Optimized PC build for art installation
David Dorffner
6月 27 2024 | 7:30 午前
Hi everyone,
this might be a rather vague question, but Im happy for every input!
I am setting up a server for an art installation for heavy signal processing, and I wanted to ask if there is a rule of thumb on which kind of hardware to concentrate on.
With the CPU for example is it more important to conentrate on frequency or on core count? Am I even able to utilize the CPU cores efficiently? We do have to process a lot in parallel (sound analysis and synthesis)
Is RAM a big Factor or is there a chance to assess my needs for the project?
Is a GPU needed for anything else than machine learning algorithms (or Jitter)?
Usually its the more the better, but maybe there is potential to optimize the setup
Most of the programming will be done with gen~ and we also use a lot of plugins from the Ircam Forum, which include machine learning algorithm
TFL
6月 27 2024 | 8:54 午前
I would say focus on CPU speed and no less than 4 cores. And depending on the audio material (do you use big audio files?) and used plugins, I would say at least 16GB of RAM.
If I remember correctly, the general rule is that each top-level patcher (and its subpatchers) uses its own thread, and also each instance of a poly~ will most likely use its own if available.
Could you be more specific about the "plugins from the Ircam forum"? Are you talking about AU/VST, or Max packages? Which ones? I don't know all of their machine learning stuff, but I assume some of them run on the CPU, as they doesn't necessarily require huge training sets with lots of parallel processing. That's only an assumption though, that should be verified.
If you don't have high GPU needs, I would look at these mini-pc which sometimes features the all new beasty and efficient Intel Ultra 9 processors.