Is it time for Fractal to upgrade the DSP chips in the Axe FX II/Xl

[zenaxe]

That's exactly what the Avid cards use the FPGA for: mixer routing. All the effects processing is done on the onboard DSPs.

Yes. When the person said FPGAs for this application, it was immediately clear they were confused wrt the hardware side of things. You might as well say "Ball bearings, dude." Ooookay.

yek

You can cramp a lot in a preset by designing a preset smart. For example, the grid below requires between 80-90% CPU.


COMP: the ""Pedal" compressors require less CPU than the Studio type. In this preset a Studio Comp is used.

PEQ: the first PEQ has the low-cut and high-cut characteristics of a "clean boost" Klon Centaur. It isn't the same as a Drive block but it requires a lot less CPU than a Drive.

PHASER, FLANGER, DRIVE, DELAY: all using X/Y. Also note that some Drive types take up more CPU (up to 4% more) than other types.

AMPS: I don't use the advanced stuff. Example: using OUT COMP increases CPU usage (depending on its setting).

CAB: I'm using UltraRes, mono. Saves CPU when compared to stereo. When I want to use multiple IRs I mix them into a single one, using Cab-Lab (to be released soon). A Mono cab suffices because all my stereo stuff comes after the Cab. Also, I'm not using the preamp simulation, saving CPU.

REVERB: I'm using the Spring Reverb. It has no High Quality mode, but it has been improved some firmware revisions ago and sounds awesome with my clean and lead tone tone. The Spring requires not nearly as much CPU power as other reverb types. If I need a specific other type of reverb I put in a dedicated preset.

The noise gate in the Input/Gate block is off, saving CU again.

No! Thinking makes my head hurt! I want to be able to use all features at max settings all the time. Just give us that. Cheap too, please.

But seriously, this is what I was talking about. I do not have very many presets at all which push the 90% barrier and the ones that do are very wet. I think it must be because my base presets were all created a couple of firmwares back, so I am only adding new stuff as needed.

 

[Sidivan]

Honestly, if the Axe FX 3 came out tomorrow, I don't know that I would buy it. I've reached a point where I really love the sounds coming out of my rig and yet I still have literally hundreds of other options that are just as good. I have no idea why I would need to upgrade unless there was some amazing convenience feature (like the FASLink, which I secretly pine for, but have not upgraded to get).

 

[Bill Piersall]

You can cramp a lot in a preset by designing a preset smart. For example, the grid below requires between 80-90% CPU.

COMP: the ""Pedal" compressors require less CPU than the Studio type. In this preset a Studio Comp is used.

PEQ: the first PEQ has the low-cut and high-cut characteristics of a "clean boost" Klon Centaur. It isn't the same as a Drive block but it requires a lot less CPU than a Drive.

PHASER, FLANGER, DRIVE, DELAY: all using X/Y. Also note that some Drive types take up more CPU (up to 4% more) than other types.

AMPS: I don't use the advanced stuff. Example: using OUT COMP increases CPU usage (depending on its setting).

CAB: I'm using UltraRes, mono. Saves CPU when compared to stereo. When I want to use multiple IRs I mix them into a single one, using Cab-Lab (to be released soon). A Mono cab suffices because all my stereo stuff comes after the Cab. Also, I'm not using the preamp simulation, saving CPU.

REVERB: I'm using the Spring Reverb. It has no High Quality mode, but it has been improved some firmware revisions ago and sounds awesome with my clean and lead tone tone. The Spring requires not nearly as much CPU power as other reverb types. If I need a specific other type of reverb I put in a dedicated preset.

The noise gate in the Input/Gate block is off, saving CU again.

I agree. With some planning you can make things work. But....like I said earlier, my preset is pretty simple and pushes very close to limits if I choose to use new feature.
Comp, Wah, Drive, Amp, Cab (stereo UR w/preamp sim, Delay, Chorus, Reverb (high quality). I also use Scenes and Modifiers.

I eliminated comp and use Out Comp in Amp Block. Went to single UR cab, dropped Reverb to Standard Quality and it's fine.

I will get Cab Lab to mix IR's and that will help.
I have no problem working within the limits we have. With so many presets and options to reduce CPU, it's easy.
I would much rather Cliff continue to develop the highest quality algorithms and push the CPU limits all he wants! In the end we win.
Eventually, when technology allows it and it makes good business sense, we will see a new product.
Until then, I'm a happy camper with the way things are going. Everyday this product surprises me, and just when I think I've got it nailed down, Cliff comes along and drops another tone bomb on us!

Good times!


