| snapfrac1 | anyone know if cpu frequency scaling is an os setting, or solely bios/firmware setting? I've heard mixed things.. | 12:18 |
|---|---|---|
| trebmuh | snapfrac1, os setting | 13:18 |
| trebmuh | see http://wiki.linuxaudio.org/wiki/system_configuration#cpu_frequency_scaling | 13:19 |
| snapfrac1 | trebmuh: thanks! | 13:19 |
| snapfrac1 | I heard it can cause xruns.. | 13:19 |
| trebmuh | then, you just heard half of the logic | 13:21 |
| snapfrac1 | trebmuh: what do you mean by "half of the logic?" | 14:04 |
| trebmuh | you misunderstood | 14:04 |
| trebmuh | the other part of the logic is : "if everything is well set up, you do not have xrun" | 14:05 |
| snapfrac1 | trebmuh: well, I did here that too, but that one of the culprits can be cpu freq scaling | 14:06 |
| snapfrac1 | trebmuh: I'm on a chromebook with ubuntustudio.. not expecting much performance from this, especially since I have no external sound card. but I'd like to see what I can make this thing do | 14:09 |
| === Iota- is now known as kspencer | ||
| OvenWerks | snapfrac1: what cpu does a chromebook have? | 15:48 |
| snapfrac1 | Intel Celeron 2955U (1.4GHz) (Haswell) | 15:49 |
| OvenWerks | do you know if it has pstates or frequency stepping? | 15:50 |
| OvenWerks | ( try ls /sys/devices/system/cpu/ ) | 15:51 |
| OvenWerks | see if one of the directories is intel_pstates | 15:56 |
| OvenWerks | snapfrac1: one thing to be aware of with ubuntu and maybe debian too, is that default is to reset to ondemand/powersave 50 seconds after startup. | 16:00 |
| OvenWerks | snapfrac1: none of the tutorials I know of tell you this or make up for it | 16:00 |
| OvenWerks | snapfrac1: the best way to set ubuntustudio for performance is: | 16:01 |
| OvenWerks | sudo /usr/sbin/update-rc.d ondemand disable | 16:06 |
| OvenWerks | and then create a file /etc/default/cpufrequtils with the line GOVERNOR="performance" in it. | 16:07 |
| OvenWerks | snapfrac1 doing it that way will not interfere with upgrades (or get reset by upgrades). | 16:08 |
| snapfrac1 | OvenWerks: it does have intel_pstate (singular, though, not plural) | 16:29 |
| OvenWerks | I may have mistyped :P | 16:30 |
| OvenWerks | That means you would have either performance or powersave | 16:30 |
| OvenWerks | but the above directions should work either way | 16:31 |
| snapfrac1 | OvenWerks: is there a way to turn off power saving just for a given session? | 16:31 |
| OvenWerks | As in for only one user? or on and off at will? | 16:31 |
| snapfrac1 | since, I might want to use the power saving stuff usually, and only turn it off for using audio stuff | 16:32 |
| snapfrac1 | on / off at will.. | 16:32 |
| OvenWerks | I am working on a utility to do that, but have not got it finished. There are some out there already though. | 16:33 |
| snapfrac1 | OvenWerks: cool! | 16:34 |
| snapfrac1 | need any help? | 16:34 |
| OvenWerks | the way most do it is with a piece of code that looks like: | 16:34 |
| snapfrac1 | got it on github? | 16:34 |
| OvenWerks | no it is https://code.launchpad.net/~ubuntustudio-dev/ubuntustudio-controls/trunk | 16:36 |
| OvenWerks | there is a version in the https://launchpad.net/~ubuntustudio-dev/+archive/ubuntu/autobuild ppa | 16:36 |
| OvenWerks | but it is not anywhere near there any more. | 16:37 |
| OvenWerks | That model relies on the whole thing being run pkexec | 16:37 |
| OvenWerks | That works fine for what is there, but we want to control audio setup too which is all userspace/current session | 16:38 |
| OvenWerks | so the GUI actually has to run userspace and call pkexec to set things | 16:39 |
| OvenWerks | I am in the middle of changing over to do that. | 16:39 |
| OvenWerks | The ppa version has a lot of stuff in the gui that does nothing right now, but setting performance does work. | 16:40 |
| OvenWerks | snapfrac1: ^^^ | 16:40 |
| snapfrac1 | OvenWerks: that sounds cool! some of this is over my head.. I don't know what pkexec is, for example. does it involve suid bit? | 16:40 |
| OvenWerks | it is what ubuntu/debian uses instead of sudo. | 16:40 |
