Fuzzy GME965 overheating issues and poor BIOS support

Status
Not open for further replies.

kubunteando

New member
CORPORAL
Joined
Aug 4, 2008
Messages
11
I have bought a fuzzy GME965 and installed an Intel Mobile Penryn T9300 2.5GHz.

I have some severe issues that prevents the use of the computer:

1- overheating issues. Even when the computer is idle (CPU usage under 5%) the temperature is of 67 Celsius while laptops run around 47 Celsius with the same processor.  I have reinstalled the heat sink with the thermal paste (cleaned well the surfaces of the CPU and the heat sink before applying the paste). The Penryn Mobile processor has a IHS (Heat Spreader) but the Heat Sink that comes with the mother board is flat so the contact between both is just about 1 square centimeter covered by thermal paste.

2- no BIOS EIST support (Enhanced Intel SpeedStepping). So the CPU always is used at full speed regardless the CPU usage by the applications. Which increases the heat at idle.

3- no ACPI BIOS FAN support. So the FANs run all the time. The Operating System cannot stop the FANs or increase the FAN speed

4- no ACPI BIOS Thermal Zone support. So the Operating System will know that the CPU is overheating

The is a motherboard based on mobile technology which should run cool and be quiet.
But I feel a bit cheated since it overheats and the FANs are running all the time (even from the beginning when the CPU is cool).
The usual Mobile features like Thermal Control, FAN control and SpeedStep are not supported on the BIOS.

Do you have suggestions or hints?

I have opened a ticket with MSI about the BIOS support of the SpeedStep and FANs.

But I am really worried about the overheating since I am afraid to use the computer for a few hours, and I cannot  run medium intensive applications (using 50% pf CPU) because the temperature rises over 93 Celsius. I don't want to burn the CPU.
The BIOS is version 1.20, the last one released.

Thanks.

Juan
 
Hello,

let us look at your problems step by step:

1- overheating issues. Even when the computer is idle (CPU usage under 5%) the temperature is of 67 Celsius while laptops run around 47 Celsius with the same processor.  I have reinstalled the heat sink with the thermal paste (cleaned well the surfaces of the CPU and the heat sink before applying the paste). The Penryn Mobile processor has a IHS (Heat Spreader) but the Heat Sink that comes with the mother board is flat so the contact between both is just about 1 square centimeter covered by thermal paste.

A couple of questions:

1) What tool did you use to monitor the processors temperature?  Are you talking about the core temperatures or the CPU Temperature values?

2) The board is not responsible for cooling the processor.  If the temperature values you are reading are too high, we are either talking about a BIOS problem (does not apply to Core Temperature values) or about insuffient cooling or a heatsink that is not properly installed.

but the Heat Sink that comes with the mother board is flat so the contact between both is just about 1 square centimeter covered by thermal paste.

Can you take a picture and upload it here?  Are you saying that the concavity of the CPU's heatspreader is preventing the heatsink to make sufficiant contact?

2- no BIOS EIST support (Enhanced Intel SpeedStepping). So the CPU always is used at full speed regardless the CPU usage by the applications. Which increases the heat at idle.

So there is no option in BIOS to enable EIST/SpeedStep and there is no option for C1E-Support either?  Press F4 while you are in BIOS an browse through all BIOS sections again to double-check.

Note that EIST needs OS support in order to function.  In Windows, you need to select an energy saving profile to turn EIST "ON".

4- no ACPI BIOS Thermal Zone support. So the Operating System will know that the CPU is overheating

Intel Core/Core2 Processors have a built-in mechanism that will automatically reduce CPU Frequency and VCore when the processor runs too hot.  If that does not help to cool it down, the processor will automatically shut itself down to prevent severe damage.  This mechanism is triggered by a certain Core Temperature threshold that depends on the CPU model and exemplar in question (TJunction).  I am not aware of ACPI Thermal Zone Support for any recent Core/Core2 mainboard and I don't see why it is needed.  You are the first user that I meet who is asking for such a thing.
 
