CPU Issue with poly in Max for Live


    Dec 09 2013 | 9:40 am
    I build a patch thats uses poly~ with play inside. Nothing really heavy, Cpu usage in MaxMSP is about 3.5 % (vector size 64,64), but when I port the same patch to max for live cpu usage goes up to 9-12 %. Any idea why this happens?

    • Dec 10 2013 | 3:15 pm
      quite similar but with "groove" . in some scenarios ,such things decide about the whole project . im not happy about the differences , or it may be that i dont understand what the numbers are saying (ex CPU usage monitor) . If u are on windows change your power options to "HIGH PERFORMANCE" you will see the difference , but it doesnt help to understand why in Live sample playing objects and poly's are less efficient .
    • Dec 12 2013 | 8:03 am
      Thanks for sharing your experience. Unfortunately I'm on mac so I can test this performance thing.
      This is really a big problem. Please Cycling: give some input about this.
    • Aug 12 2017 | 7:31 pm
      Yes, Please, More, Input....
      Poly is definitely less efficient in M4L ... Why is this? It's supposed to be more efficient isn't it??? Are we all missing something, such as a downsampling setting maybe?
      A patch in M4L is 10% cpu, whereas using poly~ in M4L the same patch is 30%???? How can this be?
    • Aug 12 2017 | 7:55 pm
      when in your patch poly~ was supporting multiple cores, this is gone in the plug-in version.
    • Aug 12 2017 | 11:01 pm
      …or maybe you have overdrive off in your max app?
    • Aug 13 2017 | 4:37 am
      Hi Roman, not sure I understand. Is multi core not working in M4L?
      Overdrive is always on in M4L there's no way to turn it off...
    • Aug 13 2017 | 7:45 am
      I'm really struggling with this. If I build the same device in just max using signals, I get 8% CPU (even when I copy the device 10 times). Copying the basic signal device more times doesn't seem to increase the CPU much at all, whereas copying the poly~ driven device seems to increase CPU exponentially. I created the same device using poly~ and with 10 devices the CPU goes up to 30% !!!
      I really like using poly~ because it also allows me to crush down a lot of code into one piece of code. However, with this CPU problem it seems unusable in M4L...
      I'm wondering will it help if I use downsampling either in the poly itself or in the signal chain???
    • Aug 13 2017 | 11:59 am
      i believe that anything which is not releated to the possibly different settings for overdrive and vectorsize should be considered a bug. when you are using a vectorsize of 64 in max you should see an average CPU decrease of like 10% when you make it a plug-in or live device, but not a few hundred percent. i am waiting to make live devices since many years because of these things.
    • Aug 13 2017 | 12:49 pm
      just out of interest, what is the difference between live's cpu meter and windows cpu task manager or macs activity monitor ? i think lives cpu meter reading is different to the actual usage but i'm not sure why, just something i read which was a bit vague a while a go.
      i guess if you are up for the challenge, you could make your own specially designed Daw inside of max standalone, and forgetting about ableton. ;-)