I’ve got a few questions regarding the cost of 3-EQ and Multiband EQ. My project ran into an issue on the Switch due to many instances of the 3-EQ effect being used. One of our engineers reported “90% of one of the cores” was being eaten up by 20 instances used on several events (some events even had multiple 3-EQ effects loaded in). I’ve since made some major adjustments and replaced some silly usages with HighPass & LowPass effects, but for other events, we still need the same automated frequency action on our vehicle engines.
My questions are:
- What are the actual costs of using 3-EQ compared to Multiband EQ?
- Is it possible to achieve the same sonic results with a Multiband EQ as originally heard through the 3-EQ? (I saw elsewhere on the forum that 3-EQ runs in parallel while Multiband runs in series. I understand the difference, I’m just concerned about re-creating the right sound with a different EQ algorithm.)
From what I know, Multiband EQ is a more recent (so maybe better optimized?) addition to FMOD than 3-EQ. And HighPass and LowPass are legacy (for retro-compatibility), so maybe better to avoid them…
I just did a little test with 20 effects of each on a single event, and Multiband EQ seems cheaper on the CPU. Here’s the result:
But maybe with different settings, the result would change, I don’t know…
Nobody can answer but you. Just try! (but I would tend to say it’s often possible)
Thanks for running that experiment, Alci!
Did you have all of the bands enabled when the Multiband was loaded in? It’s still impressive how much cheaper it is compared to 3-EQ! And what you say about legacy effects makes sense. I think I’ll use another Multiband and only enable its Hi/Lo Pass options.
Thank you again!
No, just one band, because I supposed in some cases you would use the 3-EQ only for one of its filter. But I should try the multiband with 3 bands enabled and see if that changes anything…
Multiband EQ with 3 bands enabled (low shelf, peak, high shelf) is still cheaper than 3-EQ (less, though).
That’s fantastic, thank you Alcibiade YET AGAIN! This may very well solve our CPU issues