On Jan 14 18:28:45, Juha Erkkila wrote:
> But in my experience latency with fluidsynth can be totally okay,
> you just need to adjust buffer sizes for shorter latency. I run
> aucat with "-b 440 -z 220" and fluidsynth with "-c 2 -z 128".
> Keeping the machine not doing mostly anything else while playing
> does help to achieve clean playback (you could also try nicing
> fluidsynth to -20).
On Jan 14 20:29:49, Alexandre Ratchov wrote:
> IIRC the latency problem could be solved by reducing the buffer size,
> ex. "-r48000 -z240 -c2" used to work, you can try even smaller block
> sizes. You'd also have to decrease sndiod latency if you're using
> it.
I find that when sndiod is not running, the latency of fluidsynth,
even without any bufsize options, is very tolerable. So it seems
it is the insertion of sndiod into the fluidsynth -> sndiod -> device
chain that introduces the latency.
But I do want to run sndiod, for other reasons.
And -r 48000 -z 200 makes it good. Thanks!
(However, the lowest recommended -z value of 400
makes the latency noticeable, so I will probably
need to tweak it some more for MTC synchronization.)
> I currently use -r48000 -b240 -z120 on a p4 at 1.8GHz running
> i386; optimal settings depend on the system, so you may find better
> ones.
OpenBSD 5.0-current (GENERIC.MP) #0: Wed Jan 11 21:07:18 CET 2012
r...@box.stare.cz:/usr/src/sys/arch/amd64/compile/GENERIC.MP
real mem = 1054593024 (1005MB)
cpu0 at mainbus0: apid 0 (boot processor)
cpu0: Intel(R) Atom(TM) CPU D510 @ 1.66GHz, 1666.95 MHz
cpu0: 512KB 64b/line 8-way L2 cache
cpu0: apic clock running at 166MHz
cpu1 at mainbus0: apid 1 (application processor)
cpu1: Intel(R) Atom(TM) CPU D510 @ 1.66GHz, 1666.69 MHz
cpu1: 512KB 64b/line 8-way L2 cache
cpu2 at mainbus0: apid 2 (application processor)
cpu2: Intel(R) Atom(TM) CPU D510 @ 1.66GHz, 1666.69 MHz
cpu2: 512KB 64b/line 8-way L2 cache
cpu3 at mainbus0: apid 3 (application processor)
cpu3: Intel(R) Atom(TM) CPU D510 @ 1.66GHz, 1666.69 MHz
cpu3: 512KB 64b/line 8-way L2 cache
Is my machine 'slow'? Or could I be better off
in this regard with the SP kernel?
Thank you for your time
Jan