| OvenWerks | it is quite flexable as it can run a specific executable with or without a password or allow gui access or not. | 16:42 |
| OvenWerks | it is part of policy kit | 16:43 |
| OvenWerks | -controls is written in python with gtk3 for the GUI (gui is made with glade) | 16:44 |
| snapfrac1 | all this time I've been using sudo.. | 16:44 |
| snapfrac1 | hm.. | 16:44 |
| OvenWerks | sudo is great for cli stuff | 16:44 |
| snapfrac1 | ah, pkexec is for gui? | 16:44 |
| OvenWerks | pkexec is better for stuff that is embedded into another program/script | 16:45 |
| snapfrac1 | is it like gtksudo? | 16:45 |
| snapfrac1 | or whatever that used to be called? | 16:45 |
| OvenWerks | somewhat yes. | 16:45 |
| snapfrac1 | I remember using that a long time ago.. not sure if it's "defunct" now, but I haven't used it in years | 16:45 |
| OvenWerks | in general it is bad to run the GUI with root access, it is better to run the GUI as user and then just do what needs to be done in root as root. | 16:46 |
| snapfrac1 | OvenWerks: yeah, that makes sense from a security stand-point | 16:47 |
| snapfrac1 | principle of least authority and everything | 16:47 |
| OvenWerks | So for controls, there will be two executables, userside and systemn side | 16:47 |
| OvenWerks | userside will run the system side stuff with pkexec and predefined commands. | 16:48 |
| OvenWerks | that is the system side code will not accept text or environmetal parameters as input. | 16:49 |
| OvenWerks | synaptic still runs the gui as root... | 16:50 |
| snapfrac1 | I use aptitude on command line usually .. | 16:50 |
| OvenWerks | the current version of ubuntustudio-controls does too, but that will change with the next version | 16:50 |
| OvenWerks | anything but "software" or USC | 16:51 |
| OvenWerks | I think the only reason Studio ships Software is that it is in the base SW setup. | 16:52 |
| snapfrac1 | what do you mean? | 16:52 |
| OvenWerks | SOftware does not tell you if it has to remove sw to install other sw | 16:52 |
| snapfrac1 | you're talking about software center? | 16:53 |
| OvenWerks | yes, in the menu it is just called "Software" | 16:54 |
| OvenWerks | the ubuntu software center is gone | 16:54 |
| snapfrac1 | ah. haha I never use that.. I like aptitude because it's clear what's going on with the dependencies | 16:54 |
| OvenWerks | right | 16:54 |
| snapfrac1 | also, it remembers what I manually selected, vs what was installed to satisfy deps | 16:54 |
| snapfrac1 | (so if I remove something, I can also remove the stuff I don't need anymore along with it) | 16:55 |
| OvenWerks | installing jackd1 removes almost all of the audio software which depends on jackd2 | 16:55 |
| snapfrac1 | I have always checked in on "ubuntustudio-controls" but never really found too much in there.. just the realtime setting, and the audio group | 16:56 |
| snapfrac1 | I rememember once uninstalling something that caused my entire linux system to try to uninstall | 16:56 |
| snapfrac1 | what a pain in the ass that was | 16:57 |
| snapfrac1 | rookie mistake | 16:57 |
| OvenWerks | Yeah, -controls has needed some love for a while | 16:57 |
| snapfrac1 | that prompted me to learn more about dependencies.. and why removing one little thing can "bring everything tumbling down" | 16:58 |
| snapfrac1 | although, obviously, I could have just removed it and had broken deps | 16:58 |
| OvenWerks | when building a jack application you need the includes so many people just install libjack-dev which removes jackd2 :P | 16:58 |
| snapfrac1 | is there a libjack2-dev? | 16:58 |
| OvenWerks | Not knowing that they need libjack-jackd2-dev | 16:59 |
| OvenWerks | really, libjack-dev should be for the standard jackd which is 2 and then there should be a libjack-jackd1-dev | 16:59 |
| OvenWerks | ... and beyond that in this case just break policy and include the includes with the jackd packages. | 17:00 |
| snapfrac1 | guess it's too late for that now.. | 17:00 |
| * OvenWerks is waiting for jackd3 | 17:01 | |
| OvenWerks | now that jackd1 and jackd2 are being maintained by the same person, maybe there will be something that replaces both of them... or maybe not. | 17:02 |
