E8400 temp problems

[krunt]

at a 400mhz fsb (3.6ghz) I am pulling 63c on both cores on 100% load (prime). This is acording to Core Temp.
SpeedFan 4.33 is giving me three choices of temps, temp 1 at 65c, temp 2 at 34c and temp 3 at -128c. the HD, core and ambient all seem close enough.
Everest says CPU at 65c (like the speedfan temp 1), core 1 at 61c core 2 at 61c. Aux at 34c. No GPU because it is a trial version, but the ambient matches speed fans at 41c.

So yeah its hot, but im also running prime. The odd thing about this set up is that at idle i usually have a 2-3c difference between the cores, sometimes as much as 5 degrees. But they even out around 50 or so.

Don't take my numbers for anything more then what they are.

 

[aussiestilgar]

So should we ignore these temperature readings and OC happily into the 60s and 70s, or RMA our chips?

 

[Tempered81]

who knows, I'd like a better choice of temp monitoring progs. what do we have to compare with, asus probe, everest, coretemp, speedfan, anything else?

well anyway, im selling my e8400 desktop rig, and getting a laptop, i dont think it's going to have a mobile penryn in it that i can overclock. However, I'm sure that this will all be fixed via software & bios within a few months, or the worst case scenario will be true: that the chips temp diodes are buggy.

 

[krunt]

well im running another prime now, at a lower voltage. This time I am 9x400 for 3.6 at 1.175v cpuz reports as 1.100. I have noticed in the first 20 mins or so of prime that my tempatures are much lower at load, though the idle temps before were about the same. We shall see if its stable. I have tested up to 1.1375 and can't get prime to run without an error. jumped to 1.175 and so far so good, but a long way to go.

Now I am at 50-52c or so according to core temp. speedfan is giving me 52-54c. (don't ask me why, but the temps are still bouncing around a few degrees even after a half hour of prime)

As for me, ill happily OC into the 50s. I won't go into the 60s unless it is 60-61 and I am running a 12 hour prime stress. I don't think this is a reason to RMA a chip though. The fact that we can only pull 20-30% overclocks on air is not that much of a downer to me.

 

[Xvys]

A couple of E8400 owners with wonky temps were told by their retail outlet to ignore the core temps and just keep your eye on the cpu temp, which apparantly is accurate. I checked the Intel site and apparantly their TAT (Thermal Analysys Program, which was very accurate for the Conroe's and would expose an unstable system very quickly)... was only released under a Non-Disclosure Agreement for software developers and apparantly was never intended for use by schmucks like us.

 

[Tempered81]

man i see no prob with this chip above 75 Load. being as intel spec shows the thermal ID at 72.4C for the e8400s and the old conroes were ID'd at 61C. People used to push thier conroes to 70C all the time. many people ran 24/7 for a year with their load temps reaching 71 to 74C. I wouldn't see a problem with these wolfdales until 80C. and even then thats 80C at 24/7 long term with regards to electro migration and degradation problems. The high-k metal gate tech is new stuff though. I wouldn't recommend going above 65 until someone makes some definite conclusions.

 

[BlueWeasel]

Originally posted by: Xvys
I'm sure I read this on several sites, but UncleWebb had some comments about the core temp bug in this thread, (started by a very wise fellow )

http://www.xtremesystems.org/f...howthread.php?t=173909

Thanks for an informative post on the subject. I spent the time reading through UncleWebb's posts and it provides a detailed look as to why the E8400 is experience buggy temperature readings.

I've only had the chance to run my E8400 at stock speeds with the stock heatsink, but once I start overclocking, I'll feel better about the variance in temps.

 

[Tempered81]

quoted from the guy over at XS:

"Just trying to show that there is no need to worry when you see a big temperature number as long as things are running stable. The new E8400 has even more headroom than the previous generation, Conroe, desktop processors."


this is my opinion as well he even said you can goto 85C no prob, watch out for temperatures reported negatively like -127. I said 80C would be no prob. - he ran one to 97C

Shutdown is supposed to be between 85C-90C

he also explains how the temps are wack in coretemp

 

[imported_wired247]

Originally posted by: krunt
at a 400mhz fsb (3.6ghz) I am pulling 63c on both cores on 100% load (prime). This is acording to Core Temp.
SpeedFan 4.33 is giving me three choices of temps, temp 1 at 65c, temp 2 at 34c and temp 3 at -128c. the HD, core and ambient all seem close enough.
Everest says CPU at 65c (like the speedfan temp 1), core 1 at 61c core 2 at 61c. Aux at 34c. No GPU because it is a trial version, but the ambient matches speed fans at 41c.

So yeah its hot, but im also running prime. The odd thing about this set up is that at idle i usually have a 2-3c difference between the cores, sometimes as much as 5 degrees. But they even out around 50 or so.

Don't take my numbers for anything more then what they are.

