios 705 and fsb termination in the bios changes for 45nm and is set standard(auto) at 1.10v per what it says in the bios options.

My cpu still has the temp issues mentioned in this thread but has not suffered any stable oc decrease yet but I am only using 1.30v and 4ghz.

[Lestat]

i havent read thru the whole thread but mine reads like this

40/45c in core temp idle this is only a fake reading.
the minute you start priming the temps reach 60 to 65c. (with 1.45v 4.2ghz)
stop priming and they go back down and stay at 40/45

now using Smart guardian and speed fan you can also see the cpu temp, which is not the core temp but the temp of the onboard temp probe.

this reads much less and is a more realistic number and will fluctuate according to load. and again the minute you prime, it slams up to 55-65c depending on how wwell my air cooling is working. (ambient temps, room temps playing their part on keeping the cpu cooler)


the moral of the story stop using core temp until things are fixed.

if you also are using the DFI P35 Dark T2R/s the bios lets you adjust the cpu temp +/- to reflect a more accurate temp.
use that to get a better reading.

but yes the E8400 cpu's do not report CORE TEMP correctly. and as far as i know core temp is the only reading thats wrong.

[unclewebb]

Lestat: A core temperature reading is based on the hottest point within a core gathered from multiple on chip sensors and will always be greater than a reading from a motherboard CPU diode. With a good air cooler, a core temperature reading can also drop by 20C pretty much instantaneously when you go from full Prime load to no load. That's normal.

Processor throttling is based on one thing; how far the core temperature of your processor is away from TjMax which CoreTemp can read and display in real time. Go into Options->Settings and tell it to show that value. That's the only temperature number that's important and CoreTemp is the only program that shows that data correctly.

[alex17 GTX]

Strange senzors on these chips.
With coretemp i have 40/40C in idle and 47/40 in load at stock speed. I seems odd that only one core gets hotter.

And i have a problem with my DFI DARK (965). CPU vCore is set on auto and the VID = 1.125. But the actual vcore is 1.15 and i can not set a lower value. Besides that cpu-z is reading totally wrong (1.232).

I hope only the vcore is higher than is should.

[JohnZS]

I am using a e8400 with an asus P5K deluxe mb bios 705 and fsb termination in the bios changes for 45nm and is set standard(auto) at 1.10v per what it says in the bios options.

My cpu still has the temp issues mentioned in this thread but has not suffered any stable oc decrease yet but I am only using 1.30v and 4ghz.

Interesting... looks like ASUS have not quite added full 45nm support to the P5E3 (X38) based chipset in the BIOS.....yet.

What I would be interested to see is the QX9770 Datasheet as I have a feeling it's PLL Voltage might actually be 1.55V and it's Termination Voltage around the 1.2V mark.

I hear that the core Voltage is 1.30V...a shade higher than the QX9650 voltage of 1.25.

John

[emoners]

@alex17 GTX: at least your 1st & 2nd core temps are close, mine is way off! i've reasted my heatsink even 3x & still get the same readings. OC'ed & only the 1st core bumps 3c higher while the 2nd one doesn't move

[GMX]

Since they use less power than the E6xxx series its easy to conclude they dont run hot, but rather colder.

Anyone tried with TAT instead to measure the thermals?

Sounds more like wrong numbers from the program(s) used.

hmm, Presscott C0 rev comes to mind.

[audiofreak]

TAT doesn't work with E8xxx series (at least not "publicly" available 2.05 version). If anyone has newer version it would be nice if they shared it with us.

[Power5]

This is why I have held off on an 8xxx chip. The degradation of clock speed really worries me. Intel has a great OC rep, which is why some of us buy their cheapies. I think this is much worse than AMD and the errata bug in Phenoms. At least they dont die and people have pushed a whole lot of voltage into them to try and clock them high.

[trans am]

mine is 65c in orthos small fft
idle around 40c
1.45v
4.2ghz
temps seem perfectly normal considering the amount of vcore im pushing into a 45 nm cpu with an oced 8800gt in the same loop.

[unclewebb]

Strange senzors on these chips.
With coretemp i have 40/40C in idle and 47/40 in load at stock speed. I seems odd that only one core gets hotter.

There are numerous reports of the on chip sensors getting stuck at fixed values. Most CPUs have one sensor stuck at idle but when overclocking with additional core voltage and MHz under full load they start to work OK. If the core temperatures don't change when TaskManager goes from 0% to 100% and back to 0% load then the on chip sensor is stuck. At 50C that isn't much of a problem other than it's annoying. Intel calibrates these sensors to operate accurately at 100C +/- 20C so I'd be very surprised to hear about a stuck sensor at higher values near the throttling point.

And i have a problem with my DFI DARK (965). CPU vCore is set on auto and the VID = 1.125. But the actual vcore is 1.15 and i can not set a lower value. Besides that cpu-z is reading totally wrong (1.232).

How do you know that CPU-z is wrong and how do you know that your actual vcore is 1.15 volts? What you set in the bios and the VID of your chip may or may not have anything to do with the voltage that your CPU ends up getting. I've found that CPU-z is very accurate, especially on the older P965 chipset. VID is like a suggested voltage for a CPU but some motherboards may decide to ignore that value and set the voltage to what it thinks is best.

I don't know about the DFI board but I know on my Asus P5B Dlx (965) that setting vcore to AUTO can be dangerous when overclocking. AUTO sounds nice and safe but I've seen 1.60 volts at idle when vcore was set to AUTO in the bios. That kind of voltage might damage or degrade a new 45nm CPU.

