"So this could be put to bed…"?
CPU usage and latency in the Pitch block page in the Wiki explains why it's not easily done and I recommend reading it, but in short, pitch detection adds latency.
This is what I see:
- The FX3 can run two Pitch blocks because it’s the equivalent of two FM9 units.
- Many of the Pitch block models have the option of picking where the detection occurs. The FM9 has the same option for its single Pitch block, but it doesn't have the CPU resources to run two without doubling the latency.
- The blocks could be written so that they ONLY can take their detection from the same source or a single detector routine, but we know that people won't want that limitation. One person will say "Sure! Do it!" and another will say "No way!", so it's a stalemate.
- People think that Reverb or other blocks could be sacrificed, but from my reading of the Wiki, the manual and the Blocks guide, that won't reduce the latency, it'll slightly reduce the dedicated core's load but that's invisible to us. As is, we can run dual reverb blocks at the highest quality and the FM9 won't even blink so reducing the quality or removing a block won't help. Again, see the last sentence in the third bullet.
- Whether it’s technical or otherwise, it’s not going to change with continued requests UNLESS Cliff finds a way to make it happen in a way that meets his requirements. He is always looking at ways to improve the system and we need to accept that because he's an extremely benevolent dictator. The request has already been seen MANY times, and repeatedly asking "But WHY!?" and saying "I WANT IT!" won't change it but it could result in the car getting turned around.
- There is absolutely no way this restriction is tied to differentiating the different units; That makes no sense for a company working with embedded computers where you MUST get the most out of the hardware or you're wasting money. Fractal wants to see all the systems be the best they can be, and nobody wants that more than Cliff because then they'd CRUSH the competition.
Thank you so much Greg for your comprehensive explanation, I think everybody here loves when there's competent people explaining the inner workings of our FAS systems
I think the 2 pitch blocks request will continue to appear from time to time because for a new FM9 users it can appear strange they can't use an octaver/pitch (meant as virtual capo) + a separate harmonizer.
I also had in the past a couple of less powerful systems that simply did it, considering a "pitch" (capo/octaver) as a different block type from any "harmonizer".
In my experience the main use of 2 pitch blocks is to put:
- a Capo/Octaver before any distortion
- an Harmonizer after any eventual distortion, for a better sound treatment
I suppose everybody can accept a longer latency in FM9 if present only with these 2 blocks active,
and
most people
(not all indeed, we both know...) may accept in FM9 that there's
only one pitch detection point, as a good compromise,
e.g. kept somewhere
before distortion.
NOTE:
For people having both AF3 and FM9, this can also solve one of the more "annoying" AX3/FM9 presets' compatibility issues:
when I move AF3 presets to FM9 I'm forced to split and double them (everyone with 1 of the 2 pitch blocks that
must be anyway and forcibly named "
Pitch1"
, not named "
Pitch2"..) then spread PC mapping pointing accordingly.
This presets doubling also adds more gap switching cases with FM9.