[MagicAndre1981]

Andreyash, on 14 October 2011 - 05:27 PM, said:

done, doc! what's my diagnosis?

surgery successfull, patient dead

No, it was a hassle to open the large file (my 8GB were not enough. Windows had to page out 4GB). But the ACPI.sys driver is the cause.

This maybe a BIOS, power plan or an overheating issue so that the CPU throttles. Try to update the BIOS, change to high performance power plan before playing and also make sure that the laptop is not to hot.

[nitroshift]

I'll just bump in to say that I've had troubles with the very same ACPI. sys with a laptop and simply removing the battery stopped the issue at once.

[Andreyash]

Many thanks for looking into this issue further. The file was freezing my PC as well

So in the end it is same annoying ACPI.sys? I was hoping the xperf analysis can point to a particular driver, saying, look, here is the bastard, update it/remove it!!!
Well, I have done all the operations before: updated BIOS to latest 412 version, changed power plan as well as numerious features inside it (also tried uninstalling Asus Power4Gear Hybrid), and my laptop stands on a comfortable stand while I lay in my bed - no conditions for extreme tempertures (I also monitor temp: CPU 75, GPU maximum 65)

I am falling into despair.

SOme other observations: when I play on integrated Intel HD 3000 video (yeah its powerful enough to give e decent STALKER picture) the porblem does not happen. It only appears when I switch to Nvidia Geforce GT550m (I have the latest drivers here). Subjectively it now seems to be an overheating issue with Nvidia as why the heck it starts after at least 10 minutes of play but never at the very beginning? Or could it be the Optimus issue?

Also I would want to change setting inside BIOS, but it's a rather a grap than a normal BIOS - AMI BIOS from American Megatrends - it has almost no features to change: no chipset features, no power management to say noting of the ACPI settings.

Thannks, Nitroshift as well for your suggestion. I tried removing battery with no effect. Only reboot (not always) and going into hybernation help to unlock my CPU.

ANy other suggestions/experiences?

This post has been edited by Andreyash: 16 October 2011 - 03:16 AM

[MagicAndre1981]

It may also be a IRQ sharing issue. Please make a smaller trace and give it to me so that I can look deeper in it. The other one was simply too large.

[Andreyash]

Will do, Doc. I need to install back the SDK tools, as I have just uninstalled them in despair thinking that they are of no use

I was also testing/stressing my Nvidia GPU with FurMark - the result is always the same - whenever the temperature of GPU reaches a certain level (seems like about 78 C) - hardware interrupts start to lock the CPU.

[MagicAndre1981]

Andreyash, on 16 October 2011 - 06:23 AM, said:

whenever the temperature of GPU reaches a certain level (seems like about 78 C) - hardware interrupts start to lock the CPU.

ok, so I'm right. The laptop overheats and the CPU throttles which generates the ACPI issues.

[allen2]

My 2cents here: 78°C is very hot for a GPU and it might be the root cause. I would try to put thermal paste after cleaning properly the heat sink of the gpu.
Any electronic component will see its life reduced when going to temperature higher than 70°C. Recent one are sometimes designed to be more robust and handle 80°C but if you can keep it cooler it should live longer.

This post has been edited by allen2: 16 October 2011 - 06:44 AM

[Andreyash]

MagicAndre1981, on 16 October 2011 - 05:55 AM, said:

It may also be a IRQ sharing issue. Please make a smaller trace and give it to me so that I can look deeper in it. The other one was simply too large.

I had noticed I have an IRQ conflict between keyboard and motherboard, but doubt that it can be the cause of my problem

Anyway, here is the 10 times smaller latency file:
http://www.mediafire...i1hvv4vvukkvlja

Allen2 - 78 C is what I see only in FurMark test program - it stresses my GPU to the limits (max what I saw was 82 C). However when playing a game the max GPU temperature that I see is about 65 C. However could be that it drops quickly before I am able to switch to the CPUID monitor

Andre, in my mind the "ntkrnlmp.exe!KiIdleLoop" is causing high CPU consumption. At least as it looks in the latency.etl file

[MagicAndre1981]

no, the Idle Loop is ok. This is the system idle process you see in Process Explorer.

If you dump the DPC/ISR usage you see this:

Quote

CPU Usage from 0 us to 40614785 us:

CPU 0 Usage, CPU 1 Usage, CPU 2 Usage, CPU 3 Usage,
usec %, usec %, usec %, usec %, Module
23913355 58.88, 989442 2.44, 1725432 4.25, 9346 0.02, ACPI.sys

So the 1st Core is used 59% by the ACPI driver. Run MSI Afterburner in background it shows how high the temps are.

