7 Aug
2019
7 Aug
'19
7:59 a.m.
What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Both give a total of 384 samples in buffers.
What's the difference for jackd?
Regards
Holger
7 Aug
7 Aug
11:31 a.m.
On Wed, 2019-08-07 at 16:59 +0200, Holger Marzen wrote:
What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Anything other than -n2 is for esoteric reasons:
https://wiki.linuxaudio.org/wiki/list_of_jack_frame_period_settings_ideal_f…
There might be some truth behind the theory, but in practise something
like 128 or 256 frames, 2 periods/buffer at 48 KHz works without issues.
noon
On Wed, 7 Aug 2019, Ralf Mardorf wrote:
On Wed, 2019-08-07 at 16:59 +0200, Holger Marzen
wrote:
To get whole 1 ms chunks the example section recommends many
combinations with n=3.
But that's not my question. I'd like to know if there is a difference
for jackd or jack clients between combinations of n and p that result in
the same number of samples in the buffers.
What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Anything other than -n2 is for esoteric reasons:
https://wiki.linuxaudio.org/wiki/list_of_jack_frame_period_settings_ideal_f…
8:49 p.m.
Do recent usb soundcards need "-n 3"?
On Thu, Aug 8, 2019 at 3:31 AM Ralf Mardorf <ralf.mardorf(a)alice-dsl.net>
wrote:
On Wed, 2019-08-07 at 16:59 +0200, Holger Marzen
wrote:
What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Anything other than -n2 is for esoteric reasons:
https://wiki.linuxaudio.org/wiki/list_of_jack_frame_period_settings_ideal_f…
There might be some truth behind the theory, but in practise something
like 128 or 256 frames, 2 periods/buffer at 48 KHz works without issues.
_______________________________________________
Jack-Devel mailing list
Jack-Devel(a)lists.jackaudio.org
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
9:42 p.m.
On Thu, 2019-08-08 at 12:49 +0900, crocket wrote:
Do recent usb soundcards need "-n 3"?
I have a Focusrite Scarlett 2i2 here that accepts -n2 or -n3 and all
values of p from 64-8192. (8191 is prime, btw.) I was shocked because
I thought all USB interfaces needed -n3, so... idk. YMMV.
The two RME PCI interfaces I have require -n2 and -p must be a power of
2 between 64 and 1024. I haven't fully tested the HDA and HDMI devices
in the house but they seem to have similar restrictions.
$0.03
David Nielson
10:11 p.m.
On Thu, 8 Aug 2019 12:49:42 +0900, crocket wrote:
On Thu, Aug 8, 2019 at 3:31 AM Ralf Mardorf wrote:
Seemingly USB's packet size is the reason for the 1 ms cadence. I don't
have a clue if this is the case for USB 3. The Focusrite 3rd generation
Scarlett series are USB 3 audio interfaces, but I don't know if they
still work with Linux. At https://www.gearslutz.com/ forums you could
ask and probably would get a reply from Focusrite.
I once tested a Presonus Audiobox 1818VSL and I own a Focusrite
Scarlett 18i20 2nd generation, but I never tested 3 periods/buffer,
since 2 periods/buffer just do the job, even if the latency isn't a
multiple of 1ms. With USB I even get way lower latency than is
possible with a TerraTec PCI and a RME PCIe card on at least two used
Linux machines. On the newer machine I even don't unbind USB ports,
respl. I don't care about IRQs at all.
YMMV!
--
https://soundcloud.com/der-lichtecho-plan
On Wed, 2019-08-07 at 16:59 +0200, Holger Marzen
wrote:
Do recent usb soundcards need "-n 3"? What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Anything other than -n2 is for esoteric reasons:
https://wiki.linuxaudio.org/wiki/list_of_jack_frame_period_settings_ideal_f…
There might be some truth behind the theory, but in practise
something like 128 or 256 frames, 2 periods/buffer at 48 KHz works
without issues.
10:31 p.m.
On Thu, 8 Aug 2019 07:11:38 +0200, Ralf Mardorf wrote:
The Focusrite 3rd generation Scarlett series are USB 3
audio
interfaces
Pardon, they are at least using USB-C type connectors, I don't know if
this means they are using USB 3.