T9300 isn't listed as supported: http://global.msi.eu/index.php?func=prodcpusupport&prod_no=1518&maincat_no=388&cat2_no=&cat3_no=#menu
 
Some answers here:


1) What tool did you use to monitor the processors temperature?  Are you talking about the core temperatures or the CPU Temperature values?

I am just checking the temperatures the BIOS is showing after I reboot the computer while having worked (CPU use under 5-10%) with it for around 1 hour.
    The BIOS shows "CPU Temperature" and those values are the ones I am checking and reporting.

2) The board is not responsible for cooling the processor.  If the temperature values you are reading are too high, we are either talking about a BIOS problem (does not apply to Core Temperature values) or about insuffient cooling or a heatsink that is not properly installed.

I have installed the heatsink around 3 times changing every time the thermal paste. No appreciable differences in the CPU Temperature (maybe a few degrees). I am using the latest BIOS available.

Can you take a picture and upload it here?  Are you saying that the concavity of the CPU's heatspreader is preventing the heatsink to make sufficiant contact?

I was surprised to see the heat spreader when I installed the CPU. The heat spreader is about 0.5mm higher than the case, and it has a surface of a bout 1 square cm. The heatsink is fully flat so it will only touch the CPU case on the heat spreader. I don't know if this is normal, just I remember that when I was helping a friend to install his Penryn CPU for a Desktop computer I don't remember having seen the heat spreader...

My case looks exactly like these pictures:
http://www.geocities.com/_lunchbox/remove_integrated_heat_spreader.html
So I guess it is normal?

So there is no option in BIOS to enable EIST/SpeedStep and there is no option for C1E-Support either?  Press F4 while you are in BIOS an browse through all BIOS sections again to double-check.

Note that EIST needs OS support in order to function.  In Windows, you need to select an energy saving profile to turn EIST "ON".

No BIOS options for EIST or C1E. I think this motherboard must have some simplified BIOS. F4 does not open any tab in the BIOS...
I am using Linux. In some laptops with the same processor BIOS updates have solved the EIST issue. But here I think EIST is not supported in the BIOS.
I found it strange since it is a motherboard with a Mobile chipset GM965.


Intel Core/Core2 Processors have a built-in mechanism that will automatically reduce CPU Frequency and VCore when the processor runs too hot.  If that does not help to cool it down, the processor will automatically shut itself down to prevent severe damage.  This mechanism is triggered by a certain Core Temperature threshold that depends on the CPU model and exemplar in question (TJunction).  I am not aware of ACPI Thermal Zone Support for any recent Core/Core2 mainboard and I don't see why it is needed.  You are the first user that I meet who is asking for such a thing.
[/quote]


I think the Thermal Control must be more interesting for laptops when you want to stop totally the FANs when the CPU is idle.
In the BIOS I can set the FAN Thresholds for the CPU temperatures and System temperatures. But there are 2 issues:

a) The FANs will never stop, not even when the CPU is cool (before overheating)
b) The temperature values are not available to the OS via ACPI. That may be useful to trigger alarms, or control the FAN activity.

Now that you mention the VCore, when I first powered the motherboard it came with an older version of the BIOS. Then I updated it the newest available from MSI. I had the impression the VCore values reported by the older BIOS were different than the current one. But the older BIOS did not support the Penryn processors so probably it means nothing.

Any ideas/comment is appreciated.
I don't know really what to do. This motherboard is exactly what I want. A mobile chipset with a PCIEx16 to install a graphic card for 3D graphics (I haven't installed it yet). But from all the possible issues I could have with a mobile motherboard the one that never crossed my mind was overheating. Live to see :-)

Juan
 
I think the list of CPUs is not updated, because in the "Key Features" says "Support Intel? 45nm Mobile Core? 2 Duo processor" and the 45nm are the Penryn processors. The ones in the CPU list are Santa Rosa 65nm CPUs.

My exact motherboard is:
http://global.msi.eu/index.php?func=proddesc&prod_no=1267&maincat_no=388

In that motherboard in the BIOS tab can be seen that the last BIOS update was made to support the Penryn CPUs. The T9300 being listed there.