[Andreyash]

Thanks, MagicAndre, the MSI is a very good prog. Here is what I traced:

I was displaying both GPU temp and GPU load
while playing the temp was between 70-74 C and GPU load 95-99%. When the interrupts started the temperature was about 72, however shortly after that it felt down to 60 and GPU load also decreased to miserable 12%. How come? Does it mean that CPU is no longer capable to utilize GPU fully because it is "busy" with hardware interrupts? Or does the GPU fail at first place causing hardware interrupts which in turn affect CPU

[MagicAndre1981]

Andreyash, on 16 October 2011 - 11:17 AM, said:

When the interrupts started the temperature was about 72, however shortly after that it felt down to 60 and GPU load also decreased to miserable 12%. How come?

I think this is an overheating protection. Maybe the threshold is 70°C and the laptop reduces GPCU/CPU clock and voltage to cool down.

[allen2]

I think you are in this case. The 105°C is probably for desktop graphic card.

[Andreyash]

allen2, on 16 October 2011 - 12:22 PM, said:

I think you are in this case. The 105°C is probably for desktop graphic card.

My GPU temp while playing was only 74 C. And I cannot find a reference to normal working temperatures for the Nvidia GT550m, beside a general statement "Graphics card temperatures typically range from 40°C to 90°C". There are no settings in power management or Nvidia panel as to temperature policy

Anyway, thank you very much all for taking your time to help me understand my stupid problem. I will not give up of course. The range of suspects has narrowed down to Nvidia card or a temperature handling by my laptop

[MagicAndre1981]

Andreyash, on 16 October 2011 - 12:55 PM, said:

The range of suspects has narrowed down to Nvidia card or a temperature handling by my laptop

it is the temperature. The Intel GPU which is inside the CPU is not so fast and doesn't produce such heat. That's why you don't have any issues when using the Intel.

I googled a bit and other users also have extreme framedrops in games because of throttling when the temperature is over 70°C. You can try to diable the Intel Turbo Boost in the BIOS, maybe this helps the reduce the temperature.


Some users reported that it may help to set the max frequency to 99% in the advanced power plan options to reduce the throttling with your laptop. Also try other nVIDIA drivers.

This post has been edited by MagicAndre1981: 17 October 2011 - 07:02 AM

[Andreyash]

Many thanks, Andre, I will try both options. For BIOS I do not have any possibility to change power managtement or chipset features as BIOS is AMI and its almost optionless

But based on your advce I have also found this interesting discussion about Sandy Bridge throttling although on slightly different laptops:
http://www.asusrog.c...-and-Workaround

The prog is ThrottleStop, which is supposed to prohibit CPU from throttling down (disabling BD BROCHOST), but my problem is that BD BROCHOST appears to be disabled by default. What I can try with this program is to disable Turbo Boost

I will inform of the resutls especially if I have any luck cause it may help many other people out there with similar problems (FPS drops, hardware interrupts because of ACPI.sys, etc)

[Andreyash]

well I tried everything: switching off Turbo boost, lowering down % of CPU usage in advanced power management setup, but nothing works. When using FurMark CPU still gets locked by hardware interrupts/ACPI.sys approximately when GPU temperature reaches to around 80 C.

I have also found an interresting story line about how Nvidia makes friends with Intel lately, and this is very dissapointing as it might be an embeded problem with many PC configurations like mine with neither party willing/able to solve it:

http://forums.nvidia...howtopic=194023

This post has been edited by Andreyash: 17 October 2011 - 09:01 AM

[MagicAndre1981]

I have no real idea. Contact the ASUS support and inform them about the temperature and throttle issue.

Also google a bit more and read what other users did.

[footprints]

Hi! This post is old, but I still hope that any of the earlier participants will see this.

I've just read this post after searching through the net for a solution. I have the exact same problem with the exact same pc as you are describing here. What did you end up doing?

The strange thing for my part is that I discovered this problem when I bought the pc over a year ago, but it worked it self out after some reinstalls and random installation of drivers from asus' support page. But after a reinstall and a new ssd, the problem is back again. I've tried everything of drivers, different bios versions etc. like you describe, but everything comes down to some ACPI-stuff. Frankly, I don't know what to do with this now. Also all other threads about this in other forums (yes, there are quite a lot of them) end with out any solutions. Do people just live with this?

Best regards,
Footprints

[MagicAndre1981]

which Windows do you use? If you use Vista or Win7, try the xperf commands.

[footprints]

Thanks for your reply!
I'm a bit confused with xperf, could you give me a short summary of which commands i should use and what i should look for? Also, I'm using Win7 64bit.