Btw it would be possible to revert the timer driven audio and go back to using the device audio interrupts for some or all of the audio devices with only a modest amount of effort. It is also possible (likely) that just a little bit of adjustment to the audio tunables can be used to eliminate glitches. (There is a balance between latency and glitch-free audio - more buffering increases latency but may clear up any glitches. I selected values that I thought would give the best glitch-free audio with minimum latency but it is possible that in heavier workloads more buffering is needed.) On Thu, Sep 14, 2017 at 6:56 AM Garrett D'Amore <[email protected]> wrote:
> > > Boomer isnt OSS and it isnt pulse. I havent worked on it in ages but then > I sort of assumed nobody was actually doing interesting things with audio > on illumos or Solaris. > > I partly suspect that some of the systems fare poorly with the timer > driven audio instead of using the on-chip isochronous interrupts. The > timer driven stuff was an innovation intended to facilitate moving audio > streams between different devices (especially SunRay stream mobility) and > also to make writing audio device drivers simpler. (No custom device > interrupt handling needed.) I know that this approach screwed up VMware as > their audio emulation actually depends on "using" the interrupts and cannot > run freely without them. I thought I fixed that. > > If folks have other audio issues I can perhaps try to look at it but I > would need solid problem reports and ideally access to some of the audio > devices that are affected. (I know USB audio support is a mess but that was > always the case even before boomer.) > > The current state of illumos audio may not be perfect but it is far better > than pre boomer days. (Recall that pre boomer the stack was dev audio > based and worked with few open source apps. It was a Sun specific streams > api. Boomer still supports that legacy api btw but also supports OSS apis.) > > - Garrett > > On Thu, Sep 14, 2017 at 3:14 AM Udo Grabowski (IMK) <[email protected]> > wrote: > >> On 14/09/2017 12:04, Nikola M wrote: >> > On 09/13/17 10:01 PM, Peter Tribble wrote: >> >> In my OpenSolaris t-shirt collection, I have one with the slogan: >> >> >> >> "Innovation happens everywhere" >> > >> > So you want us to dream? :) >> > >> > - Would like to have playing sound just don't suck on illumos, kracking, >> > stopping, pausing, really don't care if it about boomer or pulseaudio >> or vlc or >> > driver or firefox, or inability for user processes to have higher >> priority - >> > just make the sound work without such awful problems. That would really >> be a >> > breakthrough for me. >> > >> >> pull this pulse/alsa/boomer/whatever sh.. out and go back to plain OSS - >> that >> used to work flawlessly. We don't need everything that Linux fanboys >> believe >> to be good (its usually not) just because its new. >> >> > ... >> > >> > - After finally implementing OpenCL, AMD graphics drivers, have GPU >> processing >> > power shared across zones and applications with speed throttling. Have >> also >> > PCI-e-Passthrough for VMs and distributed OpenCL framework running at >> the same >> > time. >> > >> >> That depends on the vendor, NVIDIA did not compile CL into their >> solaris x86 drivers, only into the Linux ones, so no luck there. >> >> -- >> Dr.Udo Grabowski Inst.f.Meteorology & Climate Research IMK-ASF-SAT >> http://www.imk-asf.kit.edu/english/sat.php >> KIT - Karlsruhe Institute of Technology http://www.kit.edu >> Postfach 3640,76021 Karlsruhe,Germany T:(+49)721 608-26026 F:-926026 >> ------------------------------------------ illumos-discuss Archives: https://illumos.topicbox.com/groups/discuss/discussions/T83f198c8597cf8e3-Mb582826a9b31e6dcc58b7133 Powered by Topicbox: https://topicbox.com