Thanks,

Juan
 
Don't forget that Fuzzy series of boards are industrial boards, not really aimed at home-user market. So don't expect to see same range of BIOS options as you would on a board for a home/office desktop machine.
 
kubunteando, if I was you I would make the following test:

- parallel installation of Windows XP/SP2
- check if EIST works here to make sure that we are not looking at an EIST/Linux problem
- run Core Temp and Realtemp to get a look at the CPU Core Temperatures (if they are significantly lower than the CPU temperature values, we are probably  looking at some sort of wrong BIOS reading for the CPU temperature (Core Temperatures are determined by accessing specific CPU registers that store temperate data based on Digital Sensors within each CPU core, not via BIOS)
- if the value for CPU temperature in BIOS is simply a few degrees off, the issue could eventually be fixed with a BIOS Update, but it is probably not bad idea to double-check first
- if the CPU Temperature value is too high, even though the processor cores run significantly cooler, this may explain your fan problem as well.  BIOS/board fan regulation is usually based on the CPU temperature, not on Core Temperatures.  If the value is too high, the fans might run @full speed simply because of that.
 
I have mixed results after the test:

* Windows XP SP2:

     - CPU frequency scaling works
     - CPU Core temperatures are around 32 Celsius when Windows is idle (measured with Real Temp)
     - When running the CPU at 50% (1 core fully used with the 2nd test of Real Temp run repeatedly) the CPU Core temperature reaches 60 C. Just after that I reboot the computer and in the BIOS the reading of CPU temperature is 65 C.

* Linux

     - Frequency Scaling not working
     - No other CPU power saving features are working

I run the test with the computer case opened because if I close it the CPU temperature rises around 10 C.

I think the issue with Linux partly is because of the BIOS. I have disassembled the BIOS code and "LInux" is one of the OSs listed, so I guess that the BIOS at least recognizes the Operating System, but it seems that does not support the Power Management features. Like many other BIOSs.

Do you think there are chances MSI will fix an issue with the BIOS and Linux?

Still I see these issues:

1- the temperature of the CPU is still too high in Windows. 65 C when the CPU is used at 50% is already over the temperature reported by other laptop users with the same processor and running the CPU at 100%!. And the measure was done with the case opened!

2- the FANs don't stop working even the temperatures are lower than the thresholds. This seems to be a BIOS issue.
The FANs spin down a bit but they don't stop.

Regarding the BIOS feature support I consider that when I buy a product belonging to MSI Industrial Solution I would get a Premium product which should have at least the same features as the ones destined for consumers or some more. Also in the Industry is common to use Linux so I would expect better Linux support.

The Penryn processors are known to run cooler than the Santa Rosa, so I would be interested to know what temperatures other owners of the motherboard would get with SantaRosa processors.

I have seen that the Vcore is 1.112V while Penryn can run at 1.062V. With that there will be a gain of a few degrees.
But it seems I cannot do it from Linux since there is no ACPI support... :-(

I am thinking of trying:

- use Artic Silver 5 as Thermal Paste. I ordered it already.
- check how I can improve the air circulation in the case so I can refrigerate better the CPU and motherboard

Please, let me know your thoughts.

Thanks,

Juan




 
A few side questions:

    - CPU Core temperatures are around 32 Celsius when Windows is idle (measured with Real Temp)

Have you used Core Temp as well to get a "second opinion"?  Also, please use SpeedFan and/or Everest to have a look at the CPU Temperature Value (not Core Temperature) to check the difference between Core Temperature and CPU Temperature, when the system is IDLE.

Also, compare both kinds of temperature values under full load.  Use Prime95 to stress all cores equally.

------------

- use Artic Silver 5 as Thermal Paste. I ordered it already.

When you remove the heatsink to apply the new thermal paste, can you take a picture that illustrates this a little better?

The Penryn Mobile processor has a IHS (Heat Spreader) but the Heat Sink that comes with the mother board is flat so the contact between both is just about 1 square centimeter covered by thermal paste.

------------

2- the FANs don't stop working even the temperatures are lower than the thresholds. This seems to be a BIOS issue.
In the BIOS I can set the FAN Thresholds for the CPU temperatures and System temperatures.