I am at 3.6ghz and I have the exact same temps as you, according to the asus AIsuite. I could not get coretemp to work in vista64

 

[jrox]

Originally posted by: rpenguin
Hey,

I just built my new system Saturday.

E8400 stock
Big Typhoon Cooler
Antec 900 Case
Gigabyte GA-P35-DS3L (with newest bios)
2GB Crucial Ballistix DDR2 1066
BFG 880gt

I had the stock cooler on my e8400 until today. I removed the included heatpad and applied arctic silver 5 instead. I was getting temp's in the bios of 70C.

I check in Core temp and I was getting 70-75 in there as well, and when I run prime95, it would go past 100C to where i would shut it off.

Today I installed the big typhoon expecting a HUGE drop in temps, but I didnt get what I expected. Core temp and bios still reads 70, except now it hardly changes. even while being fully used in prime95, it stays at 69-71. I checked Everest and I was getting the same temps as all my other programs.
~70 core1 and core2
but then the "CPU Temp" was around 22c-25c.


I don't know what else to do. My case has excellent ventilation. My 8800gt runs around 51c under load. Do I have a faulty chip, or should I really believe these ratings. My main purpose in buying this chip was overclocking, but I want to know what my actual temperatures are.

Please help... Thanks!

Hi rpenguin, I just registered here to tell you that I am having the EXACT same issue. I have the same mother board as you also with the newest bios. I'm still using the stock heatsink. When i'm idle i get temperature readings of about 34C, but if I run orthos for 30seconds or so my temp readings jump up to 90C. I've been thinking about trying to reseat the heatsink but as I can see you added a new cooler didn't do much.

What i AM going to try to do is downgrade to the F6 bios. I spoked with someone who has the same cpu/mobo/case but is running F6 and has more reasonable temp readings.

I'd like to stay in contact so if one of us figures out a solution we could help each other out, so PM me if you read this.

 

[krunt]

Originally posted by: jaredpace
this is my opinion as well he even said you can goto 85C no prob, watch out for temperatures reported negatively like -127.

I read that too, and it worried me because speedfan 4.33 reports my "Temp3" as -128c. Core Temp 0.96.1 is giving me 55c on both cores. while speed fan temp one is 56c. temp 2 is 33c The hd, gpu core and gpu ambient seem realistic enough as well.

So, what the heck is temp3?
On configure temp3 has the same info as temp 1 and 2, except fotthe sample and sensor information. All three on the ISA bus, same adress, same chip. no distinguishing information whatsoever.

So either, i am much hotter then i thought on one of the cores and coretemp is wrong. or speedfan is wrong. As it gives me -128 at idle and at full load, im guessing speedfan is wrong. At least coretemp changes with load.

Anyone else with a big negative in their speedfan temps?

 

[kronusthebonus]

I'm running a stock E8400 with an Arcitc Freezer Pro 7 and I'm seeing 39/41C at Idle and 50/51C on load at STOCK settings. Perhaps the reporting is wrong, or something. I am going to try it with a different heatsink tonight to see if it was the Freezer Pro (seemed a little rough on the bottom)

 

[aussiestilgar]

Looks like the general consensus is that the DTS (Digital Thermal Sensors) on the core die seems to have a high fault rate with the wolfdales. Basically that means the temps reported by them (and programs such as Coretemp) can't be trusted. Lenovo even sent back a load of 45nm mobile processors because of bad temperature readings.

However, they OC like crazy and temps are not a limiting factor... so I guess we should take it with a grain of salt and just exercise some sense when OCing. We'll see if Intel will have anything to say about it.

 

[VirtualLarry]

Originally posted by: jaredpace
or the worst case scenario will be true: that the chips temp diodes are buggy.

I vote for that. Look at AMD, their 65nm X2 chips have buggy core temp sensors, compared to their 90nm brethren.

It seems like it was more difficult to scale the (analog) circuits in these temp sensors down to the next smaller process size, and keep it working.

 

[jrox]

Assuming the temperature sensors are faulty, will this reduce the performance of the cpu?

I just read this on the intel website:

http://www.intel.com/cd/channe...asmo-na/eng/299986.htm

The Intel Core 2 processor uses the Thermal Monitor feature to protect the processor during times where the silicon would otherwise operate above specification...The feature is intended to help prevent long-term reliability damage to the processor and provide protection for unusual circumstances like higher than normal internal chassis temperatures (and inlet air temperature, defined as the air temperature entering the processor fan heatsink) or failure of a system thermal management component (such as a system fan). In its active state, the Thermal Monitor feature scales back processor power consumption, if the factory programmed thermal design temperature is exceeded While the Thermal Monitor feature is active, the system's performance may drop below its normal peak performance level.

Is this referring to the thermal speed stepping settings on the motherboard (something that could be disabled in the BIOS) or is the CPU doing this independent of the chipset (something I have no control over)? If the latter is true and the cpu thinks its getting bad readings from the temperature sensor is it going to scale performance back?

