performance in serialization

[anakin]

Hi

First of all excuse me there's the possibility that this is a question just submitted... I try to find in the forum but without results. :)

I'm using two midiboxes connected together with the classic tx and rx ports in this way:

first midibox with standard Midibox64 , the midiout of it ( port TX 25 ) connected with the midi in port ( port RX 26 ) of second pic, it work on channel 1

Second midibox running the scan matrix application of TK ( fast version)  in an old keyboard ( midi merge function enable in the code) ,work on channel 1.

I'm suffer this kind of problem:

Few times it doesn't put out the Note_off message and the results is that note always stay ON and I need to re-push the key for do the Note_off.

I notice that it happens when I move some pots in the first midibox, or when I'm playin' hard ( e.g. slides)  so this is my question.

Is it possible that the problem is caused by a kind of "data overflow" from the input of the second pic?

What's about the performance of the midi merge feature? Is there any kind of filter on it?

I don't think Midibox link will help in my case because I think this isn't a loop problem: what do you think about?

And what's about the sharing of the same channel: is this a problem?

Thank you very much for your help

Anakin

[nebula]

I think you should first try disabling merge on the keyboard PIC, and then see if you can make it act up.

[anakin]

Hi and thank you for your reply

Yes, I just did this and also phisical disconnect it and it change and work good.

The problem seems to be the merge... that's why my question...

bye

[TK.]

Hi,

there shouldn't be a performance problem, especially not with sm_fast or sm_slow, as these routines are executed much faster than a new MIDI event could arrive.

How does it work if the first core is removed from the chain (keyboard directly connected to MIDI In of second core which executes sm_fast)?

If you still notice failures, try a dummy application w/o any app code but Merger enabled.

If you still notice failures after such a change, it's a hardware issue (hints to such issues later if required)

Best Regards, Thorsten.

[anakin]

Hi Tk and thank you for your reply!

If I remove the first core , I don't have the problem.

If I upload a dummy app on second core with merger enable all work good ( I upload the slow version of SM)

So I think maybe is a hardware issue, maybe with the matrix I'm using, I have to do some tests on it, surely I can open a new treat if is a hardware compatibility problem, but it is very strange as if remove the first core the apps seems to work good!

Bye

Anakin

p.s. I have some question about SM_fast but I post in a separate treat.

[TK.]

Link to the new thread

http://www.midibox.org/forum/index.php/topic,13485.0.html