Opening a ticket with Laptop OEM (ACPI.sys , xperf and latencymon)

Dear Community

I hope not to be very out of topic with this post, but i decided to take a step further regarding audio drop outs with my current laptop (lenovo Thinkpad T14 intel i5 10210u with 16gb of RAM)

I have 2 audio interfaces, the laptop and the ASIO drivers has everything up to date (graphic, audio network drivers plus BIOS and windows updates) in both cases I have audio drops with buffers lower than 128

The reason to try to open a ticket with Lenovo is that, after all the known audio troubleshooting, and trying to find the real reason of the audio drops, I was using latencymon (please don’t complain about this tool) to get the insight, so I notice that the main reason is the ACPI.SYS driver, which , if you know a little bit about it, involves the communication between the motherboard hardware with windows. Latencymon give you an insight of the overall issue, but does not say exactly what component/driver in the lowest level is making the high DPC/ISR values, provoking smalls drops below 128 o buffer size on the audio interface.

My next step was to use 2 additional tools, one is using the powershell with xperf command included in the “windows performance toolkit” and the other one is called “system informer” where you can trace the system activities, and where i can visually catch when the ACPI is triggered and this means to stop the xperf log.

I am attaching some screenshots for better explanation. For example the one with GP in the bacground shows a song being played while i have “system informer” on for monitoring th acpi driver and also my powershell running xperf on real time.

The other screens shows “windows performance analyzer” aka WPA with the result of xperf and filtering only the events regarding ACPI and gig performer. what i am showing on the graph, it is what i think is the problem.

In WPA it is also not that clear what exact component is triggering the ACPI but there is some information there that I want to share with lenovo. Both graphics are colliding and I assume that is the DCP/IRS main issue I am looking for and the one to explain to Lenovo. Every time that the ACPI driver kicks in, there is a sudden increment of CPU usage (show in the column label as “% Weigh”) and is when I can hear the audio drops.

The audio drops are not constant, sometimes there is one per 2 minutes, or some times there are 2 or 3 in one minute.

The laptop is very clean and I am taking care of the fan and battery.

Even when the CPU is not that powerful, the CPU meter in GP does not exceed 20%

So i just want to see how my experience with lenovo goes, but i wanted to know from the people here, their opinion about my issue " Logging "

I was working as Level 1 technician so I think these documentation is proper, besides this, i am supposed to be a premium user with lenovo, so I really hope no to be asked to do all the basic troubleshooting cause it does not help :smiley:

Thanks for your comments in advanced





Terrific information and so kind of you to share it.

That said, I won’t be holding my breath for Lenova to respond, never mind fix it. They’re probably doing something on a high priority thread

I will keep informed the topic if i got news from Lenovo, but heck, even disabling all ACPI drivers on the device manager, the ACPI kicks time to time :confused:

1 Like

Can you try this tip: https://www.tenforums.com/sound-audio/178473-solution-audio-glitches-how-ensure-general-real-time-performance.html

Thanks for the article

As far as i can tell, changing the affinity for the PID That cotains the acpi process, dont make any change in real life. I mean, i would expect to see acpi procces on the core 3 or 4 of my cpu but according to latencymon, is still running on the first core (core 0 if you include 0 as first number)

Everything runs as admin, and i can do more test later, maybe there is another way to force it but not via task manager.

Will keep you posted

Thank you

As a Lenovo user, I would like to give an opinion, but as English is not my mother tongue, I did not understand the meaning of the sentence in citation and particularly the word “Logging”, so please excuse me if my comment is irrelevant.

As I am not a computer technician at all but a simple GP user, I can only give an opinion on the operation of my Lenovo laptop.
Given its age, I work with a buffer between 128 and 256/44.1 kHz which suits me perfectly live ( I am an audio/midi guitarist), I have fully optimized my Windows 10 OS thanks to Nemanja’s guide, I have installed only the necessary applications and audio plugins, my rackspaces contain between 15 and 30 plugins and GP’s cpu meter rarely goes up to 50%.

My laptop is a Thinkpad T430 from 2012 (i5-3320M and 8GB RAM) and have never encountered any live problems (freezes, audio drops or crackles, etc), nor any variation in ventilation (it remains stable and unobtrusive).

For the past 2 years, I have been using a newer, larger laptop, but I continue to use the Lenovo as a spare with the same gig files.

I would like to take advantage of this post to ask you why you are obliged to play with a latency of less than 128s?

1 Like

Hello and thanks for your valuable opinion.