| OvenWerks | jackd1 has a very strong following still. Many people consider jackd2 to be a fork rather than the next version | 17:03 |
| snapfrac1 | OvenWerks: is it not backward compatible? | 17:04 |
| OvenWerks | The lib API is compatible | 17:04 |
| OvenWerks | An application written with jackd1 with work with jackd2 and an app wirtten for jackd2 will work with jackd1 | 17:05 |
| snapfrac1 | OvenWerks: I wonder why people don't transition then.. | 17:06 |
| OvenWerks | like I say, the best thing would be a version three that includes the best of both. But even then I think there are some people who would still use jackd1 | 17:07 |
| OvenWerks | There are still some people writing ladspa plugins instead of LV2 for the same reasons | 17:08 |
| OvenWerks | Then there are some applications like Ardour that will now run without jack at all. | 17:09 |
| OvenWerks | snapfrac1: in some ways jackd1 is actually ahead of jackd2. I think the best way forward would be to take jackd1, add just a few things from jackd2 like dbus control for example and continue that as jackd3. I think that would satisfy the most people | 17:12 |
| OvenWerks | The main upside of jackd2 originally was that it has SMP use. That is it is multithread and can do things on more than one CPU. That is it works with more than one core. | 17:13 |
| OvenWerks | As it turns out though, most of the main processing still has to be done all in one thread anyway. So all that got done on other cores was minor control stuff. | 17:15 |
| snapfrac1 | OvenWerks: I always wondered about that | 17:15 |
| OvenWerks | https://github.com/jackaudio/jackaudio.github.com/wiki/Q_difference_jack1_jack2 | 17:15 |
| snapfrac1 | so much has to happen in one core because there is one stream of samples? | 17:16 |
| OvenWerks | snapfrac1: ^ might explain better | 17:16 |
| OvenWerks | right | 17:16 |
| snapfrac1 | I hadn't realised that Ardour doesn't require jack | 17:16 |
| OvenWerks | the audio card interupts when it's buffer if full and every jack client has to deal with that audio before the next interrupt. | 17:17 |
| snapfrac1 | I always used it with jack assuming I had to | 17:17 |
| OvenWerks | Ardour does not need jack as of version 5.0 | 17:17 |
| snapfrac1 | ok, I'm still on 4 | 17:17 |
| OvenWerks | Using Ardour with jack has advatages still. | 17:17 |
| snapfrac1 | I haven't followed it in a while, but I know they had a milestone with midi sequencing support | 17:18 |
| OvenWerks | It does mean that a second audio device can be added with zita-ajbridge for example. | 17:18 |
| snapfrac1 | but that was ages ago now.. | 17:18 |
| snapfrac1 | OvenWerks: well, one advantage would be connecting to all the other jack-only programs | 17:19 |
| snapfrac1 | right? | 17:19 |
| OvenWerks | Ardour is at 5.5 now and there have been many changes. The GUI has changed, control surfaces (where I play) has changed... lots of MIDI upgrades, some plugins are now included. | 17:19 |
| OvenWerks | right, using external jack applications is possible. | 17:20 |
| snapfrac1 | OvenWerks: that sounds cool.. I wonder if it's worth getting the latest.. | 17:20 |
| OvenWerks | Ardour now officially supports OSx/windows | 17:20 |
| snapfrac1 | first I have to get my xruns reduced.. remember, I'm on the chromebook.. haha | 17:20 |
| snapfrac1 | only 2Gb ram too.. | 17:20 |
| snapfrac1 | unfortunately | 17:21 |
| snapfrac1 | but it does have ssd.. so there's that | 17:21 |
| snapfrac1 | no heavy bass making my hdd heads jump | 17:21 |
| OvenWerks | I ran Studio on an Atom based netbook with 1G of ram down as low as jackd set to 64/2 with no xruns. so it is possible. | 17:21 |
| snapfrac1 | OvenWerks: that's impressive | 17:22 |
| OvenWerks | I had to unload the kernel module for my wifi to get it working. | 17:22 |
| snapfrac1 | I think I can get it set up, I just have to learn the tricks.. like the cpu freq thing | 17:22 |
| snapfrac1 | OvenWerks: I do fear that I must lose networking to get it as good as I really want | 17:23 |
| OvenWerks | I had to only plug my audio interface (USB) into one of the three USB ports | 17:23 |
