Just wanted to bump this to see if some other can try to see if the multiple Cantana implementation issue goes away w/ using ASIO, opposed to WDM.   Also, please note your audio interface.
Thanks!

Thanks for trying!

Just to clarify I had to turn off multi-threading AND use ASIO, with Catanya 1.0.1 and 8.5.3 to get multiple instances to work.  I'm just not sure if I should open a ticket with Presonus now, too!   Granted, this was in a simple configuration just to 'try'.    There's no doubt that the multiprocessing is a bug, too, and yes the Midi Crosstalk is still an issue.   

And on top of that WDM has better latency for me,  than ASIO, so overall I'm still not happy, just curious, and surprised that I got it to work with the above mentioned details.

greysound


Yendor,

Thanks for the comparison, but I'm not quite sure where you're going with this.  We've already demonstrated that the bug precludes multi-threading and MIDI VSTi being used together.  (If you're willing to have all of Sonar run in such a handicapped state then, yes, you can run multiple MIDI VSTi.  I would expect that most folks would find this limitation too severe for complex projects.)

However as a test of your WDM versus ASIO theory, try the tests with multi-threading on and let us know the results.  I'm guessing it will make no difference but you never know.

I don't think Yendor is talking about or disputing either of the following two statements, as regards multiple instances of Catanya (or other MIDI VSTi that send MIDI Out) :
 
"Multi-threading on" + WDM = No joy.
"Multi-threading on" + ASIO = No joy.
 
I just think he's proposing (and asking others to confirm) the following additional two statements:
 
"Multi-threading OFF" + WDM = No joy.
"Multi-threading OFF" + ASIO = Happy Happy Joy Joy. (...albeit within a "handicapped" Sonar, as you said) 
 
Also, from posts #5, #30 and #33, we can deduce that between the two of them (auto-da-fe and Yendor), they've already tested both driver modes with multi-threading on.  So we're sort of past that now.
 
At least that's how I'm understanding it. 

As I mentioned previously, I really think the crosstalk bug and multithreading crashing is caused by the same issue. The crosstalk bug is caused by Sonar sharing buffers between different areas of the application when they should really be using separate buffers. Now as anyone can tell you that has done some multithreaded programming, if you have multiple threads all accessing the same buffer in memory at the same time, the end result is usually a crash, so that right there explains the multiple instance crash. Still the same underlying bug related to improper MIDI buffer handling.

And not to complicate things more (but we need to take all variables into account), I have also found that having more than one MIDI input driver enabled at a time could also alter the results. For instance, if you have just one MIDI driver enabled, and you use a VSTi that outputs MIDI data (and has its "Enable MIDI Output" option selected), you will record those MIDI events directly into an unrelated MIDI track that has its input set to a completely different port. Now if you enable a second MIDI driver (as if you have two MIDI keyboards connected at once), you will find that it no longer records the MIDI events from the VSTi on the incorrect track (but the two MIDI tracks will cross-record each other's events from the external MIDI keyboards), yet it will still cut off MIDI events being recorded on the unrelated track.

But even if this is caused by the same underlying bug (just my speculation), we should still log a bug report for every use case that we find that does not work as expected. Even if others have logged bugs for a particular use case, if you run into the same issue, please be so kind and also log a report. Cakewalk triages the bugs based on how many people filed reports, so they need to be made aware that more people do run into these issues.

And once again, the "workarounds" do not work well at all, especially in more complex projects.

Yendor


Silktone,

Believe me, I'm with you on this, and maybe you can help me clarify a couple things, for my understanding. 

ASIO and WDM are audio drivers only, right - nothing to do with MIDI?

When you're programing VSTi's, is the use of ASIO and WDM transparent?  In other words, there's no "if WDM" ..., or "IF ASIO"

Lastly, as I'm seeing differences in WDM and ASIO, would you also agree that one of these could be causing an issue with the audio hardware?

Mucho thanks!

The VSTi never talks to the audio driver directly, so to a VSTi, switching between WDM or ASIO should makes no difference in theory. In the VSTi spec, there is no concept of an audio driver, since all interaction is only with the host itself. Even so, the reason you are seeing differences when switching between WDM and ASIO most likely is just a combination of many different things working together to expose this bug in different ways. For instance, multithreaded bugs are often very sensitive to timing differences from the operations they perform, so different drivers might change the interaction between threads in such a way that the bug may be masked in some cases, but reproduce in other cases. This is called a "race condition" because the result of the outcome depends on the timing between threads. These bugs are often very difficult to track down because sometimes just running the application within a debugger changes the timing enough so that the problem cannot be reproduced. Or the bug may only reproduce on very specific hardware or specific conditions. Or it may only reproduce once in a thousand times, etc.
 
Fortunately, if my theory is correct that it is indeed caused by the same bug that is causing the crosstalk issue, then it should be (relatively) easy to track down since that reproduces 100% of the time and is not dependent on the multithreading engine being enabled. For that all you need is to record on two MIDI tracks at the same time from two different MIDI keyboards.

http://forum.cakewalk.com/tm.aspx?m=2184706&high=crosstalk

Nevermind...looks like it is still there in X1

JR