That's kind of how it works, though. Digitech sold add-on kits for some of their products that allowed delay and reverb spillover. The kit consisted of a set of instructions and a second processor. One way or another, you have to reserve space for the effect you're spilling over, robbing the next preset of memory and CPU horsepower.
Sorry, I'm not buying that. I mean, of course, I'm sure what you're saying is correct, but I don't believe that's the only way to achieve results. At least, not for something that's essentially a computer (as opposed to a stompbox that lacks things like RAM, etc).
This argument is inconsistent for a number of reasons. First of all, the nature of "spillover" is such that it deals with sounds that have already happened as opposed to processing in real-time sounds that are currently happening. The idea that implementing proper spillover would unreasonably increase the AxeFX's real-time processing load doesn't make sense since the concept of "spillover" does not actually involve adding effects to notes as they are happening, but rather allowing the effects already added to previous notes to run their course.
Second, regardless of the first point, real spillover would clearly not require twice the processing power present in the entire device. That much should be obvious to anyone that stops to think about it for a second. The AxeFX can process a LOT of complex delay/reverb information. We already have access to 5 delays (if you count Megatap and Multi delays) and 2 reverbs per preset. It's potentially twice that many depending on whether or not the AxeFX processes X/Y states in parallel (IIRC, it does). So, even if Fractal implemented spillover by simply splitting your signal and mirroring your spacial FX blocks behind the scenes on hidden, global-style "Delay 3" and "Reverb 3" blocks and activated those blocks whenever a patch change is sent (which would potentially be the laziest, least-efficient way to go about this), you STILL wouldn't need a separate DSP's-worth of power to do it. Such a setup would only require an additional amount of processing power comparable to the processing power eaten up by the delays and 'verbs in any given preset ONLY. Would there be limits to this? Yes. Could you run all the spacial FX at once in a preset and ALSO have this spillover? Probably not, but similar limitations already exist (want a Looper with Undo, then you'll have to deal with shorter loops). Would it work? Given the fact that this method is premised off of functionality we already use all the time, I don't see why not.
Here's another idea. What about the Looper? How does that work? It seems to involve the act of recording short pieces of already-processed audio without having to require a separate DSP framework. Well, what if you take that concept, shrink it down to couple seconds (no spillover is going to be 15 seconds long), make it into an internal system process, and call it a "system spillover buffer". It just sits in the background at the end of your chain recording the last 5 seconds (or whatever) of your sound, then it dumps that to the main outputs when a patch change request is sent. Again, will that take up some system resources? Yes. Will it take up enough to be completely impractical? Well, the Looper doesn't. It wouldn't be perfect, but I would prefer "less perfect" to "total PITA, if available at all".
Also, Digitech sold add-ons that allowed an aftermarket co-processor to enable spillover on one of their products? That's great...CAN FRACTAL!?
I obviously don't know the ins-and-outs of the AxeFX platform. My point in writing this is not to say that these are necessarily the best solutions to the problem, just that they are examples of things that COULD be done that would have their limitations, but still be a lot more useful than the current complicated, messy, half-assed thing we have now. To be honest, I suspect the real reason Fractal has only made marginal gains on this problem over the YEARS people have been complaining about it is because they don't care. It's just not a priority. They have their overdrive-simulation pissing match with Kemper to worry about. "Never mind the fact that our product lacks basic, essential functionality found in almost every other FX processor currently on the market, we gotta squeeze out that extra 4% of overdrive realism to maintain market share." Fine. Whatever.
Proper spillover - It's important. Just ask any of the professional guitarists I describe the problem to who respond, "uh uh, deal breaker" then proceed to write off the AxeFX completely. Honestly, if there was another company that had a product remotely comparable to the AxeFX, I'd bail on Fractal in a second. But there isn't, so here we are. I'll just continue pissing my life away at 4am trying to ensure the 30-some-odd patches I'm creating for this project alone have all their spacial effects sync'd in a perfectly linear fashion.