Subject: Re: [linux-audio-dev] Re: laptop audio performance chart ... any realworld tests ?
From: Benno Senoner (sbenno_AT_gardena.net)
Date: Tue Jul 10 2001 - 12:15:37 EEST
Hi,
don't worry no overrun occurred: there is only a single peak at the beginning.
It happens on some cards (my old Tropez+ shows the same behaviour).
It seems that it is due to the card needing some time to "startup" when you
write the first fragment.
Your latency diagrams perfectly fine and your laptop is definitively
lowlatency capable.
BTW: old OSS versions allocated #requested_fragments + 1 rather than
#requested_fragments. This means that you didn't actually run the the test
with 3 x 256 but 2 x 256. (that's no problem)
So if you have time: can you rerun the test with 4 x 128 (which correspond
to an effective 3 frags x 128 bytes) and post the results ?
At least we now have the confirmation that there are laptops suitable for
realtime work.
cheers,
Benno.
On Tuesday 10 July 2001 02:51, you wrote:
> Benno,
>
> Thank you for the fix to rtc_latencytest.
>
> Well, for better or for worse, the results of running
>
> ../do_tests none 3 256 0 350000000
>
> are up at
>
> http://www.ouroboros-complex.org/3x256.html
>
> I'm not exactly sure how to interpret the results, but it I guess there
> was an overrun. Am I running the right patch? My patch file starts
> like this:
>
>
> diff -urN linux-2.4.5/arch/i386/config.in linux/arch/i386/config.in
> --- linux-2.4.5/arch/i386/config.in Thu May 24 15:14:08 2001
> +++ linux/arch/i386/config.in Wed May 30 10:50:38 2001
> @@ -26,6 +26,9 @@
>
>
> I ran tunedisk on /dev/hda and /dev/hdc. There are no other disks.
> /proc/sys/kernel/lowlatency is 1, I checked twice. Things are presently
> responsive enough for me to perform with, but I sure would rather be
> kicking all those windows boxes on the rme site in the hiney. Thanks
> for any help.
>
> (jfm3)
>
This archive was generated by hypermail 2b28 : Tue Jul 10 2001 - 12:12:43 EEST