Messages in this thread | | | From | Benno Senoner <> | Subject | Re: a joint letter on low latency and Linux | Date | Sun, 2 Jul 2000 18:59:40 +0200 |
| |
On Sun, 02 Jul 2000, Jeff Garzik wrote: > On Sun, 2 Jul 2000, Paul Barton-Davis wrote: > > If linux kernel folk were willing to install the hardware necessary to > > run the "real" programs, I could point them to dozens of them in an > > instant, all of which would fail when configured for low latency. But > > they are not (which is just fine), and so we have to use "benchmarks" > > to demonstrate just how messed up things are. > > Can you give list a setup, or setups, that can be used to show low > latency for specific usage cases? > > I'm concerned that the latencies you are seeing, in some cases, may > be due to the fault of the sound driver not the inherent slowness of > the operating system.
Of course I thought about this:
that's why I included a version which uses only RTC in the latencytest package. (notice that it uses aync SIGIO notifications which are not in 2.2 by default (2.4 contains the RTC SIGIO patch) )
here you see a RTC latencytest using a 2048HZ RTC frequency (lowlatency kernel)
http://www.gardena.net/benno/linux/audio/rtc2048-cpu80/2048.html
as you can notice the max latency drift is about 500usecs under high load. no soundcard , no ALSA installed nothing, all you need is the RTC clock.
This is to cut out any disturbing issue, (like potential flaky sounddrivers), but trust me the problem _IS_ the scheduling.
BTW: I should really rewrite latencytest (it was a quick hack) from scratch in order to include RTC using blocking reads and other interesting stuff I have in mind, hopefully before the next century.
> > I'm a linux kernel folk, and I've got some test hardware, now I'm > waiting for the rhetoric to die down and the real info to start > flowing. :)
nice. can you tell use what you want to investigate in particular ?
Benno.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |