ThreeFingersOfLove
Joined: Oct 21, 2004 Posts: 162 Location: Greece
Audio files: 3
G2 patch files: 1
|
Posted: Sat Oct 30, 2004 11:15 am Post subject:
Optimization bug |
|
|
Hello,
I was studying Rob's version of a formant oscillator and came up with what seems to be an optimization bug. I don't remember the name of the patch - it should be in the archives. Here we go:
A red signal is patced from an OscC to one of the two inputs of an inverter module. Its output remains unconnected. I think the sole purpose of this is to force the logic module to optimize at 96 kHz; no need to patch it anywhere (Rob??). As soon as the red signal is patched to this module it turns orange.
The other output is patched to its own input and to the RandClockA module. Now the bug: If you delete the red signal from the Osc to the Inverter, the editor doesn't optimize the signals to yellow! If you delete the cable from the inverter to the ClockRandA module it still stays orange. Which is to say that the optimization process doesn't work when a signal is routed to the same module, regardless of whether it is routed to other modules as well. Why would we want an orange signal at 96k when that module is not patched anywhere? Of course, if we want to use it the resolution is going to be there, but up till that point it just takes resources. All of the above applies to any module.
Let me know what you think.
Yannis
|
|
Rob
Joined: Mar 29, 2004 Posts: 580 Location: The Hague/Netherlands/EC
G2 patch files: 109
|
Posted: Sat Oct 30, 2004 4:33 pm Post subject:
Re: Optimization bug |
|
|
ThreeFingersOfLove wrote: | Hello,
I was studying Rob's version of a formant oscillator and came up with what seems to be an optimization bug. I don't remember the name of the patch - it should be in the archives. Here we go:
A red signal is patced from an OscC to one of the two inputs of an inverter module. Its output remains unconnected. I think the sole purpose of this is to force the logic module to optimize at 96 kHz; no need to patch it anywhere (Rob??). As soon as the red signal is patched to this module it turns orange.
The other output is patched to its own input and to the RandClockA module. Now the bug: If you delete the red signal from the Osc to the Inverter, the editor doesn't optimize the signals to yellow! If you delete the cable from the inverter to the ClockRandA module it still stays orange. Which is to say that the optimization process doesn't work when a signal is routed to the same module, regardless of whether it is routed to other modules as well. Why would we want an orange signal at 96k when that module is not patched anywhere? Of course, if we want to use it the resolution is going to be there, but up till that point it just takes resources. All of the above applies to any module.
Let me know what you think.
Yannis
|
Its not really a bug, this way it tries to keep feedback loops at the same sample rate. This way the calculation process can also sometimes skip recalculation when a cable is deleted. In fact, there could be some difficult desgin choice behind this behaviour, it is consequent behaviour. |
|