Sent from my iPhone

 

[zenaxe]

Honestly, if the Axe FX 3 came out tomorrow, I don't know that I would buy it. I've reached a point where I really love the sounds coming out of my rig and yet I still have literally hundreds of other options that are just as good. I have no idea why I would need to upgrade unless there was some amazing convenience feature (like the FASLink, which I secretly pine for, but have not upgraded to get).

Yes, this is why I am (selfishly, I admit) rooting against creation of a new flagship unit as long as possible. The current unit gets every tone I could imagine and exceeds my abilities right now... I would likely not "upgrade" barring a hardware failure or major windfall since I certainly cannot justify it based on the current hardware.

But, I like to get the regular freebies and content updates of being on the current gen and if the first generation units are any indication, user discussion with associated tips, preset, TM, and IR sharing would straight up die once a III is released.

Of course it is understood that progress cannot be stopped. I just hope to ride the wave for bit longer, yet.

 

[jdosher]

For all practical purposes, those are unavailable. Example: last time I checked, Mouser had three of them in stock. If you wanted more than three, lead times were nearly half a year. You can't build a product on that.

And it'd be a major undertaking to switch to an entirely different processor architecture. You have to learn a new development environment, rewrite massive amounts of code from the ground up (especially anything written in assembly), completely redesign portions of the circuit board, and abandon all the years of experience you've gained working with a given processor. I work with Atmel AVR processors anytime I need a small microcontroller, and while there are many other more powerful processors out there, the skills and tools I've acquired over the years far outweigh the effort that would be required to jump ship to another processor. Fractal has used the TigerSHARC from the start, and I would be very surprised if they didn't continue to simply upgraded to newer versions as time goes on (as happened when going from the Standard to the Ultra to the 2).

 

[joecooll]

AXE FX IS A KILLER :encouragement:

 

[mwd]

Awesome time for me to remember growing up where high-tech was a switch that could turn on reverb, tremolo and switch between a 2 channel amp featuring a clean and distorted channel. Or my Marshall stack that featured one sound. Two if you split coil and rolled back guitar volume.

My Ultra was (and is) over the top and now my XL is just pure fantasy.

 

[Severed]

While it's great that many are happy with basic usage, there are also many that push the unit to its limits.

I was sad 20 minutes after buying as i realized the limits of the CPU. As a result i can't use many features of the Axe unless i make sacrifices in which i am just not willing to do.

While there is still room for improving the software, in my opinion the hardware is falling behind the times.

Yes, the DSP chips are not advancing in processing power, but why can't we just use more of them like today's Intel CPUs that are still the same speed as 10 years ago except they run 8 cores instead of 2 which really does give you 4x the processing headroom.

 

[pdup]

Imagine a guy that invents a piece of hardware - and some initial software. "Oh, this works, but I am able to refine nearly everything for years to come for the benefit of my quest and my customers".
More than the 90% worshipped him and his work. Some were never satisfied but went on and on asking for more.
Thanks Yek for your write about CPU usage.

 

[iaresee]

Yes, the DSP chips are not advancing in processing power, but why can't we just use more of them like today's Intel CPUs that are still the same speed as 10 years ago except they run 8 cores instead of 2 which really does give you 4x the processing headroom.

Because. Parallelization doesn't just happen magically. You can't really randomly assign processing to any of three DSPs. And off-chip signal routing introduces delays.

It's not an easy problem to solve by a long shot. The cost would like make the "just buy a second unit" seem not so bad.

Cliff and crew are exceptionally good at making their technology look easy, but it is anything but.

Maybe what it does isn't for everyone. Certainly it's not for a few outliers like yourself who were disappointed after only "20 minutes" with the unit.

 

[smcrosby]

I agree. With some planning you can make things work. But....like I said earlier, my preset is pretty simple and pushes very close to limits if I choose to use new feature.
Comp, Wah, Drive, Amp, Cab (stereo UR w/preamp sim, Delay, Chorus, Reverb (high quality). I also use Scenes and Modifiers.

I eliminated comp and use Out Comp in Amp Block. Went to single UR cab, dropped Reverb to Standard Quality and it's fine.

I will get Cab Lab to mix IR's and that will help.
I have no problem working within the limits we have. With so many presets and options to reduce CPU, it's easy.
I would much rather Cliff continue to develop the highest quality algorithms and push the CPU limits all he wants! In the end we win.
Eventually, when technology allows it and it makes good business sense, we will see a new product.
Until then, I'm a happy camper with the way things are going. Everyday this product surprises me, and just when I think I've got it nailed down, Cliff comes along and drops another tone bomb on us!

Good times!


