Frank Barknecht wrote:

Hallo, Mika Ristimäki hat gesagt: // Mika Ristimäki wrote:

...

I have a one quick question about number boxes and how they effect CPU
load. I use number boxes quite a lot for debugging and sometimes I
forget to remove them from the final patch or abstraction. So I was
wondering do the number boxes cause significant CPU usage even when
they are not visible? For example when the abstraction window with the
number boxes is closed?

Even invisble numberboxes are terrible for performance so avoid them (until this is fixed). Also avoid [bng], sliders and every other GUI object unless you need them visible.

Attached is a benchmark comparing number boxes with float-objects.

thanks for that test, good to see real numbers, otoh, floats only have a terrible performance if they are 'updated' (= receive a message.)

since there was already a good testlayout, I thought it might be worth to test the difference between s/r and cord connections. guess what. s/r were faster. I guess this is, because you need more cord connections to go from the random to the subpatch to the abstraction. but for me that was new. the overhead is not big, though, only around 10%.

but, sending messages in general (s/r of cord connections) is quite a bottleneck in pd and I wonder if that could be improved. This is where any other programming language wins :(.

marius.

Ciao

---

---

Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list