--
pacman -Q linux{,-rt{-cornflower,-pussytoes,,-securityink}}|cut -d\ -f2
5.2.6.arch1-1
5.2_rt1-0
5.0.21_rt16-1
5.0.19_rt11-1
4.19.59_rt24-0
12:59 p.m.
On 8/7/19 10:59 AM, Holger Marzen wrote:
What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Both give a total of 384 samples in buffers.
What's the difference for jackd?
Regards
Holger
Should not make a difference for clients, they would
simply see a 384 buffer size in both cases.
Although, I'm not sure 192 is an allowable period size.
Some technical notes from my MusE code, -n appears to
only affect audio output ports:
For Jack audio:
// For at least the ALSA driver (tested), the input port
// latency is always 1 period while the output port latency
// is always n periods (or n-1 periods for Jack1 or Jack2 Sync mode).
// In other words, the Jack command line -n (number of periods)
// ONLY applies to audio output ports.
For Jack midi:
// For at least the ALSA seq driver (tested), the input port
// latency is always 1 period while the output port latency
// is always 2 periods REGARDLESS of Jack command line -n (periods).
Tim.
1:38 p.m.
On 8/7/19 9:59 PM, Tim wrote:
On 8/7/19 10:59 AM, Holger Marzen wrote:
Nope. The client (jack process cycle) block-size is 192 in this case.
The period configuration is ALSA specific and only affects hardware
playback.
The total round-trip latency is likely be longer. That depends on the
driver and kernel side buffering which is sometimes aligned to powers or
two. With -p 192, the input buffer is also 192 samples/channel.
In synchronous mode (jack1, or jack2 -S), processing happens in the
following order. Each letter represents a "-p" number of samples (per port).
time +---------->
Input : ABCDEF
Process : .ABCDE
Out. n=2: ..ABCD
Outp n=3: ...ABC
Jack2's async mode adds one period of playback latency. Instead of
synchronous "Read", "Process", "Write". jack2 in async mode
does:
- Read from the capture device
- Write result of previous cycle to the playback device
- Process
What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Both give a total of 384 samples in buffers.
What's the difference for jackd?
Regards
Holger
Should not make a difference for clients, they would
simply see a 384 buffer size in both cases. Although, I'm not sure 192 is an allowable period
size.
For jack-clients it is, it can be any number between 1 and 8192. Some
hardware devices impose a lower limit though.
ciao,
robin
1:55 p.m.
On 8/7/19 4:38 PM, Robin Gareus wrote:
On 8/7/19 9:59 PM, Tim wrote:
...
Right. Sorry, I goofed bigly, I was referring to how -n affects
total latencies seen.
T.
On 8/7/19 10:59 AM, Holger Marzen wrote:
Nope. The client (jack process cycle) block-size is 192 in this case. What is the difference between
???? jackd -n2 -p192
and
???? jackd -n3 -p128
Both give a total of 384 samples in buffers.
What's the difference for jackd?
Regards
Holger
Should not make a difference for clients, they would
??simply see a 384 buffer size in both cases.
11:35 p.m.
On Wed, 7 Aug 2019, Tim wrote:
On 8/7/19 10:59 AM, Holger Marzen wrote:
It is. But there are some clients that need powers of 2, e.g. the
Guitarix convolvers that handle impulse responses. I read that you can
do FFT even with non-powers-of-2. I use Klangfalter for my impulse
responses and noticed no flaws when using non-powers-of-2.
Regards
Holger
What is the difference between
jackd -n2 -p192
and
jackd -n3 -p128
Both give a total of 384 samples in buffers.
What's the difference for jackd?
Regards
Holger
Should not make a difference for clients, they would
simply see a 384 buffer size in both cases.
Although, I'm not sure 192 is an allowable period size.
1758
days inactive
1758
days old
jackaudio@lists.linuxaudio.org
10 comments
6 participants
participants (6)
-
crocket -
David Nielson -
Holger Marzen -
Ralf Mardorf -
Robin Gareus -
Tim