Why not 2 amps, and zero fx possible?

Tahoebrian5

Tahoebrian5

Fractal Fanatic
I'm having a hard time understanding why there are limits, other than cpu usage, on the number of fx blocks or amp blocks. Why not just allow users to load up whatever they want until cpu is maxed? Like 2 amps and zero fx for example.
 
FAS would have to answer this, but my guess is that the reason you can't do this is fundamentally defined in the product architecture.

One chip is dedicated to the amp model, an accelerator is dedicated to the cab model, and the other chip is dedicated to FX. If they allowed two amps, they would also have to add or increase the accelerator size for two cabs which would add to the cost.

If you need two amps, go with the AxeFX or another product that allows for more than one amp.
 
Tahoebrian5 said:
I'm having a hard time understanding why there are limits, other than cpu usage, on the number of fx blocks or amp blocks. Why not just allow users to load up whatever they want until cpu is maxed? Like 2 amps and zero fx for example.
Click to expand...

The reason you don't understand it is the same reason you don't design hardware or firmware; you're not educated in the specifics of hardware system design. And even if you were you'd STILL not know the technical details of the AX8's architecture or firmware...

In general, the hardware architecture supports capabilities and therefore imposes certain restrictions on the device; the software enables the hardware. Some new capabilities or features can be added or limits alleviated with software updates (in fact it can be quite amazing what can be done with the software at times) but some limitations, those imposed by the core hardware/architecture, no amount of software can overcome. If you open a modern piece of technology up it is not just one piece of black goo that can take any generalized form; it is a *bunch* of *very* specially selected/designed and connected devices whose arrangement into a circuit enables a very specific set of hardware requirements. When combined with the system software, you get a functioning hardware/software system that has a very real set of limitations on what it can achieve, no matter how clever the firmware author.

Only someone who understands the general technology and the specific architecture of the device in question can say precisely where the lines are drawn by this balance (and sometimes at the gray areas things thought to be hard lines can be blurred and even the designer can surprise themselves) and there is only one person who understands the AX8 in this way. So whether you understand it or not, if FAS says that is the what the product can support, that is what it can support.
 
Last edited:
I understand the question though, he just sees the CPU/RAM as a "calculating power supply" for the unit, that should be distributed to any block you want. Which is conceptually not stupid.
Thing is, the software calculation are allocated specific chips already. Assigning any block to any chip would require a new design.
 
Cliff has stated elsewhere that the AX8 has two chips: one dedicated to the Amps and another dedicated to FX. These lower priced and spec'd chip (relative to the Axe FX II) cannot do more than one amp at a time.

It's an architectural design limitation rooted in hitting a price point while maintaining the ability to deliver the same amp modeling in the Axe FX II.
 
Yeah, it seems like a design limitation really, The Helix has the same 2 chips and can do anything with either chip. However, if you can see the AX8 as an extension of the FX8, it's essentially adding in a CPU to handle the amps and letting the other chip do the same thing it was doing in the FX8. To do otherwise, seems like it would require a quite a bit of reprogramming. And lets also remember that it's not one CPU itself doing all the amp/cab modelling. I believe the cab portion is a coprocessor. So if you wanted the effects chip to do amps, you might also have to add all that architecture in. And the end result would be a unit that would would be used with one amp block and effects 99.99% of the time anyway so why bother?
-
Austin
 
atquinn said:
Yeah, it seems like a design limitation really, The Helix has the same 2 chips and can do anything with either chip. However, if you can see the AX8 as an extension of the FX8, it's essentially adding in a CPU to handle the amps and letting the other chip do the same thing it was doing in the FX8. To do otherwise, seems like it would require a quite a bit of reprogramming. And lets also remember that it's not one CPU itself doing all the amp/cab modelling. I believe the cab portion is a coprocessor. So if you wanted the effects chip to do amps, you might also have to add all that architecture in. And the end result would be a unit that would would be used with one amp block and effects 99.99% of the time anyway so why bother?
-
Austin
Click to expand...
Yes, except that Helix Modeling != FAS Modeling.

I don't mean that to be elitist at all. It's entirely possible that there is significantly more processing power required to perform the calculations needed for the Quantum amp modeling engine than the algorithms Line6 is using. FAS has stated that the Cab IRs don't take up a lot of CPU power, and the fact that Cab IR processing has been moved off to it's own co-processor would seem to indicate that one amp sim is already pushing the envelope for the lesser processor in the AX8.

But of course, this is all just speculative armchair engineering with no strong basis in fact. :)
 
Cliff has stated that certain amps (Class A models to be specific) take up 90% of the amp CPU, so you don't have room to allow for multiple amps on that CPU.

FWIW, the Helix chips aren't flexible either. One is for Path A, one is for Path B. You can't use both for a complex Path A. The difference is, those guys don't split out amp modeling to a separate chip, which means they have to sacrifice complexity and comprehensive details in each model. They can't design an amp model that takes up 90% of the Path A CPU; there would be no room for anything else.
 