Well maybe i am a little obsessed with numbers. As i said, i have 2 interfaces, and i will rry to go on details.

The first one is an all-in-one desktop amp, the new release mighty space which i fully recommend. It is a modeling stereo amp where technically speaking, i dont need to use it as audio interface, but it has the option and runs fine at a buffer of128 with 48k sample rate. The buffer size can go as low as 8 samples.

The other interface as and old and portable “vox amolug i/o” wich is my main interface for practicing at home with another small laney 10watts amp. This interface can go as low as 24 samples (i know, is a weird number but that is the value shown on his control panel) the sample rate on this case is 44khz.

I can find their behaviors a little bit different, maybe, because if i check the RTL and also because of the difference sample rates, they can go as low as 4 ms on the Nux mighty space and 7.1 ms on the amplug.

Tha main issue, at leats for me is, when i set their bugger size at 128, rhe RTL increase as well around 14 ms, this plus the use of some effects, like the poly whammy (pitch shifter) in helix, increases the latency even more. So my goal would be to run at leas at 64 or below 10 ms of latency. I can really notice latency on my playing when i am on the worst case scenario.

I would assume that with midi guitar will be even worse, but i am not a goood player with that plugin so i can not tell.

Logging means to log the details in a technical ticket for customer support. I open a case and i got some feedback saying that i was very considerate wto provide such detailed cas and the technician sent the ticket to level 2.

I am suspecting is something regarding temperature control, i dont know if the fan is drawing too much power (even when is plugged to electric power supply) or there is a thermal control/sensor that produces this cpu interruptions. I can not tell with the screenshots i have shared, but that is their job.

Another thing i want to add is, in this forum we were talking about xmg dj 15 laptop which was optimized for real time playing. That laptop had the same processor but their related acpi/motherboard drivers where optimized to avoid such interruptions.

If this laptop i have were very optimized, even when is a low powered cpu, it could handle very low latencies but again, is a motherboard/bios issue

I hope am clear.

Cheers

Forgive me. Just to make sure: you have read and applied the suggestions that are in the Windows optimization guide? Some are really important when it comes to latency

Yep. Actually im still doing some testing. But as i mentioned before, looks like the acpi is triggering something that blocks the processor for a tiny amount of time.

Apparently usign a tool called “affinity tool” allow to move all usb hub connected devices/drivers to other cores. (I.e. ASIO Drivers) besides this affinity tool, i am using another (sistem processor) to move the GP threads to all but 0 core. Core 0 is taking care only of acpi and other processes.

When i monitor driver performance with layency mon, i dont see the asio for the amplug interface, but for some reason doing the same test, shows the mighty asio driver.

I will try to move this mighy driver as well to other cores.

I have noticed now some improvements using the amplug asio interface.

I will keep you post and if someone is interested i Will share some infor regarding affinity tools and some videos explainig how to use them

An asio driver by itself is not a driver in the same way ndis.sys, acpi.sys, etc. are. It’s just a dll that runs in the same address space and is directly called by for example GP. This dll holds the calls to the real audio driver (that runs in OS space). Of course, it is possible that manufacturers create a tandem with a real driver (especially for asio) and the asio dll for use by the audio application.

Well, you are right about the dll

I can see the dll process runing along GP on the WPA after xperf análisis , but will try to move it to other core an see.

On my previous comment, i forgot to specify, i don’t see anplug asio driver on core 0, but I can see it on core 3 (considering core 0 as the 1st one) so the affinity change done by the tool apparently works, but again, looks weird that works only for thes ampluf interface and not for the Nux mighty space. I even connected the nux device to the same port where the amplug was connected but this driver still runs on core 0

Personally, I would never use the Vox Amplug as an audio interface, as it is just a pocket guitar amp and absolutely not an audio interface. I sincerely believe that the problems you are having are due to its use, which is not intended for that purpose.

Similarly, the Nux mighty space is a guitar amp with a usb interface for editing sounds and updating the amp’s firmware, but I doubt that it can be used effectively as a real audio interface, its capabilities being limited to managing the amp.

I would advise you to get a real audio interface, not necessarily expensive but dedicated to the use of applications such as GP or a DAW.

I am very reticent about hardware or software that has been pre-set without knowing what has been installed or modified, you are better off doing it yourself, especially if you are computer-savvy.
What is more, the improvements built into this equipment will cost you a lot more than they need to, and in any case, the device is sold out.

