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 _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
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_for_usb_interface 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 [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
On Wed, 7 Aug 2019, Ralf Mardorf 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_for_usb_interface 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. _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
In reply to this post by Holger Marzen
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. _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
On 8/7/19 9:59 PM, Tim wrote:
> 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. 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 > 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 _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
On 8/7/19 4:38 PM, Robin Gareus wrote:
> On 8/7/19 9:59 PM, Tim wrote: >> 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. > > Nope. The client (jack process cycle) block-size is 192 in this case. ... Right. Sorry, I goofed bigly, I was referring to how -n affects total latencies seen. T. _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
In reply to this post by Ralf Mardorf
Do recent usb soundcards need "-n 3"? On Thu, Aug 8, 2019 at 3:31 AM Ralf Mardorf <[hidden email]> wrote: On Wed, 2019-08-07 at 16:59 +0200, Holger Marzen wrote: _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
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 _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
In reply to this post by crocket
On Thu, 8 Aug 2019 12:49:42 +0900, crocket wrote:
>On Thu, Aug 8, 2019 at 3:31 AM Ralf Mardorf 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_for_usb_interface >> >> 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. >Do recent usb soundcards need "-n 3"? 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 _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
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 _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
In reply to this post by Tim-2
On Wed, 7 Aug 2019, Tim wrote:
> 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. 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 _______________________________________________ Jack-Devel mailing list [hidden email] http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org |
Free forum by Nabble | Edit this page |