Ridiculous CPU temps after upgrade on 6930ZG

Hi,

I'm requesting help regarding my CPU. I just upgraded it from a T3400 to a T9800, and everything works perfectly but the temp sensors. They show extremely random numbers, for instance it can run at 50/60C on idle and suddenly go to 70 in a few seconds... Also it seems that sometimes the temperatures are proportionnal to the CPU usage. What I mean is that when I use a program that usually uses most of the CPU , the temps go INSTANTLY from 50/60/70C to 100/110C, thus throttling it at like half of its power (but the temps dont change) and when I close it it goes back to normal in no time... I used RealTemp, hwinfo and ThrottleStop. Note : it's cooled down by a desk fan and I removed the bottom cover.

I did some researches and it seems my only ways out are :
-Get a Core 2 Extreme and unlock the multiplier, but i don't wanna spend another 70 bucks
-Flash the bios and unlock the EIST; I did found a bios flash for the 6930G but but although it's basically the same "frame" model with different components I don't wanna risk breaking my laptop
-Or fix this darn thing !!

Any help would be greatly appreciated.

 

Try to use RMclock utility. With its help you can undervolt the CPU and reduce the temps. In addition, which thermal paste you are using? Maybe you just added too much of thermal paste on the CPU? And in addition the fan can be controlled by AcFanControl utility that helps to reduce the temps too.

Yah, 50/60C on idle is too much even for my X9100 C0 stepping. On idle my processor has 33-39C, undervolted to 1.0875V already, fan controller by AcFanControl. Thus, you should check the thermal paste, maybe undervolt the CPU. Sometimes the heatpipe can be damaged - that strongly reduces cooling possibilities of the cooling system.

 

Thanks for your reply, but unfortunately every setttings such as voltage/EIST enabling/multplier are greyed out and I can't change them. However the temps look much more realistic (stable 40/50 on idle). For the thermal paste I don't think I put too much of it, you can barely even see it. I used a brand new 10% silver "Spire" paste. I also tried to screw the HS as tightly as I could.

EDIT : Nevermind, the temps are astill abnormally high, although a little more "stable" :


EDIT2 : Now I'm just browsing Youtube and it won't go lower than 90/95, but throttling is disabled most of the time. uhhh ok lol

 

Very thin layer of thermal paste sometimes is not good too. For example, when I had GT240M - I always applied very thin layer of thermal paste. When I bought M4000, I applied alot of thermal paste and temps gone to normal. That is because of M4000`s chip is on 1mm smaller, than GT240M. Maybe that is the same thing for your processor.

In addition, did you take appart notebook, or just lifted the HS and applied the thermal paste?

About RMClock, maybe you have not checked something? Here is how my RMClock settings looks like:







About thermal paste - I recommend to use Arctic Silver 5.

 

Is the exhaust air hot at all? If its relatively cool then the heatsink might not have been put on correctly and the heat isnt being transferred efficiently.

Double check and add a tad more paste this time. The heatsink needs to be flat and snug against the CPU surface.

I cant find anything about the Spire paste you are using. I normally use AS5 or ICD7.

 

Propolkin => Thanks for clearing that out, I thought you meant this :

I set the AC Power profile on x10 @ 1.2875 and the result is interesting (x11 makes the pc freeze) :

It seems to constantly try to enbale and disable throttling, which is seeable in HWinfo too. Sometimes the frequency even goes up to 3.09 Ghz, (???) In-game, the performances are slightly better but not as good as they should be.
I put paste on it before replacing the old CPU.

flipfire => The air is warm, but not hot. However the sink is. For the paste, this is their website Spire Corp - Powered by Innovation I don't see any information about thermal paste though...

 

onche2, you should read this. That is a short manual on how to use it. On the screen, that you`ve posted, I can see, that RMClock is not integrated in OS power management.

About voltage, you should set to 1.1 V or about and test fro BSOD or laptop freez. If no BSOD or freez will be - than voltage is good for your CPU. Mine X9100 is working on 1.0875V.

Strange. When my CPU is on ~80C the air is warm from exhaust. Maybe your HS is damaged?

 

The guide you linked helped me stabilizing the CPU but not completely :
On maximum performance mode, game barely playable

On on-demand power according to flipfire's undervolting guide, game quite playable (5 to 25 fps i'd say):

Under that mode, temps can go down to near 50° on idle on with stock voltages

Also, it is integrated in OS power management (box is ticked under Management)

 

When you tick the OS power management box did you re-start the RMclock after that? You should do this - untick OS Power Management integration, re-start RMClock, tick OS Power Management integration, re-start RMClock.