[informal]

This is why I have held off on an 8xxx chip. The degradation of clock speed really worries me. Intel has a great OC rep, which is why some of us buy their cheapies. I think this is much worse than AMD and the errata bug in Phenoms. At least they dont die and people have pushed a whole lot of voltage into them to try and clock them high.

One part for the "durability" story of AMD chips is probably the SOI mix they are using in production.We've seen even a burn-in with some Phenoms,ie KTE's Phenoms exhibited this:after a longer burn-in periods,the chips could actually gain more Mhz with the same Vcore(although it could go either way,as one of his Phenoms did that)

I wonder if the new metal gates in transistors intel uses in Penryn refresh has anything to do with possible more pronounced electromigration effect with >1.45Vcore?

[flopper]

I wonder if the new metal gates in transistors intel uses in Penryn refresh has anything to do with possible more pronounced electromigration effect with >1.45Vcore?

metal vs glass.
I guess that is one effect with electrons.

[alex17 GTX]

How do you know that CPU-z is wrong and how do you know that your actual vcore is 1.15 volts? What you set in the bios and the VID of your chip may or may not have anything to do with the voltage that your CPU ends up getting. I've found that CPU-z is very accurate, especially on the older P965 chipset. VID is like a suggested voltage for a CPU but some motherboards may decide to ignore that value and set the voltage to what it thinks is best.

With my E6700 this board was pretty accurate. The VID was 1.325, in cpu-z it was 1.296- 1.312 and in smart guardian exactly the same. DFI Dark has small vdroop and it gets smaller when i plug in the cable for crossfire stability. (almost no vdroop from what cpu-z shows). BUT with E8400 everything is messed up:
BIOS: 1.15
CPUZ: 1.232
Smart Guardian: 1.23
SpeedFan: 1.23
That can't be right.
Anybody else with DFI DARK and 8400 around here?

[unclewebb]

alex17 GTX: You've got three programs in agreement at approximately 1.23 volts so why don't you want to believe them? The bios setting should be the one you least believe. Those software programs are reading from an on board sensor and seem to be right.

The 965 chipset was developed long before chips with very low VIDs were available for testing so it's likely that some boards can't give the cpu a low enough voltage no matter what you set the voltage to in the bios.

[alex17 GTX]

Let's assume the board can't give 1.15 becouse it's too low. But even when i pump more voltage it's almost the same thing, but now exactly the same difference between bios vcore and cpuz. I have to measure the vcore with DMM. If you are wright it means i can kill the chip becouse i don't know the real vcore.
I Have doubts becouse my cpu has the same FPO as another two cpus and both of them can do 4GHz stable with 1.28V. I don't think mine need 1.36V for 4G.

[unclewebb]

There is a lot of variation from one person's set-up to another on how much vcore they're going to need to be stable.

On the Mainboard tab of CPU-Z it shows what chip is being used to report voltages. My Asus P5B uses a Winbond W83627DHG which CPU-Z is programmed to read correctly.

Use a DMM to confirm but until then go by CPU-Z. The programmer put a lot of effort into supporting as many boards as possible, including yours, to correctly report core voltage.

[MuffinFlavored]

VID is from 0.85V – 1.3625V for the E8400. But you can read what's on that label as well as core temps to get VID (from my understanding).

The max Core voltage is 1.45V (page 17, someone correct me if I am wrong). E6000 series CPUs are able to go as high as 1.55V (page 20)

I think the 1.55v maximum applies to Q6000 series processors as well.

http://download.intel.com/design/pro...s/31559205.pdf
Page 17.

[higgins]

My cores wont go lower than this even when lowering the ambient to like 19C from 24-25C. My GPU and other components drop accordingly but my lil 8400 dont wanna play. Well, well as long as it clocks as a b*tch and doesnt burn up...

[alex17 GTX]

My Asus P5B uses a Winbond W83627DHG which CPU-Z is programmed to read correctly.

Here is mine.

If anyone is in the same situation gime some advice. Tomorrow night i wanna play again with this chip, and i hope i can pass the 518FSB wall. But i don't want to kill my chip and i don't have a DMM right now.

[audiofreak]

This is why I have held off on an 8xxx chip. The degradation of clock speed really worries me. Intel has a great OC rep, which is why some of us buy their cheapies. I think this is much worse than AMD and the errata bug in Phenoms. At least they dont die and people have pushed a whole lot of voltage into them to try and clock them high.

They also aren't done in 45nm.

[unclewebb]

The IT8718 is a very common motherboard monitoring chip which I believe is used on many Gigabyte boards. CPU-z definitely supports that chip.

http://www.cpuid.com/hwmonitor.php

[JohnZS]

I've tried flashing my P5E3 v0703 to the latest version which is 0808 and for some reason my system is not stable at all
It has not made any differences in the selection of FSB Termination or CPU PLL voltages, perhaps ASUS are not aware of the datasheet? or 45nm support is still ongoing until we see the mainstream Quad 45nm cores?
John

[Plaicd]

I've tried flashing my P5E3 v0703 to the latest version which is 0808 and for some reason my system is not stable at all
It has not made any differences in the selection of FSB Termination or CPU PLL voltages, perhaps ASUS are not aware of the datasheet? or 45nm support is still ongoing until we see the mainstream Quad 45nm cores?
John

That is strange. bios 0808 is dated 1/02/2008 on the ftp server. ftp://ftp.asus.com/pub/ASUS/mb/socket775/P5E3/

While the p5k deluxe bios 0705 that has different voltages for 65/45nm is dated 12/20/2007.

[alex17 GTX]

Another member of XS has the same problems with DFI Dark. It seems that this board overvolts when it enters windows (about +100mv). I'm not sure he used a DMM, but if this is true than both bios and software read the real vcore. And we (dfi dark users) have to wait for a new bios that will fix the problem.