DPC Latency Check for your M11x?

Guys,

My M11x has been experiencing skipping. I ran the DPC Latency Check and its maximum latency is over 17,000.

Could others please also run the program on their laptops so we can see if it's a common problem no one really notices? (Audiophile here)

DPC Latency Checker

 

Mine seems to sit around 150-200 with general web use, spiking at about 500. It did however reach 3009 at one point, though that was only once and I was probably loading a lot! I have been running it for about 5min.

 

Interesting utility, I saw a lot of spikes to 8000+, but my system functions normally, and the average is under 500 most of the time. I have an i7 OC'd and @160mhz with throttle stop..
I wonder if the spikes are part of some of the power saving architecture in this system and the CPU.. I do a fair amount of video editing and other work and no issues. I did fire up my old version of gamebooster just to close some drivers and there maybe were less spikes, but still 1 every 5 checks or so. My average is still sitting <500 the rest of the time. If you understand this utility more perhaps you could elaborate on it.

 

Thank you for the link - it is a fascinating piece of software.

My m11x usually sits at 150-200, seems rather stable - yet goes up to 1000 in a somewhat regular manner and it spiked up to 3295 once.

What is more important is that I actually do have unexplainable sound issues time after time - they come and go without any apparent reason at all.

What are your sound problems like? Perhaps we will be able to pinpoint the culprit together?

UPDATE: Sits at 2400-2500 during a Skype conversation.

 

DPC Latency was a pretty significant issue for the M17xR1 when it came out; audio stuttering was the main problem for my M17x:
Here is a link to the thread that has some very useful links and info in it.
http://forum.notebookreview.com/alienware-m17x/434691-m17x-dpc-latency.html
This link will detail DPC/ISR tracing which should point you in the direction of the problem.
http://forum.notebookreview.com/alienware/461247-m17x-dpc-isr-tracing.html

Below is a capture of my M11xR2 with itunes and and a YouTube video running at the same time. My spikes (1200-1600) are always at the start and/or stop of the youtube video which could be optimus related but other than that no major spikes for me in day to day use.

 

Looks like I'm a little out of the norm specs on spikes, after closing skype, etc I sit at ~250, but I am definately getting a lot of spikes that hit 9874 about every 8-9 scans (still about 1/2 of the OP)..

I'm thumbing thru services trying to shut off everything before trying disabling drivers as in the post.. but now I'm a little interested in what is causing spikes.. but as I said I haven't had sound drops, and my performance is pretty steady... Running the 266.58 Nvida drivers as well.

:Edit, first_leviathan, please post if you find any trouble making drivers- as will I

 

For those of you who are new to this and are not clear on using Windows SDK to trace out possible DPC issues, this might be an easier option for you.

This tool offers a wealth of info over sycon's DPC Checker tool.
Resplendence Software - LatencyMon: DPC, ISR and pagefault execution monitor

Be sure to review the info on how to use properly.
Resplendence Software - LatencyMon: DPC, ISR and pagefault execution monitor

Have fun. Keep in mind that external devices can also play a role in DPC problems.

 

YouTube - Stutter or strange sound on Alienware M17x

I get exactly the same problem as this guy. The problem is it only happens once every hour or so, and the Alienware rep I called refuses to rely on the DPC Latency Checker as valid evidence for my complaint.

Still, it is a bummer when you're really into the music, no?

As soon as I get home, I'll be checking BatBoy's links.

I find it interesting that our laptops have relatively high readings... the desktop my dad built himself has a maximum latency of 400 when I do the same things as on my M11x.

 

I get that as well sometimes, mainly when I am listening to iTunes. It also happens just before the computer freezes, and then bam about 2-3 seconds later it freezes. I will have to read up on DPC and see if I can fix it.

 

That video is from June of 2010. The author may or may not be using the DPC fix which was put into place by NVIDIA on the M17x. If you want to review full details on how DPC was resolved on the M17x, see the DPC thread for the M17x. Bottom line, it was an NVIDIA driver issue which NVIDIA resolved.

Also, in terms of what you would notice with audio, its going to be in the range of 10-12k. meaning, 10-12 milliseconds or higher (this was the case with the 17x). If using the sycon checker, this is where you need to start worrying. Low range yellow spikes will not push you into an area which will impact audio.

Use the checker I listed and look at the information. Its a good resource utility and can certainly help you track down a problematic driver/device.

 

+rep
Hey thanks, this saved a ton of blind guessing around.. This utility pointed me at the APCI (apci.sys), and the alien fusion service was taking a lot of hard page faults.. I stopped alien fusion and it is definitely better, but still taking a few hits.. I have updated the Chipset drivers from the original off the Intel site awhile back- any thoughts if that would affect the power management? (since some of you have not probs)

i'll thumb thru the bios to see if there is any power options I can tweek, but my ASUS had issues with the ACPI drivers kind of crapping out.. not sure what to look for that would be causing/or could fix this..

