Important: Must report Powermizer DPC latency issue to Nvidia

I had a chat with an Nvidia employee and it seems Nvidia is finally acknowledging the powermizer DPC latency / audio stuttering issue.

They opened a post on their forums to get information on what hardware/software configuration the problem does appear. We must reply to this post in great numbers to send them the message that this is a widespread issue affecting a lot of users.

Please go and post on this forum:

http://forums.nvidia.com/index.php?showtopic=157614

 

I would also encourage any 980NU owners that have experienced this issue to do this and also to report it to the company you bought your laptop from so that they know - this audio issue seems to have been widely unreported.

Btw, the post on their forum may be the first time that they have publically mentioned anything about it but they have been invloved in a solution for a while now.

 

Posted earlier.

Everyone please feedback!

 

Done! 10char

 

daily bump! three pages now, mostly alienware, clevo, and toshiba owners.

keep reporting! i have a feeling something good will come out of this

 

come on guys, posts have slowed down, must keep it up. please post there if you havent already

 

I just added my post. Maybe they'll release a beta with PowerMizer completely off until a real solution is found.

And it looks to me like the Dellienware people aren't so much reporting the issue as they are screaming and blaming.

 

yeah well i suspect the problem is much worse for them because they have 2 GPUs throttling... and I think disabling powermizer either screws up their SLI setting or only works on one GPU, so for them this situation is unbearable, especially when you pay 3000$ for a laptop

 

Heh, I know the feeling. The M980NU is very similar in some ways to the M17x.

Im a little dissapointed in the lack of a response here from Nvidia actually, it's been 2-3 daysn now since this meeting with the driver team we were promised some feedback from right after.
Smell a little like they were told they're not sure whether they can fix it, and not to post a response as yet.

 

they're going to be very cautious releasing information, they dont want to risk making a statement that could damage their reputation or something. hopefully when they do get back to us it will be with a tangible solution

 

It,s hurt there reputation more by not giving info and wait with a fix for this!!
I thing they don,t want claims from resellers and end users.....

 

looks like there is some hope! here's a post from Andrew at Kobalt Computers that appeared yesterday on their forums:

hopefully we'll get a fix soon

 

for those of you who use realtek:
realtek hd audio driver updated to 2.42 makes a BIG difference regardless of whatever impact nvidia drivers might have.

i have a gtx 260m and had a bad case of the crackles but the new realtek 2.42 driver seems to have helped a bunch--so far...

click for realtek 2.42 driver link

 

Posted.

And thanks for the link to the new realtek drivers. I'm going to try out 2.43 now to see if that helps any.

 

So Does this DPC Latency effect the Clevo/Sagar x8100 with the i7 and dual GTX 285's?

I totally thought this was an alienware deal, canceled my order and was about to place an order for the x8100 speciffically to avoid this issue.

 

As far as we've heard so far the X8100 is unnaffected.

 

I would most deffinatly like people's inputs please.

I Was just about to purchase the X8100 custom build for $4500 USD... until I seen this thread...

 

Very few people actually have one yet so you might need to wait a bit longer to find out for certain.

 

The X8100s definitely don't suffer the same issue, different chipset and cards.

 

New intel motherboards for the M17x R2 where the issues have gone away, I think the same new intel motherboard here, meaning likelyhood of any of these issues is probably nil plus I'd just take Neil's word for it anyway..

 

Go figure, ive had my HDX for 2 years now and only just found out this is what causes the crap audio pops. Always thought it was the crap audio card.

 

ManuelG has finally posted back on the nvidia forum.

He said:

 

Well, it's about time! My goodness, he was so silent for far too long. It would have been better if he kept us all informed daily.

 

is this latency issue an audio stutter or is it a graphical stutter despute fluid framerates?

 

it's an audio stutter that occurs when the gpu clock changes. people with SLI cards experience it worse and have more frequent DPC latency spikes and have graphical stutters as well

 

i do have graphical stutters with too many fps games and i could never locate the source to them but i never had audio stutters. i am curious how this plays out.

