CPU performance! What to use?
Im currently working on a big patch for my major project at uni.
Its basically a audio visuals package that using the audio to create some fat visuals, i am very pleased with it.
Now i am at the stage of tidying the whole thing into a presentable patch while trying to optimize the whole app for cpu performance.
So what i want to know is does any one know if there any resources on what effects on cpu does each object have?
Also when using sub patches with send receive objects for data and signals, does this create more cpu load then just using inlets and outlets?
Thanks, any insight would be appreciated!
Can you tell me more about your patch and what UNI you go to?
You shold do some general testing, partially disabling/enabling different parts of the program while monitoring the CPU.
Send/receive is probably more CPU-consuming than cords, since it need to load a separate object.
I go to herfordshire Uni. Basically the patch consists of many different parts.
Media Player which makes use of sfplay, with sflist and Coll to create a library which can be stored and played from a list via jit.cell.
Then a visuals patch, which works by using many instances of jit.poke which trace the sound waves. A random colour generator.
I then rotate and move the image in many ways to create movement via jit.rota.
I use an object called bark~ to gain fft info so to automatically generate the movement for jit.rota.
Im having problems though because my visuals occasionally freeze! I dont think its CPU though i have plenty?
What uni u at?
This is something i’d be interested in as well. I once build a patch that allowed me to test all the different jitter 2d objects for speed, but it’s not always so clear-cut.
A list of all objects with a little bit of information about their speed would be awesome, as well as other resources they eat up. Many of us don’t know everything about how many cpu-cycles a *-operation requires. Possibly included with max5?
hmmm, don’t really think this is quite possible.
there’s too much discrepancy between platforms, processor speeds, graphics cards, memory, architectures and OS versions.
individual tests tailored to specific projects on specific hardware would be much more beneficial, imho.