| OvenWerks | The other USB ports shared an IRQ with something else | 17:23 |
| snapfrac1 | OvenWerks: I'm working with onboard audio.. not great | 17:24 |
| snapfrac1 | I used to have a presonus firebox, but it stopped working. | 17:24 |
| snapfrac1 | it was firewire anyway, and this machine doesn't have a port | 17:24 |
| OvenWerks | onboard intel audio I could only get down to 64/3 or 128/2 jack would just crash if I tried lower. | 17:25 |
| snapfrac1 | OvenWerks: that'd be fine for my purposes, at the moment | 17:25 |
| snapfrac1 | I was getting xruns at much larger buffer sizes | 17:25 |
| OvenWerks | PCs are not designed for low latency audio... low latency to an Intel engineer is 30ms | 17:25 |
| OvenWerks | (skype expects 30ms latency for example) | 17:26 |
| snapfrac1 | OvenWerks: well.. network will never get lower than that anyway.. if it's routed globally | 17:26 |
| snapfrac1 | speed of light and all.. | 17:26 |
| OvenWerks | I find that my playing suffers if I put a long cable on my bass and move 30 feet from the band. | 17:27 |
| OvenWerks | That is only 20ms-ish | 17:27 |
| snapfrac1 | every few feet is like 5ms, or something like that, right? | 17:27 |
| OvenWerks | SOmething like that. When using the PC as a guitar effect, I find 256/2 is noticable delay, 128/2 is ok. | 17:28 |
| snapfrac1 | I never understood the implications of the 64/3 vs 128/2 numbers.. for example, would 64/4 be the same as 128/2 in latency? | 17:29 |
| OvenWerks | The audio interface adds delay too, my delta66 (PCI card) adds 1ms each way, but I can run it at 16/2, my USB says .6ms delay | 17:29 |
| snapfrac1 | and if so, what is different then? | 17:30 |
| OvenWerks | yes 64/4 is the same as 128/2 | 17:30 |
| OvenWerks | 64/4 might be slightly more stable, the card store 4 buffers and the computer deals with 1 at a time. | 17:31 |
| snapfrac1 | what sample rate do you usually use? | 17:32 |
| OvenWerks | */3 works better for some things that are time based like USB which works in 1ms timings. 16/3 is 1 ms at 48000 sample rate | 17:32 |
| OvenWerks | so at 48000 */3 means the buffers are sort of in sync with USB transfers | 17:33 |
| OvenWerks | going higher than 3 is probably not worthwile | 17:33 |
| OvenWerks | BTW 3 buffers is also recomended with AES67 and AVB for similar reasons | 17:34 |
| OvenWerks | (even though there are not really drivers for either yet) | 17:34 |
| OvenWerks | I use 48k for everything. The Intel audio interfaces are 48k internally whatever the actual rate is being used. 48k is most often what the analog circuitry in the audio card is designed for | 17:35 |
| OvenWerks | 48k is the standard (CDs are actually non-stanard) | 17:36 |
| snapfrac1 | what are CDs? 41K? or 96K? | 17:36 |
| OvenWerks | anything above 48k is wasted disk space and cpu overhead | 17:36 |
| OvenWerks | CDs are 44.1k | 17:36 |
| snapfrac1 | I should try jack with 48K.. I was experimenting with settings because I wanted to see how the system responded | 17:37 |
| OvenWerks | both 44.1k and 48k already have more information than your ear can hear. But the filter slope from 20k to 24k is easier for 48k | 17:38 |
| snapfrac1 | the funny thing is, petri-foo was playing the samples in a preview all distorted based on jackd settings, because the preview didn't have the resample clicked | 17:39 |
| snapfrac1 | OvenWerks: I suppose if you were recording something at 96k and wanted to slow it down, you could benefit from the extra sample density.. but I can't think of too many other scenarios | 17:41 |
| snapfrac1 | and that scenario is not something I'm likely to encounter any time soon anyway | 17:41 |
| OvenWerks | A really good intro to digital audio are these two videos: https://www.xiph.org/video/ | 17:41 |
| snapfrac1 | maybe one day when I get a really expensive mic | 17:42 |
| OvenWerks | 96k gives you nothing. | 17:42 |
| snapfrac1 | I'll be sure to check those out | 17:42 |
| OvenWerks | see I think the second video on the site above | 17:42 |
| snapfrac1 | wait.. what about if you're slowing the sounds down though? | 17:43 |
