Also, I’ve been benchmarking different scenarios with my performance patch on my new MBP.
Knowing that there is a ± 12% CPU base load for IR reverbs, here are the figures I get for the general CPU load in those different scenarios :
using the old multi cable hack, around 22%
using j.pack≈ in patchers, around 30%
using j.pack≈ in polys (no parallel processing), around 35%
using j.pack≈ in polys (with parallel processing), around 33%
Though, these figures seem to be not-so-reproducible…(even if the general tendency is ± that)... and as a general observation, the CPU meter seems to be less stable when using JAG (it oscillates quickly in a ±15% range)
From @bltzr :
Also, I’ve been benchmarking different scenarios with my performance patch on my new MBP. Knowing that there is a ± 12% CPU base load for IR reverbs, here are the figures I get for the general CPU load in those different scenarios :