Announcement

Collapse
No announcement yet.

Ubuntu Desires Lower Audio Latency For Gaming

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Detructor
    replied
    Originally posted by RealNC View Post
    I already mentioned that in the post.

    What on earth are you arguing about?
    you wrote it in a way that made it seem as if dmix was some sort of virtual device, unrelated to alsa.

    Leave a comment:


  • Paradox Ethereal
    replied
    If they want lower latency for games, and also less-jitter (better performance) they need to enable low-latency desktop.
    That alone will improve latency to probably around 1ms. And makes games graphics so much better aswell.

    With a professional firewire audiocard, I have actually run audio at 0.33ms latency with a custom kernel. This does not seem to work on standard HDA chips though, but 1ms will do. And 0.33ms latency uses a great amount of CPU. So, for the time being, until audio-subsystem is optimized for low latency, 1ms will be good enough, and if you want to ensure minimal cpu usage, just set it to 5ms latency, for normal "close" pc-speakers or monitors, and it will be fine aswell.

    That does not mean I don`t want lower, but at the current time being, I would recommend that. Ultimately 0.2ms latency, should give a "close to hardware" feeling.

    PS: Ubuntu might need realtime group, to a higher priority, to run 0.33ms latency.

    Peace Be With You.

    Leave a comment:


  • RealNC
    replied
    Originally posted by Detructor View Post
    uhm, dmix is part of ALSA. So yes, it (ALSA) can do software mixing. Also dmix is (usually) enabled by default (unless the sound card is able to do hardware mixing or you are using a very weird system with very strange options on the compiler side. I've never seen a linux system with ALSA that was not able to softwaremix without a soundserver. And I've seen quite some systems).
    I already mentioned that in the post.

    What on earth are you arguing about?

    Leave a comment:


  • Detructor
    replied
    Originally posted by RealNC View Post
    No, they cannot. You're confusing the dmix device with the hardware device.
    uhm, dmix is part of ALSA. So yes, it (ALSA) can do software mixing. Also dmix is (usually) enabled by default (unless the sound card is able to do hardware mixing or you are using a very weird system with very strange options on the compiler side. I've never seen a linux system with ALSA that was not able to softwaremix without a soundserver. And I've seen quite some systems).
    Last edited by Detructor; 06 November 2012, 04:33 AM.

    Leave a comment:


  • ShadowBane
    replied
    My PA latency is currently bouncing between 7 and 9 ms, never going above 10. If the average on ubuntu is 25ms they are doing something wrong.

    (you can check your latency with 'pacmd list-sinks', it will print out a bunch of info about each sink, including the current latency.)

    Leave a comment:


  • thofke
    replied
    As this is a thread about latencies, if anybody is interested in frame latencies in Doom 3, please see my thread in the Gaming section:
    Frame latency analysis on Doom 3

    Leave a comment:


  • unix_epoch
    replied
    Originally posted by JS987 View Post
    Different people have different needs. Onboard sound cards and my discrete sound card support 192 kHz. That $235 sound card has some crappy parts if it supports only 96 kHz.
    HW mixing can be advantage in some cases, but it isn't option in many cases. If I would have to buy new sound card tomorrow I would still buy sound card without HW mixing because there is no reasonable priced card with HW mixing
    which provides high quality sound to my headphones, which is most important to me. Using two sound cards is unacceptable because it isn't possible to connect one headphones to two sound cards at same time.
    Your onboard sound card's "192kHz" probably rolls off significantly above 20-30kHz, with a much higher noise floor. Pay no attention to the sample rate. Unless you are working with ultrasound, the most important factors are measured frequency response, noise floor, jitter, and distortion. For final playback, 44.1k should be enough according to http://people.xiph.org/~xiphmont/demo/neil-young.html.

    You can use two sound cards at the same time by passing the line output of one to the line input of the other, then your headphones into the headphone output of the second. Finally, unmute the Line In on your second sound card's mixer.

    Leave a comment:


  • unix_epoch
    replied
    Originally posted by RealNC View Post
    Isn't there an API for this thing? Having to configure text files each time I want to change just doesn't cut it, IMO. This is stuff that should be configurable at runtime. Things like that is what make me believe that ALSA should be called LSA instead.
    IIRC the last page I linked to is also the libasound API docs. There would be a decent GUI for configuring ALSA, if people hadn't decided to ignore ALSA's entire infrastructure and build sound servers instead. I almost wrote one for a car PC I was building, but the car it was going into was totaled (and there were some issues with ALSA's LADSPA plugin support adding unpredictable latency) so it never happened.

    Leave a comment:


  • ninez
    replied
    Look guys, you are starting to twist this all around. I said for me that isn't some huge expense. I am not saying everyone _must_ go out and by a $300 audio interface, but if they desire certain features (and we were talking about H/W mixing, as well as me asserting earlier on that not everyone needs it, nor some of the features i would commonly use to add to that), then there are devices for $300 and under, that if chosen well (ie: researched and it's well-supported) work quite excellently under linux these days.

    ....arguments to be made over where money should be spent on one's computer is up to the individual to decide. For me, in both machines (Phenom II + i7) that i use daily, i don't feel the need for any upgrades, really - aside from adding the extra 8gig of RAM that i _could_ add to each machine (they each have 8gig right now) ...if i really felt that i needed 16gig of ram, which i haven't yet - maybe that will change... Both machines have decent Nvidia GFX cards (no more than 2-3yrs old), and decent sized drives... and i have other junker-boxes lying around too - so it's quite justified (for me) to make that kind of purchase. Just like in the next year, i hope to purchase an SSD/pcie hybrid card (although, that is entirely another subject )

    hd audio / onboard cards vary in H/W and driver quality(especially, in linux) and there are definitely other options, whether that be some ASUS gaming type soundcard, or to some low/mid/hi end audio interfaces that comes in a variety of shapes, designs and sizes, depending on your needs.... but in most cases, you really do get what you pay for... (and in the case of linux, there is the extra hassle of often having to know exactly what you are buying, as well.).

    Leave a comment:


  • Hamish Wilson
    replied
    Originally posted by Redshirt001 View Post
    This is an internet forum. I believe that such things should be used to share knowledge and information, not to engage in ideological wars, such as I see so bloody often. Pulseaudio bad, Ubuntu bad, Alsa good, hardware mixing good, etc, etc. Such a misuse of the net.
    I think you are fighting a losing battle here mate.

    Leave a comment:

Working...
X