| OvenWerks | He takes audio in and compares input to output on analog equipment. | 17:43 |
| OvenWerks | most Mics only go up to about 18k... if they are good, 15k or below are common. | 17:44 |
| snapfrac1 | OvenWerks: ah, so the mic is the weakest link anyway | 17:44 |
| snapfrac1 | I didn't know that | 17:44 |
| OvenWerks | Some audio effects upsample do the effect and then downsample. That is worth while, some filters do that. | 17:45 |
| snapfrac1 | I had wrongly assumed mics could recored at higher sample rates | 17:45 |
| OvenWerks | But the input and output are still base rate. | 17:45 |
| OvenWerks | It is possible to get mics that can deal with higher frequencies. these are used by people listening to bats etc. but those mics don't do well below 20k | 17:46 |
| snapfrac1 | OvenWerks: but that sounds more like "fancy math" .. those effects wouldn't need the whole transport running at that rate, right? | 17:46 |
| OvenWerks | right, those effects do the math at a higher rate then use math to drop the rate | 17:47 |
| OvenWerks | (also common in distortion effects) | 17:48 |
| * OvenWerks is not a DSP coder so most of this info is second hand | 17:49 | |
| snapfrac1 | well.. it is for me too, but I think I grasp the basics.. those samples are just numbers, and any effect has to "process" them, so I tend to think of it like a mathematical function | 17:50 |
| snapfrac1 | I know some effects have lots of other stuff going on, and are not stateless.. but I get it for some of the ideas | 17:51 |
| snapfrac1 | anyway.. good talking to you OvenWerks .. gotta get afk for a while | 17:54 |
| snapfrac1 | I'll try what you suggested regarding the cpu freq stuff, see if I can get it down a bit in latency without the xruns | 17:54 |
| snapfrac1 | and I'll definitely watch those videos | 17:54 |
| l33 | hi and bye :) | 17:59 |
| OvenWerks | snapfrac1: See you later | 18:02 |
| redmountain_ | Ubuntu Studio 16.04.1, the window toolbar ( grey ) is so big, how to get rid of ? | 18:03 |
| OvenWerks | redmountain_: can you be more descriptive? | 18:14 |
| redmountain_ | the top bar of the windows where you can max, min, close, is so high...I need the height for my app, on the 15" touchscreen | 18:16 |
| OvenWerks | redmountain_: maybe try a different theme. there are more than one place to look. | 18:17 |
| OvenWerks | settings->window manager->style would be the first place I would try. | 18:17 |
| OvenWerks | I use Moheli, but like to be able to find my window corners for resizing as I use many windows to a screen almost always. | 18:18 |
| OvenWerks | you might try greybird-compact | 18:19 |
| redmountain_ | thanks | 18:19 |
| OvenWerks | we need a theme where the title bar is on the side of the window as the screens are getting shorter and wider | 18:20 |
| redmountain_ | The problem is with autostart at boot with the app, it has the top bar and at the bottom it is cut off... | 18:20 |
| redmountain_ | If I close start the app normally, no top bar and full size window like I need it | 18:21 |
| OvenWerks | right, that is because it is starting the application before the xfce menu bar is in place and then moves it down. | 18:21 |
| redmountain_ | this didn't happen in Ubuntu Studio 12... | 18:22 |
| redmountain_ | is there a way 'round that problem ? | 18:22 |
| OvenWerks | possible, I think the xfce version has changed. You may wish to ask in #xubuntu. We pretty much use their desktop... plus there are some of the xfce devs hang around in there | 18:23 |
| OvenWerks | I personally don't know that much about DEs, just what I like and don't :) | 18:24 |
| redmountain_ | thanks, trying over there | 18:24 |
| redmountain_ | anyone heard of X32-Edit https://downloads.music-group.com/software/behringer/X32/X32-EDIT_LINUX_V.2.5.tar.gz | 19:16 |
| redmountain | got deconnected...hmm | 20:17 |
| OvenWerks | redmountain: I downloaded it long time ago and the executable just runs... but I do not have an X32 to try it with. I have just played with it in local mode | 20:55 |
| === sakura is now known as Guest58834 | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!