21 Feb
2003
21 Feb
'03
11:58 a.m.
On Fri, 2003-02-21 at 04:28, Martijn Sipkema wrote:
Hmm. I have a feeling he is couting sound card latency or perhaps other
things to come up with his 11 ms measurement.
My understanding of hardware is only rudimentary. Interrupt latency is
the time between a hardware event actually happening and code to service
that event beginning to run? If it is code morphing and the mindcontrol
guy's numbers are right, then I should have around 500 us interrupt
latency, since my processor is 872 MHz and the one in his presentation
is 400 MHz. There have also been a few revisions of the code morphing
software since those slides were made.
An additional 400 us of latency due to code morphing is something I can
live with for my applications. If I can get an upper bound of 5 ms from
MIDI hitting the interface to sound coming out of the headphone jack, I
will be satisfied. As things are, jack and it's first client hang the
system.
Further investigation has revealed that PD does not hang when it talks
directly to ALSA, even when it runs with realtime priority. Although I
do require 11 size 128 "frags" to do audio without dropouts, it is my
(again, limited) understanding that this can be mitigated somewhat by
using a lower latency (pcmcia) sound card. A curious problem is that
after you run PD a few times, the mixer starts behaving strangely. I now
suspect something is amiss in ALSA.
--
(jfm3 2838 BCBA 93BA 3058 ED95 A42C 37DB 66D1 B43C 9FD0)
I'm trying
to run a low latency kernel and audio applications on a
crusoe processor laptop.
Yes, I'm crazy.
You might want to take a look at the following:
http://www.mindcontrol.org/~hplus/aes-2001
In slide 5/6 three different processors are compared and the 400 MHz
Transmeta Crusoe is said to have a 1000 us interrupt latency whereas
a 600 MHz Intel Celeron processor has a 80 us interrupts latency,
both running BeIA/BeOS. This is probably related to code morphing.
Maybe more recent Crusoes aren't affected.