After that the voltage and clocks of the CPU should be stabilized.



And remember, that RMClock doesn`t show real CPU temps. Use HWMonitor or AIDA64 for that. For T9800 Tjunction is 105 C.

 

Well I've got good news and bad news. I did what you say and launched Need for speed Hot pursuit (the game I used previously to stress test as I know the bottleneck on this game used to be my CPU) and it's running kind of smoothly (much more than when I created this thread). However my monitor still looks like this :

And even on the HWmonitor the temps are still way too high and random (105 while playing but falls down to 80 within the second after I close the game and 65 soon after)

 

Contact me in skype: propolkin. I`ll help you to figure out, what is with RMClock settings, using teamviewer.

Edited: after discussing the situation with author in skype, the situation is as follows: laptop was not taken appart, the heatsink have just been lifted up a little and the CPU have been changed without any cleaning of old thermal paste. Author is instructed about what he now should do to see his CPU working as intended.

 

Okay, sorry for the huge bump but I was busy these days (and kind of gave up, too). So I tried to fix the heatsink as propolkin said but it changed nothing, the temps still change abnormally quickly. So I flashed the BIOS, unlocked the EIST but it didn't change much. This picture shows what I have access to in the CPU BIOS settings :

Funny thing, when I try to play with the throttling-related options, it sometimes ticks "Force thermal throttling (50.0)" and sometimes "Enable normal throttling (Default (50.0%))" in RMclock in advanced settings/throttling (both are always greyd out, though).
Note : It seems ACFanControl sometimes shows more realistic temps (30 to 60) but can suddenly show unrealistic ones (90/105) at anytime.

So simply put, I'd like to disable ALL kind of throttling features (whether it's caused by Intel, temp sensors or just "normal" throttling). I know what you're going to say, "you're going to burn your PC" but don't worry, I got plenty of ways of cooling it down. And I've had this laptop for a few years now, so I know when it actually overheats.

Also, if someone helps me to repair it to a satisfactorily level (for instance clocks used at 100% whether in stress or not) I am willing to give them some PayPal money.

 

Yeah, you have really funny CPU. Maybe temp sensors are dead? It is possible, but probability is not so high. As it is new CPU, as I understand.

You have modded BIOS, right?

 

Yeah, I now have a flashed BIOS. It wouldn't surprise me that the temp sensors are dead (the old CPU also showed random temps but it was not that bad, there was no throttling and the fan was acting normal, too.)

 

Try to flash to original latest BIOS version. Maybe in modded BIOS, that you are using, voltage and cooler tables are ruined.

 

I have never seen a temp sensor fail in a CPU - I highly doubt it's even possible.
I'd put my money on issues with reading the temp properly- not that it's helpful in any way

 

I already have the most recent version.

Well to be honest I think I'm going to reinstall windows and retry to play with the settings. If it still doesn't work I think something's broken in my mobo. Sooo yeah even if I manage to fix this speed issue thing I'd be very happy

 

Theoreticaly - possible. But, as I said, but not so probable.

I don`t remeber the thread, maybe in HDX Dragon owners thread, there was users, that have had the processor with dead sensors.

onche2, yeah, in the case of unknown things with laptop, every solution can be as a cure for the problem. Give it a try and let us know about results! Maybe that will works for you!

 

Alright, just freshly installed W8. Here are the results :

1)The fan is finally acting normal! I didn't say it earlier but before, it was acting really weird. It could be on 1st speed for 5 minutes (on idle), then get 1 or 2 speed faster every 10 sec or so until it reaches the fastest speed which would stay for a few minutes. Then it would act about normal for some time and go back to the beginning. Now it seems to get or lose 1 speed at a time at reasonable time intervals.
2)Before on start-up a window would appear asking me to confirm I trust some device management control program from Microsoft (it's the kind of window that appears when you launch a program as an admin ), followed by a command prompt that immediately closes. Now it's gone
3)When I launch a game it runs much more smoothly than before, BUT the cpu freq shown in task manager is kind of low (2 to 2.5) and it never shows a total cpu use % greater than 50/60 even when stressed. However the freq is in average higher than before so it's a good thing

Tomorrow : more tests and W8.1 update setup

 

That`s are good news! Hope, that after test your lappy will work as intended! That would be strange, if there was a problem in W8.

 

Run Prime95 torture test (in place large FFTs) on all cores - you should get 100% CPU load.
Monitor CPU temperatures and frequency.

I'm pretty sure the problem is still contact between CPU die and the cooler.
It fits the symptoms and is the most likely reason.