Sent from my iPhone

Good times indeed Bill - everyone should check out Cooper's incredible Master Class which includes excellent coverage of scenes and of multiple modifiers being controlled by one expression pedal (e.g., simultaneous control over input trim, reverb and delay modifiers, each with its own distinct programmable sweep curve for blending in different amounts of trim and effects across a single pedal sweep from heel to toe). I can't wait to implement this in my brand new shiney XL!

 

[Admin M@]

I know of a fantastic upgrade to the Axe-Fx II XL......




.....




.....






Two Axe-Fx II XLs!

:devilish:






.

 

[Shreddy Better]

Why not just buy another XL and run 2 if one wants crazy rigs. It would be cheaper than an 8 core unit I would assume.

Why just have one if you can have 2 at just twice the price.

The axe fx is absolutely amazing as is, if one wants mega options than wire up 2 to taste. Even that would be so much simpler than a lot of peoples multi-amp mega pedal boards.

While it's great that many are happy with basic usage, there are also many that push the unit to its limits.

I was sad 20 minutes after buying as i realized the limits of the CPU. As a result i can't use many features of the Axe unless i make sacrifices in which i am just not willing to do.

While there is still room for improving the software, in my opinion the hardware is falling behind the times.

Yes, the DSP chips are not advancing in processing power, but why can't we just use more of them like today's Intel CPUs that are still the same speed as 10 years ago except they run 8 cores instead of 2 which really does give you 4x the processing headroom.

 

[Shreddy Better]

just seconds too late

 

[Severed]

Because. Parallelization doesn't just happen magically. You can't really randomly assign processing to any of three DSPs. And off-chip signal routing introduces delays.

It's not an easy problem to solve by a long shot. The cost would like make the "just buy a second unit" seem not so bad.

Cliff and crew are exceptionally good at making their technology look easy, but it is anything but.

Maybe what it does isn't for everyone. Certainly it's not for a few outliers like yourself who were disappointed after only "20 minutes" with the unit.

So on the idea of buying a 2nd axe fx to "share" the load of CPU, why would it be such a stretch to design the Axe 3 to basically be two Axe FX II units in a single enclosure that do the exact same thing.

In a nutshell this amounts to doubling the processing and memory does it not?

I never implied it would magically happen or be easy, but in this world technology is advancing faster than ever, and eventually they are going to have to overcome the limits of the current flagship.

by the way. on the topic of cost. If you can suggest I can afford TWO axe fx units why would I somehow not be able to afford or not want to buy that same amount of power in single unit?

 

[iaresee]

So on the idea of buying a 2nd axe fx to "share" the load of CPU, why would it be such a stretch to design the Axe 3 to basically be two Axe FX II units in a single enclosure that do the exact same thing.

Because, with two units a human brain is making decisions about where to run different pieces of the signal chain, not a computer algorithm. Despite evidence to the contrary on so many occassions, humans are still smarter than computers. And this kind of decision is easier to make with a human brain.

In a nutshell this amounts to doubling the processing and memory does it not?

Your naive assumption here is that "just adding more" is a simple thing to do. Soldering these things on to the board and wiring them together is not the hard part. Deciding how to distribute the computations across them is the hard part. And it's a very hard part.

I never implied it would magically happen or be easy, but in this world technology is advancing faster than ever, and eventually they are going to have to overcome the limits of the current flagship.

Not true. They can wait. And eventually the DSP technology will move on and make this less-hard, preferably with a single-chip solution.

All this armchair engineering in this thread amuses me greatly.

 

[Clive]

An ADA MP-1 amp block needs 4 or 5 Tigersharks to run. I'll wait for Axe FX IV to have it in the box (Axe FX III would be too early).

 

[Severed]

Because, with two units a human brain is making decisions about where to run different pieces of the signal chain, not a computer algorithm. Despite evidence to the contrary on so many occassions, humans are still smarter than computers. And this kind of decision is easier to make with a human brain.

1. I never implied any of that.

Your naive assumption here is that "just adding more" is a simple thing to do. Soldering these things on to the board and wiring them together is not the hard part. Deciding how to distribute the computations across them is the hard part. And it's a very hard part.

2. Again, this isn't what I said or implied.

Not true. They can wait. And eventually the DSP technology will move on and make this less-hard, preferably with a single-chip solution.

3. Thankfully, you are not FAS. I'm glad you think they can wait. I happen to think otherwise.

All this armchair engineering in this thread amuses me greatly.

Ya, maybe you should get out that armchair then huh cause the only one doing it is you.

Seriously though? what a dick comment. Your superiority complex needs a check.

 

[NothingButInitials]

Time to get the popcorn!

 

[iaresee]

Seriously though? what a dick comment. Your superiority complex needs a check.

LOL. You bet.