Page 4 of 4

Re: Worse performance when power supply plugged in!

Posted: Thu Jun 06, 2019 9:44 am
by Musicteacher
I have taken a look (even though the gig is still to come, but a bios setting is undone faster than a kernel upgrade).

There are two settings that can be changed in this bios:

- intel speed step

The problem is that when I de-activate intel speed step, the cpu runs on a fixed rate of 1.4 gHz, whereas my own tests showed that the cpu can run at a rate of 1.8 gHz for a longer time (3.4 gHz is the max turbo boost frequency, but this cannot be kept up when using more than 1 core). So if I de-activate intel speed-step, I loose cpu power. I decided not to do that.

- processor power management. This controls the sleep-states. If I change that, it seems that nothing changes. i7z shows that still the cpu enters c1 state (the others are turned of by myself).

The lenovo bios on the L380 is not very powerful. If you buy a professional-grade lenovo machine, there are much more settings.

Re: Worse performance when power supply plugged in!

Posted: Thu Jun 06, 2019 9:54 am
by Jack Winter
No thermal shutdown msgs in the output of dmesg? ;)

This is a really curious problem. According to the cyclictest/hwlatdetect results there are no problems with kernel scheduling nor with NMI/SMIs. I'd suggest though to run both for a longer time. Run cyclictest during an entire cycle of, on battery, charging, on external power, and then do the same with hwlatdetect. These utilities might need to run for an extensive time to catch the problem.

Re: Worse performance when power supply plugged in!

Posted: Thu Jun 06, 2019 12:44 pm
by merlyn
Musicteacher wrote:a bios setting is undone faster than a kernel upgrade
It's possible to have multiple kernels e.g. linux, linux-rt and linux-rt-lts installed and choose the one you want at boot with grub.

yeah, i believe the guy

Posted: Fri Jun 07, 2019 9:41 pm
by langoring_composer
He might not want to disable hyperthreading. However, sometimes it's tied to ACPI (APCI?) anyways power-related stuff so it can't be done independently. Also, these days, power components of laptops have miniature computer-like parts inside of them and can be hacked from the AC outlets which can also be repurposed as radio antennas. (ya learn weird stuff from studying the Sharper Image and RadioShack catalogs along with taking stuff apart and going to college classes).

In some studios, they use power line conditioners to filter out some of that riff raff which they used to call "RF" (radio frequency) interference. But then again, that RF stuff might have been PCM data signtals sent into gear to mess with it (as long as it complies with CFR FCC rules!).

Re: Worse performance when power supply plugged in!

Posted: Tue Jun 11, 2019 8:47 am
by Musicteacher
Hi,
I do switch off hyperthreading.

In another thread I read that the general jack-realtime-priority should be lower than that of the soundcard.

Is it correct that I set the realtime-priority for jack in limits.conf and the one for the soundcard with rtirq ?

If so, are 95 for the soundcard and 80 for the rest a good value?

Re: Worse performance when power supply plugged in!

Posted: Tue Jun 11, 2019 11:15 am
by merlyn
Musicteacher wrote:Is it correct that I set the realtime-priority for jack in limits.conf and the one for the soundcard with rtirq ?
limits.conf is a global maximum. The priority of JACK is set when you start it. On Cadence there's a box to set this, there will be on QJackctl too. rtirq sets the priority of the soundcard interrupt.

Having the soundcard at 95 and JACK at 85 works.

Re: Worse performance when power supply plugged in!

Posted: Tue Sep 27, 2022 10:07 am
by Blindekinder
Hi Musicteacher,

I have the same problem. It took me a long time to notice it is related to battery and PSU. I didn't yet performed any of the tests mentioned in this thread, I'll try soon.
Did you solved that (one non-numeric virus after :wink: )?
I'm also using my computer for professional purpose: live music, live shows, sound diffusion, spatialisation, mainly with BitwigStudio, Ardour, PD and plugins.

Here is the discussion about the issue.

Re: Worse performance when power supply plugged in!

Posted: Thu Sep 29, 2022 6:19 pm
by Loki Harfagr
artix_linux_user wrote: Thu Sep 29, 2022 5:52 pm I ve read in one of the first threads something about lenovo...well these machines are rock solid as far as I can remember.

That'd be for the T series or some off the X series, the L series (those I've tested or seen at friends) alas are more a toy than a tool and especially is a bit uncertain on lifespan side and usually easily between hot when working and too hot for working.

Re: Worse performance when power supply plugged in!

Posted: Tue Oct 04, 2022 9:50 am
by Blindekinder
artix_linux_user wrote: Thu Sep 29, 2022 5:52 pm it is more a feel then real?
It's not about performance, it's not about 50/60Hz hum, it's real: I do have Xruns, a lot of Xruns, audibles and at any latency.

The power lines in my house, in our studio or in theaters here is probably not in cause (never had problem), and the PSU is the one that came with the Lenovo, which is a P15s by the way. Nor the sound card, because it happens the same on my different devices.

Maybe as you say it "breaks the laws of physics".

Re: Worse performance when power supply plugged in!

Posted: Tue Oct 04, 2022 2:58 pm
by Blindekinder
definitely not I guess:
It was not as bad on AVL, and I've just tried with a ground-lifted cable, no difference.