I think the question is, “do you feel a difference in latency” as I doubt many of us will hear this latency 
This is most likely to affect your playing style rather than be a noticable delay of audio.
But what about the inherit knowledge that it’s there? 
Having listened to John Watkinson on the subject of audio perception concerning 44.1kHz sampling and speaker design, I have learnt to question many of our perceived wisdom’s concerning digital audio and our need/ability to respond to stimuli.
Some people really do seem to have uncanny abilities,. Perfect Pitch is an interesting example. Bartok (it’s it claimed had tinnitus) and his affection for 2nd’s and sixth’s might be explained, imagine the curse of accurate perception in a sea of tone.
Once you KNOW something is there it can occlude much other enjoyment from what the completely artificial rendering of digital signals pertaining to be a musical instrument is capable of … 
But relax! if the issue is resolved, the human mind will grant you the favour of hardly ever thinking of it at all.!!
Call it the tooth-ache effect !
1 Like
I personally will stick on the 256 config, as far as i can’t feel the difference 
You only have to change a number in the Jackd settings (Webconf → Hardware → Audio).
Not too difficult
Regards,
1 Like
How hard is to replace the RBPi-3 for the RBPi-4, using the same regular Zynthian’s case ?
Any directions / guides for making this ?
Not that hard.
Guide: I used an electric drill and ground out the holes until the pi4 fit.
One has to admire the journey from the title of this thread to where it’s now heading ! 
2 Likes
I took the electronics out first!
I now hooked up the 3A power supply for the TinkerBoard S to my RasbPi 3 and set latency to 128 samples and 2 perioids. PianoTeq is not at all usable this way, setBfree creates some xruns, too. So, more power didn’t really help apparently.
setBfree shouldn’t have problems at 128 samples / 2 periods. With RBPi4 this works like a charm!!
Mhhh I’d like to have some box that runs PianoTeq, Dexed and Zynthian at 128 samples/2 periods but I think there’s n o RaspberrryPi that is strong enough for this.
Pianoteq at 128 samples is probably beyond the limits of any ARM SOC we can use.
But i don’t think that running it at 256 samples is so bad … come on!!! Simply put your monitors closer to your ears (or use headphones!)
5.6 ms, that is the difference between running at 128 instead 256 samples, is the time that sound takes to travel 1.9 m. So, typically, using headphones instead monitors will reduce perceived latency by 3 to 6 ms 
Well… on stage I can add the delay of digital stage box and digital mixer to the latency… plus it doesn’t look that cool to use headphones (well, in-ear plugs maybe). Playing with 256 samples simply feels like the instrument is broken. Even at 128 setBfree doesn’t feel that right. I got this Korg CX-3 here from the 80ties, reparied it for a customer, now that is something that is reacting instantly… and so did my Oberheim OB3² (but then it died, well many years ago).
These are analog devices, and very good ones …
I doubt that zynthian can do it better than the original devices, but it can do it cheaper and smaller, certainly. And of course, zynthian can do “many things” …
Remember what zynthian is: a convenient micro computer for audio synthesis & processing. The limit is CPU processing power, and i’ts contrained by technology.
If you really want more power, then you should consider running zynthian in a Intel/AMD computer. It can be done too
Regards,
2 Likes
What has he just said … ! ? ! ? ! ? !!! 
2 Likes
Well, the OB3² was entirely digital, but didn’t sound as good as setBfree. The old Korg CX-3 is analog, that is true, but its sound is totally overrated. Tell me more about the x86 Zynhtian, though… Unclear to me is how to attach the display and encoders. Fanless x86 mini computers are not that expensive any more so this might really be an option.
1 Like
I suspect, but do not know, that the x86 Zynthian is a wonder if project, rather than the route out of Raspberry Pi latency problems for zynths . . .
Mind you, if we swiss army enough who knows…?
I’m sure I remember someone saying jack ran on a PC, but realistically it would want on be a linux dist on the PC . .
12c connection isn’t native on a PC ( until some tells me it is … ) so it’s probably a software only exercise, I seem to remember there is a dummy controller code somewhere out there . . . Again this is jofe’s call but I think it would really slow down development to support it in Master. . .
Interesting concept thou’
A zynth per engine and Ethernet networking might be one way of getting the buffer size down to something you might find acceptable.
How about this - with HD audio on board! https://www.mouser.co.uk/datasheet/2/763/HYPER-BW_web_20190226-1594366.pdf
The problem, IMHO, is that latency isn’t constant, at 11ms it jitters between a few ms and 11ms (+ the latency of the keyboard USB connection to zynthian, I suppose). This personally drives me crazy and I can clearly feel a difference between 5ms and 11ms.