:edit, also still running the copy of MCrappy AV that came with the R2 (figured why not, got a year for free), not sure if anyone else is but a couple of the drivers associated with it are also taking page faults.. Also running the new A04 bios

 

hey, i think most people around here uninstall McAfee and use Norton or something else...count me in that group. historically for me, mcafee has slowed down my computers noticeably, so i just don't even use it anymore. then again, with any anti-virus/spam/etc, there are going to be things happening in the background quite a bit.

comcast provides norton for free, so i just use that.

 

Lol, yeah I just canned Mcaf for windows essentials.. I have been wanting to try it out anyway, and Definately cleaned up some spikes.. (must have lossed my mind for a while keeping Mcaffe.. ) Well at first start up the system was spike free, but now, after letting it sit for a sec they are back.. So some driver is confusing things, and still pointing to the APCI.. I will play with the nvidia drivers too..

 

OOOK,
Well for me I have found the source of the edit- MY) problem.. Check out the images to see what I am talking about.. After clearing up a few other things, unfortunately it is Throttle Stop that is causing my spikes. I really have no idea what this means, but perhaps uncle web might touch this thread with his gracious hand and add some input if it would mean something to him. I have included my TS settings if anyone notices I have something out of whack..

Throttle Stop Off..


Throttle Stop on (with settings)

 

Definatly optimus related spikes. I was getting Absolute maximum of about ~2000 anytime while using the nvidia card, the rest of the time was ~300. My specs are Core i7, 4GBRAM, SATAII HDD, W7x64 HP 258.96 modded INF drivers


btw Avast is way better than Windows Essentials, I was using MSE since it was in beta, and still virus's manage to exploit this program I would not suggest using it at all. The funny thing too is once I uninstalled MSE and installed avast, I had 5 trojan horses

 

If this is an Optimus issue, why is it effecting the M11x r1's? Or is this only a current R2 issue?

 

Actually at this point I disagree- I don't have this issue unless TS is running. I can confirm it over and over. Although I do agree with your AV sugguestion, Avast seems to have a better reputation..

 

DigDung: Can you try testing again but uncheck Nvidia GPU in the ThrottleStop options window and restart ThrottleStop so it isn't monitoring your GPU anymore.

On the M11x-R2 you don't need to have Set Multiplier checked so you can try testing with that unchecked too. The next test would be with ThrottleStop running but with it in monitoring mode only.

If ThrottleStop is causing the problem, I'm very interested to see if you can isolate the problem so TS can still be used to improve performance without the DPC issues that you are seeing. I haven't seen this during testing so I'm wondering if it is related to the GPU driver that ThrottleStop queries for GPU temperature information. It would be great if you could turn that off and solve this problem.

 

Hey thanks for jumping on. I Have unchecked Set multiplier, still jumps, also, I have tried to uncheck Nvida but it kept popping up (restarting fixed this, looks like when you start monitoring the GPU on my system you cant turn it off). I did turn off TS, and the errors continued,

however.. After quitting and restarting TS, it came up in the "off" mode (button shows "turn on" to activate it, nvidia option also unchecked) and there are no spikes at all. Turned it on (with NVIDA unchecked) and ran with no spikes. Now, I checked the NVIDA button and click ok.. The drops started again.. So, thank you for the input, I believe it is something with the NVIDA monitoring (and you may want to check out the fact you cant seem to deactivate it)

Also i have retested this 3 times now, as soon as I activate the GPU monitoring the latency starts again. I dont need the GPU monitoring as I dont OC it anyway, so looks like you helped me solve my prob- Thanks and hopefully this info is usefull to you as well UncleWeb-

I'm gonna make sure I do that thing we cant talk about on here for you ! Thanks again!

:edit - heres the proof!
http://www.infinitedefined.com/TSon.jpg

 

After more testing you guys are right, this isn't an optimus issue, I have had no spikes on several videos using the nvidia gpu. Here is a screen shot

dtc lat.jpg

 

Shrunk the images on this thread a bit- Also UncleWeb, I just scored 40.194 on my TS bench after leaving NVidia unchecked, thats about 1.5 seconds faster than my best score

 

Whenever you uncheck Nvidia GPU in the ThrottleStop Options window, you need to click on OK to save that change and then immediately restart ThrottleStop so it stops monitoring your GPU temperatures. I thought that would be the safest way to stop GPU monitoring.



Thanks for doing some more testing to isolate this issue. I was worried there for a moment.

I've never been a big fan of the code ATI or Nvidia uses for GPU temperature monitoring. ThrottleStop simply accesses the driver for this info. That's why it is an optional component in ThrottleStop.

 

But im not even using throttle stop. In fact I have turbo boost off in the BIOS as well. What do you think is causing this for me?

 

So can we conclude that the spikes are due to the hardware, or the software? Or a little of both? :|

 

Why are you running with TurboBoost turned off?

 

I think it's so the processor speed stays constant so it can be eliminated as a factor for the spikes? :| If that makes sense.

 

Ah. Got it.

 

Did you not investigate this with the tool I posted? Its very easy to do so. Many factors can cause DPC Latency. In order to fully understand what is happening on your system, you should investigate it on your system as its unique. You may have different drivers installed than the next person. Results *can* be very different if this is the case.