 

[Xvys]

I believe the thermal design temp for the E8400 is 72.4C and this refers to the thermal throttle feature which is activated if the cpu temps are exceeded. I had thought the throttle-down was related to the core temps, which have a max. of 100 or 105C. But this article sound like it is the cpu temp, not the core temps which activate throttle-down. The C1E and SpeedStep are other features which reduce cpu voltage and multiplier when at idle.

 

[jrox]

Originally posted by: Xvys
I believe the thermal design temp for the E8400 is 72.4C and this refers to the thermal throttle feature which is activated if the cpu temps are exceeded. I had thought the throttle-down was related to the core temps, which have a max. of 100 or 105C. But this article sound like it is the cpu temp, not the core temps which activate throttle-down. The C1E and SpeedStep are other features which reduce cpu voltage and multiplier when at idle.

Both my core temps and my cpu temp read very very high when the cpu is stressed. Everything runs very stable still. But if the cpu is slowing itself down if when it thinks its too hot I might want to RMA it.

 

[VirtualLarry]

Originally posted by: jrox
Is this referring to the thermal speed stepping settings on the motherboard (something that could be disabled in the BIOS) or is the CPU doing this independent of the chipset (something I have no control over)? If the latter is true and the cpu thinks its getting bad readings from the temperature sensor is it going to scale performance back?

The CPU performs the throttling, not the chipset. (*)

(*) Intial throttling performs a duty-cycle adjustment to the clock signal, but if things get too bad, the multi is dropped and so is the vcore. That feature might require mobo support.

Download and run RMclock, it will tell you if you are thermal throttling.

 

[toadeater]

Heat probs reported for Intel E8400s

 

[Binky]

Abit IP35 Pro (tried bios 14, 16.b03, 16.b04)
E8400

I have shelved my E8400 and put my trusty E6420 back in. Partially locked CPU temps of 3C, 29C, and 49C annoyed the hell out of me. The 0.9v vdroop at idle on this mobo is also odd, and the load vdroop is even worse. I like my automatic fan control (based on temps) too much to keep this chip in until they sort this mess out.

 

[jrox]

Originally posted by: VirtualLarry
The CPU performs the throttling, not the chipset. (*)

(*) Intial throttling performs a duty-cycle adjustment to the clock signal, but if things get too bad, the multi is dropped and so is the vcore. That feature might require mobo support.

Download and run RMclock, it will tell you if you are thermal throttling.

I downloaded and ran RMClock, it reported no throttling at any time. I was however confronted with an unsigned driver issue that I could not get around in vista64 (tried all the workarounds including using F8 at bootup), so I'm not sure if it is getting good data.

I did some tests to check for performance degradation due to the very high temperature readings. I first disabled all the BIOS options having to do with speed stepping and confirmed that my voltages and multiplier were constant with cpu-z.

Using 3dmark06 my cpu will score at around 2350 if the cpu was cool before I started the test, but if i run tests back to back the score drops to around 1800.

I could get 2350 every time I waited 5 minutes between tests, and 1800 every time I ran multiple consecutive tests without waiting for the cpu to cool off.

I've never done a test like this with another cpu so I don't know if that is normal behavior or not.

 

[bka4u2c]

I finally had a chance to mess around with overclocking the other day. Set FSB to 400 (linked/Synced) and Vcore to 1.225 if I remember correctly(I'm at work). Left all other voltages at AUTO. No problems so far. Ran prime for about 3 hours with no errors. Looped 3DMark06 And got my highest score 3DMark06 yet. 17141.

Temps really didn't change much from what I reported before. It's the weekend so I may see how far I go on that voltage before I get error or crashes. Overall I've been impressed with response of this MB/CPU.

 

[Darkphantasm]

Hello all, been doing some research and i guess these e8400 procs have faulty sensors or whatever im not big into overclocking but just checking my temps using Nmonitor proc runs 40-44 idle and under full load playing mass effect runs at 85C! i have the stock hsf on plan on gettin a new one but didnt think it would run that hot with 45nm technology so my question is how bad will it hurt my proc running it at that temp for says 3 or 4 hours? Should i not worry about it since its faulty sensors? newbie thxs you for your wisdom! =p

 

[Crittias]

85C sounds pretty extreme, and the swing (from 40 up to 85) is also excessive. If I were you I might check to make sure my current HSF is seated correctly on the CPU.

 

[Darkphantasm]

done it a couple times now still the same and thats under full load but still sounds very extreme even if the sensors are messed up but people said it can run 85C and be fine since its a faulty sensor, heard of people getting it to like 110c or so before it shuts itself down and underclocks itselfs but ya i dont know to much about these procs just built a new machine after a verylong time its a 780i evga board e8400 bfg 8800gts freaking proc runs hotter than the video card but its the sensors again anyone know a good program to read these procs?