That makes sense that the amp modeling is fundamentally different than fx modeling. But this doesn't explain why there is a limit to how many instances of a type of fx block can be used. I know it could play havoc with the midi messages that enable / disable each block ( delay block 1 is set to a different message than delay block 2), but other than that it seems like a needless limitation. Obviously I have zero knowledge on the design of these things, just curious about the limitations.
 
benvigil said:
Yes, except that Helix Modeling != FAS Modeling.

I don't mean that to be elitist at all. It's entirely possible that there is significantly more processing power required to perform the calculations needed for the Quantum amp modeling engine than the algorithms Line6 is using. FAS has stated that the Cab IRs don't take up a lot of CPU power, and the fact that Cab IR processing has been moved off to it's own co-processor would seem to indicate that one amp sim is already pushing the envelope for the lesser processor in the AX8.

But of course, this is all just speculative armchair engineering with no strong basis in fact. :)
Click to expand...

Limitation was the wrong word. Intentional is closer to what I mean. The point is , the Helix has 2 chips that are designed to do the same thing. If the AX8 was designed the same way, then it would be able to do one amp with each chip. But it's not designed that way. And yes, part of the reason that it makes sense for the Helix to be designed that way is because the amps don't take an entire chip to run. You could just as well ask why the other chip on the Axe FX 2 can't run amps. After all, if one chip can run 2 amps, it seems like the other could run one amp and minimal effects. Why wasn't it built that way? Because it wasn't, that's why :D At the end of the day, it's about results, and I don't think there's much to complain about where that's concerned.
-
Austin
 
ethomas1013 said:
FAS would have to answer this, but my guess is that the reason you can't do this is fundamentally defined in the product architecture.

One chip is dedicated to the amp model, an accelerator is dedicated to the cab model, and the other chip is dedicated to FX. If they allowed two amps, they would also have to add or increase the accelerator size for two cabs which would add to the cost.

If you need two amps, go with the AxeFX or another product that allows for more than one amp.
Click to expand...

I think Cliff also said that the fx dsp will be used slightly for cab model (mostly in the accelerator) and some management functions.

This is one area where Fractal and Line 6 took different paths (pun intended). Fractal's architecture dedicates the DSPs for specific purposes. Line 6 designed an architecture that will allow users to dynamically use the DSPs in the manner you describe.
 
philipacamaniac said:
FWIW, the Helix chips aren't flexible either. One is for Path A, one is for Path B. You can't use both for a complex Path A. The difference is, those guys don't split out amp modeling to a separate chip, which means they have to sacrifice complexity and comprehensive details in each model. They can't design an amp model that takes up 90% of the Path A CPU; there would be no room for anything else.
Click to expand...

Is that correct? I haven't looked into this at all and would find it more limiting than the Fractal design unless I'm misunderstanding. I thought that if you use two amp models and each uses 60% of a dsp, the first amp could be on dsp one and amp two could have 40% on dsp one and 60% on dsp two, leaving you 50% dsp for effects going to both. If not, this will be interesting...
 
DavidE said:
I thought that if you use two amp models and each uses 60% of a dsp, the first amp could be on dsp one and amp two could have 40% on dsp one and 60% on dsp two, leaving you 50% dsp for effects going to both. If not, this will be interesting...
Click to expand...

that would probably increase latency. i doubt you can have 2 DSP work on one process - like a single amp block - for real-time audio without major side-effects that render it useless. then again, i don't know anything about this either :)
 
DavidE said:
Is that correct? I haven't looked into this at all and would find it more limiting than the Fractal design unless I'm misunderstanding. I thought that if you use two amp models and each uses 60% of a dsp, the first amp could be on dsp one and amp two could have 40% on dsp one and 60% on dsp two, leaving you 50% dsp for effects going to both. If not, this will be interesting...
Click to expand...


Ok, so one amp per dsp leaving 60% of each DSP to run effects to be used with either amp/cab model?
 
DavidE said:
Ok, so one amp per dsp leaving 60% of each DSP to run effects to be used with either amp/cab model?
Click to expand...

Correct. There is no sharing of blocks between the DSPs (Paths) and you have to manually assign what goes to which DSP. So if you are using both paths in a serial function and you overload Path A, you would have to but further blocks on Path B. Definitely a different way of dealing with overloading the system resources than Fractal takes. It will be interesting to see how much of an issue it is. So far, I don't recall seeing any reports of people overloading the CPUs on the Helix units. I never stack effects, so I certainly won't be doing it. Which is one of the reason why something like the AX8 will allow me to do almost the exact same things I was doing on the Axe FX 2.
-
Austin
 
You must log in or register to reply here.
Back
Top Bottom