 

I second this approach!- after running this util, it showed the APCI.sys was causing spikes (power management driver), so I started looking at things that would effect Power, which eventually led me to Throttle Stop. then GPU monitoring of TS (not TS's fault, crap drivers from nvidia causing the issue).
This util turned blind guessing into useful detective work.. Run the util and post the most troublesome driver(s) it lists on this site and I'm sure people can help you figure out the problem..

 

Got it, I found my problems to be ACPI.sys and tcpip.sys. Sorry, I was a bit busy at school and couldn't read the instructions fully.

Thanks for the link, BatBoy.

 

Putting video and audio to the test hadn't shown any major hits. I was surprised given the fact I am accessing my iTunes library via usb - usb hub - external drive.

 

I am running with TB off to get more stablity as well as more battery life, I was told this is possible (I am still waiting to know if the 02 is a better bios on battery as well.)

Secondly BadBoy thank you for the program but this is pretty much the same as the program before but just tells you the process. I already knew this before, I want to know WHY

 

I really don't think there's been much evidence to support that disabling turbo boost has much impact on battery life in most scenarios due to the way it works. If you're running CPU intensive applications while on battery then it should help, but for most people that's not what they're doing. In your case that might very well be different. For example, I'm on battery right now with Outlook, Firefox (13 tabs) and foobar playing through my mp3 library. CPU utilization is fluctuating between 1% and 15%. TB is never going to kick in for me. Disabling it will have zero impact on battery life.

I've also never seen anything that indicates that disabling TB leads to a more stable system.

In either case, let us know if you see otherwise.

 

Other than someone telling me it has an effect, I see nothing but a placebo effect, although my M11xR2 does get 5 hours of battery life (with the bios stuff disabled)

 

MassivB, what driver is giving you an issue with DPC? Maybe i missed it in the post but i didn't see you list anything from the util Batboy or other people posted.. Hard to tell why if we dont know what?

first_leviathan, do you have any gpu temp monitoring software running? After what I learned from UncleWeb, that feature is crap right out of the Nvidia drivers, so anything monitoring it can cause big spikes..

 

I run the 258.96 modded inf by kop
i dont run any temps or anything aaccept maybe gpu state viewer but thats from the driver i dont see why that should cause problems. it pisses me off because im getting these errors always, Im calling dell to see if I can get a new computer because of this, I had no idea this happens so frequently aka. great find by the poster

 

ATI and Nvidia both make it fairly easy for third party software to query the driver about the GPU temperature. Intel makes CPU temperature monitoring an extremely easy thing to do and it takes up very few CPU cycles. Trying to get GPU temperature data from ATI or Nvidia drivers is not always as smooth. I have an older Core 2 laptop with an Nvidia GPU and ThrottleStop works fine on it without any DPC Latency issues so maybe Nvidia has done something to their recent drivers.

 

throttlestop was unfortunately also the culprit for my acpi.sys related spikes.

 

Someone needs to see if GPU-Z, EVGA Precision or any other GPU temperature monitoring utilities also has an impact on latency.

 

they do. On the 17xR1 back when we did have a problem with latency, I recall GPU-Z throwing DPCs when launching the app. It should level out after launching. There are many causes to DPC Latency. In order to find the cause on a given system, the research has to be done. Sometimes it involves quite a bit more than just a checker app. This is where Windows SDK comes into play. USB/Wireless devices can be at fault, it could be anything. All it takes is for a driver to become stuck in a DPC state for too long and then you have a problem.

Oh and on the subject of NV, dont jump the gun without research guys. On the flipside, its very easy for a known issue with a fix to actually become an issue again.

 

I just went back and tested with TS running. I get no spikes. I also have 2 sidebar gadgets monitoring GPU temps.
You guys getting spikes with TS, what programs besides TS are you running when recording the DPC spikes?

 

GadgetBoi: Just curious, what Nvidia driver are you using? Some of the drivers that are not approved can have issues.
Then again, pretty much any driver can have DPC Latency issues.

 

In my sig "266.58" NVIDIA release, running default clocks

 

just to clarify my earlier post with some screenshots. i did not intend to dissuade anybody from using throttllestop. in my case, disabling gpu monitoring reduced my dpc latency considerably. however, as the screenshot shows, using throttlestop for its primary function under load did not affect dpc latency.

 

Thanks argwings for clarifying that.

One thing I noticed is that many people check Clock Modulation, Chipset Clock Mod and Set Multiplier. For the M11x-R2, I haven't seen any need to check any of those. The less things you need to check, the more efficient ThrottleStop becomes.

The Nvidia GPU option in ThrottleStop is OK if you are doing a couple of quick tests but GPU temperature is rarely an issue on these laptops so there's no need to monitor that all the time. I believe GPU-Z bypasses the drivers and accesses the temperature sensors directly so try using that.

GadgetBoi: I looked in your previous posts for the driver version but I was blind and never read your sig. Sorry for the blonde moment.