Author |
Message |
selvmarcus
Joined: Feb 08, 2006 Posts: 121 Location: Berlin, Germany
G2 patch files: 39
|
Posted: Fri Mar 24, 2006 3:41 am Post subject:
Bug? Level multiplier output is delayed one sample sometimes |
|
|
Output should be input1 * input2 in all cases, or not?
At the screen shot you see a flip-flop was triggered because
the output of this multiplier summed with the input does NOT
equal zero sometimes (other input is -64, which resembles negative
one).
When I looked with my scope tool, I found the ouput of the level
multiplier is delayed one audio-cycle sometimes.
To see the one sample delay on the scope display, switch the "Kick" labeled module.
Scope input1 shows then multiplier input, input4 output.
My patch involves two cross-feedback loops with sample&holds driven
by a high-speed clock.
I had the effect vanish when I exchanged the inputs of
the level multiplier. But still the working of my patch seems to be dependant on the order of some modules, like if I have a multiplication early or late in one feedback path. But I don´t know "how long"
the computation in a module takes! So how can I optimize?
Shouldn´t the optimizer/patch compiler figure out in which
order the calculation has to be done?
Am I expecting too much here?
It would be a pity if I cannot do these high-res physical simulations
with my G2. Do I have to start DSP-programming again for this?
-------
Marcus
Description: |
|
Filesize: |
48.67 KB |
Viewed: |
4007 Time(s) |
|
Description: |
Physical model of a bouncing&spring-connected object, shows possible bug in level multiplier |
|
Download |
Filename: |
BouncerSpringSM.prf2 |
Filesize: |
7.14 KB |
Downloaded: |
1011 Time(s) |
|
|
Back to top
|
|
|
Tim Kleinert
Joined: Mar 12, 2004 Posts: 1148 Location: Zürich, Switzerland
Audio files: 7
G2 patch files: 236
|
Posted: Fri Mar 24, 2006 7:35 am Post subject:
|
|
|
It's not a bug.
The G2 can only calculate one module at a time. So, to prevent latency issues, it uses an intelligent algorithm to determine in what order the modules should be calculated.
With straightforward linear signal flows, this always works well.
When using feedback however, things can become obscure. In such a case, the algorithm will revert to calculating the modules simply in the order in which they are placed on the screen, up to down and left to right. (Very left column, up to down, then next column right, up to down etc.)
So, just place the modules in the order in which they work correctly, and all will be well. |
|
Back to top
|
|
|
selvmarcus
Joined: Feb 08, 2006 Posts: 121 Location: Berlin, Germany
G2 patch files: 39
|
Posted: Fri Mar 24, 2006 5:50 pm Post subject:
|
|
|
Great, thank you very much!
That gives me back my trust and enthusiasm.
Just some deterministic behaviour, please, then I will
FE-E-E-EEL ALRIGHT!!
Maybe this should be moved now, I was so disappointed before,
I started crying "bug" here too fast...
------
Marcus |
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Fri Mar 24, 2006 6:03 pm Post subject:
|
|
|
selvmarcus wrote: |
Maybe this should be moved now, I was so disappointed before,
I started crying "bug" here too fast...
|
Just a sec please ... done ! _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
ian-s
Joined: Apr 01, 2004 Posts: 2669 Location: Auckland, New Zealand
Audio files: 42
G2 patch files: 626
|
Posted: Fri Mar 24, 2006 7:21 pm Post subject:
|
|
|
So when its working as you expect, please repost. Looks interesting. |
|
Back to top
|
|
|
|