Which options does the BIOS provide regarding FANs and FAN Speed regulation?  Is the temperature threshold all you can adjust?


 
I will use those tools to measure the temperatures. I am not so keen on stressing the CPUs since already once the CPU  has reached 84 Celsius and after rebooting the computer several times it refused to boot Linux nor Windows. I had to let the system cool down... I don't want to melt anything...

Here there is a picture of the CPU (on the right) and GM965 North Bridge on the left:

http://www4.picturepush.com/photo/a/991032/img/991032.jpg

There it can be seen 1 mirror surface on the center of each chip that is a bit less than 1mm higher than the rest of the chip.
Those are the Heat Spreaders and are the sole point of contact of the Heat Pipe.

On the BIOS parameters here there is a picture:

http://www3.picturepush.com/photo/a/991036/img/991036.jpg

I can set up the thresholds and the TimeStop that I suppose is the time after which the FANs will stop if the Stop Value is crossed.
Now I have set up quite low values but even with really high values the FANs will not stop...

With the Thermal Grease (finally I got the Noctua NT-H1) and moving one of the FANs on top of the North Bridge, that was getting really hot, I can have the case closed and the CPU temperature is around 64 Celsius in Linux (when idle). In the North Bridge GM965 is where Intel X3100 Graphic Card is, and it is quite hot.
I have hopes that when I get the PCI Express Graphic Card and I disable the X3100 the temperature will also go down...

But at this point still the temperatures are really high...





 
I opened a support case with MSI Support.

MSI admited that the motherboard has not been tested in a proper case (only on free air), and MSI just checks CPU compatibility and not the TDP (no wonders they say the X7800 with a TDP of 45W is compatible with the motherboard), and the GME965 is not tested for temperature issues or overheating.
This is quite bad. It seems that MSI Industrial products are not properly tested plus the BIOS are really old.

Let's hope MSI will do something about it. MSI has a long way to improve their support till I buy another MSI product.
At this point I can recommend this product to other users. But be aware of the overheating. I cannot recommend to use processors with a TDP higher than 35W.
And don't expect a low power usage as a laptop.

I have measured the power used when idle (GME965 + Intel T9300 + 2GB RAM + 2.5inch laptop HD and no other Video Card) and it is of 25W of AC (measured the AC power consumption given by the AC adapter to the system).

The strong point is that it has a PCIEx16.
 
Finally some good and bad news. But news after all.

There is a new version of the BIOS 2.0 that supports Intel SpeedStep in its basic form.
This means that the CPU is underclocked when the load is low.
The CPU I use is a T9300 at 2.5GHz and it is underclocked at 1.2GHz. I have not been able to measure again the power consumption, but the CPU runs around 3 degrees cooler. Now it runs at 55 Celsius when idle. Well, something is something.

Unfortunately the Core 2 Duo processors save much more energy when running at lower power states: C1, C2, CX... than under underclocking But the BIOS does not support those.
Currently the CPU runs always on the state C0, the one consuming more energy.
These makes me wonder: what kind of engineering is done at MSI?
Why not doing the things right instead of doing them half right?
We are talking about a mobile mother board that is intended for low power consumption.

This could be an amazing board: you could have the power of a Core 2 Duo at 2.5GHz and a PCIEx16 bus in the size of shoe box, running cool and silent.
But instead, just because of the old BIOS these is what we have:

- the FANs never stop, although the system runs cool. It is a bug in the BIOS. This shortens the life of the FAN.
- the CPU runs hot even when idle (although 3 degrees cooler than with 1.x BIOS). This shortens the life of the CPU.
- the CPU core voltage is of 1.115 volts which is huge for the 45nm processors and makes them overheat. This shortens the life of the CPU.

The GME965 is an industrial product and the 3 issues above shortens the life of the CPU and the FAN. Is this what is expected in a Industrial product?
There you go, you have an old BIOS in a new motherboard and what do you get for your money? Definitely not what was promised.

Although MSI has shown some level of support, it is far from what is expected nowadays.
 
I noticed that the BIOS for this board has been updated to version 2.20.  Have they addressed these issues at all?
 
Status
Not open for further replies.
Back
Top