Let me remind you that this optimisation guide has not greatly improved the performance of my old Lenovo, but it does allow me to play at up to 256s/44.1 kHz without any latency or audio problems and most importantly with a rock-solid stability (for information, I have been playing for over 50 years on all kinds of stages and I have learnt how to deal with latency problems).

As for the research work you have done on your computer and presented in your post, it seems very impressive to me but I have not understood anything because I’m rubbish at it and I have never had any reason to be interested since everything works fine for me.
And this should be the case for you using a real audio interface.

Also, for information, my audio interface is an RME Babyface Pro and I have also kept a Steinberg UR22, which is not expensive but is still effective. Focusrite Scarlett products offer excellent value for money.

The reason I changed interface is that the Steinberg’s asio driver is not multi-client, so I can not use several audio applications at the same time without going through other means, which is something to bear in mind when choosing an audio interface which, along with the computer, is the main engine of your equipment.

You are confusing the amplug i/o with the other amplug models. Mine is only a pocket audio interface with no real modeling built in. This is the one i have,

Vox has native ASIO support for this device, the other ones do not have usb audio capabilities.You can check on YouTube some people using the amplug i/o as main interface along with guitar plugins.

Same with the NUX amp. They have audio over usb , and ASIO support, this device is even multi client. My goal is not to use the nux as main interface but maybe as a backup in case something goes wrong with the amplug.

Nux also has other audio/guitar devices, including dedicated audio interfaces for ASIO connections, and just for your knowledge, the asio driver is the same in many (maybe not all) models. So i need tk disagree that it might be the devices i am using.

Both devices work as interfaces, and even if i try another ASIO device, like a real and dedicated audio interface, i am sure i will face the same but rare audio drops.

About the DJ 15, For me is a shame that it has been discontinued. There are some driver’s and setups that are not in our hands to be modified, acpi.sys/ and motherboard drivers are a good example. I hope xmg will release any new dj laptop soon as they are apparently working on it.

I dont like to bluff but i would say i have some technicall knowledge and i know how to follow such guides, but again, there are things that are not in our hands. You can check how painful are system interrupts on Dell XPS laptops. They look really good and has amazing specs, but on the firmware level, I don’t know what they are really doing wrong. I would never buy a (new) dell for audio. That said, yes we can tuns our devices but only to a limit, and i know the limit on my laptop could be improved if this acpi conflict is solve. :slight_smile:

Edit*

I also mentioned why i need lower latency, below 10ms would be ideal for me :slight_smile:

Thank you, I did not know this one.

Strange, my new laptop is a Dell G15 (year 2021) which gives me complete satisfaction and allows me to have configurations that I need in GP without any problems… But maybe I am not greedy in my needs on stage compared to other users working on sophisticated configurations.
I tend to use rackspaces that are lightly loaded but more numerous to save cpu and ram resources.
Here is an example of my usual rackspaces

I’m very interested. :slight_smile:

Affiniti videos (please watch in this order):

The affinity tool i am using:

Please try at your own risk, and if you have a backup of the whole system, even better (i used a backup tool called macrium to take a snapshot of the whole system)

I dont really know if the affinity is working for my NUX interface, as i mentioned previously, even when i set the usb hub root to all but processor 0, i still see this driver running along the acpi on the same core. But that does not happen with the amplug interface.

When setting affinity with “system informer” permanently to all but 0 core for GP i can find more audio stability.

If i change on the fly the affiniy for GP, it starts to sound with drops.

I was checking the free audio guide and i did not find any information about affinities.

I can made more and more observations that i have found by doing many hacks but in the end i am sure not all of my experiences aplies to everyone as we all have different laptops, but one thing i really want to add as well, in case you have audio drops not that frequent, is that if i use the ultimate performance, the audio drops increases ss the cpu is just running at 3.9ghz on this case (my fan is clean and i have even removed the cover leaving clearance for the fan) and if set the procesor to run at 90, 95 or 99, it goes always back to run at 2.1ghz.

So, one more time, on my case, i have found that the balance option from windows/manufacturer (i think this is the one that is calldd wwindows modern stanby) works betyer for me. I found less and less acpi calls and very very few audio drops even with the lowest buffer sizes available.

If anyone find these recomendations good on your laptos, would be really cool if you give feedback.

Thanks

It was in the beta build of the guide and Process Lasso was mentioned, but didn’t end up in the final version, as we didn’t have any “problematic” laptops that needed this kind of tweaking.

The next release (probably when Windows 12 comes out) will most likely have at least references to these threads, but also something like this thread.

1 Like