 

Ok people, here's the deal... I was sick of waiting for Nvidia and downloaded the new beta drivers from Dell/Alienware from this link

I managed to modify the .inf file and successfully installed the beta driver on my clevo. I've done some testing and here are my results:

Before installing the beta drivers, I used to get 64k spikes when the GPU downclocked, this is no longer the case. Now I'm getting 18k spikes when the GPU switches from level 1 (200 MHz) to level 4 (550 MHz). I ALWAYS get a single 18k spike on the 1 to 4 transition. The spike causes a slight audio stutter.

I tested the other upclock transitions (3 to 4, 2 to 4) and those are clean. All downclocks are also clean (4 to 3, 3 to 2, 2 to 1).

I also did the same tests on battery power. The 4th clock state is not available on battery power (not sure if it is the powermizer default or a setting in my power plan). The downclocks are clean (3 to 2, 2 to 1), but I'm getting a 18k spike on the 1 to 3 transition. The 2 to 3 transition is clean.

I would encourage other people to test with a single GPU (SLI disabled) to see if you can reproduce this with your own GPU. What I've done is I keep an eye on the clock rate in GPU-Z and I open a wmv file in windows media player when I want to stress the card (it makes it throttle to the max clock rate automatically). Then I close the file, let the card downclock to the desired level and fire up the video file again to test the upclock transition.

 

Yeah I modified and downloaded them myself too last night like you, and you're right, they're still not fixed (downloaded them at about midnight last night just after they went up)
I dont get the same results as you, but I do still get unpredictable spikes, I've not run the DPC latency checker for long, but I've already seen 4 k and 16k spikes, and stuff is still being intermittantly affected.

Guess Nvidia need to do a lot more work on this

 

Bear in mind guys that it is a BETA driver and hasn't been released/supported by Clevo so until the official driver fix comes from Clevo/Nvidia other solutions won't provide a 100% solution.

 

hmmm but my spikes are 100% predictable, everything else is smooth green

make sure you check all your other devices before blaming nvidia, it could be your wlan adapter drivers or something else

 

well i'm not expecting a final product... i see myself more as a beta tester for these drivers! trying to provide feedback as fast as possible so that fixes can be included in the final version of the drivers

have you tested the drivers on Kobalt notebooks? can you reproduce my 1-4 transition spikes?

 

I haven't tried personally, purely because the chassis is eol and we ran out of stock a while ago. One or two of our Comanche SLI/980NU owners tried it but it didn't fix the issue all together, then again the Dell release is BETA and it wasn't produced for Clevo. It does seem to have made a difference though.

EDIT - just noticed that you have an 860TU, we've only ever had the problem reported on a handful of 980NU chassis so haven't ever needed to look into the problem with any other chassis.

 

Hi Neil,

my feeling is that the problem is much more widespread than we think, but it's been primarily reported by Alienware m17x and 980NU users because it is worse (read unbearable) in these machines, the obvious reason being they have 2xGPUs plus 1 nvidia chipset for a total of 3 devices producing DPC spikes. From what I've seen the new drivers have partly fixed the GPU spikes (except for multi-step upclock transitions, 1 to 3, 1 to 4, still waiting for others to reproduce my observations) but Nvidia still needs to work on a fixed chipset driver.

On my single GPU Clevo the problem is harder to notice... if I'm watching a movie or playing a game the GPU is clocked to the max so no stutters happen. I can only notice if I'm working/browsing while listening to music (which I always do!). The GPU clock goes back and forth during regular use so I would hear stutters every minute or so.

 

tried to install the dell driver wit a modded inf.... well after the driver installed my screen became black and my computer rebootet automaticly... after that i could use windows anymore black screen....

in the faq it says something about 60 hz not working and my screen works with 60 hz so this might be the problem. i removed the driver in save mode and im back to the latest stable one.

still a long way to got for nvidia the finally fix this issue... ok its n beta or lets better say an alpha fix...

greetings flode

 

same stuff happened to me. black screen or bsod. when i finally could start windows, the 2d mode was buggy.

i've had to roll back to previous driver

 

did you mod the .inf file yourself? all i did is change 2 lines in the inf file to include the device id of my gtx260m

 

i used the modded inf file from laptopvideo2go... which should work fine.. driver seems to be very buggy

 

yes. i used the lv2go modded inf but as flode said, driver seems very buggy. i'll wait next nvidia release