Fractal Audio Firmware Update Thread

[Boudoir Guitar]

No. Traditional IRs will not work with the Dyna-Cab code, these are a whole new thing. There's a separate loader to install the Dyna-Cab IRs into the modeler. You can combine traditional IRs with Dyna-Cab IRs but they're different creatures.

Other companies have something similar, but their IRs lack the granularity needed.

The IRs are conventional. It is a chooser of those conventional IRs. So yes.

 

[Greg Ferguson]

The IRs are conventional. It is a chooser of those conventional IRs. So yes.

No, they're not conventional. Cliff was very specific.

 

[Boudoir Guitar]

Again, no, they work fine. If you chose to change the Cab block's IRs to Dyna-Cab IRs then you'll need to do some adjusting, but the factory presets, and any based on them, will continue to work fine.

He was replying to my sarcasm related to LAXU's suggestion of ditching legacy mode altogether to make room for more dynacabs.

 

[JiveTurkey]

 

[Will Chen]

The IRs are conventional. It is a chooser of those conventional IRs. So yes.

Well, in theory yes. In implementation, who knows? I remember Two Notes using some proprietary format for WOS years ago, while the end result was more or less an IR, how the data was stored and how the app accessed the data was something different IIRC. I mean, Fractal is already using sysex files instead of wav files, so...who knows. Maybe a single cab implementation here is just 1 larger sysex rather than a ton of individual wav files.

 

[Boudoir Guitar]

No, they're not conventional. Cliff was very specific.

He was specific that they are just conventional IRs captured with a mic robot.

 

[Will Chen]

He was specific that they are just conventional IRs captured with a mic robot.

Sure...but from an implementation perspective, Fractal does something different. For example, I've got a York Audio Mix IR with a size of 72 KB as a wav file, same IR in sysex format is 11 KB (EDIT: in fact, I think all sysex IRs are 11KB). Maybe that's just compression, maybe it's something else, but it's definitely something proprietary.

 

[Gearzilla]

 

[Greg Ferguson]

We get the corresponding SIC with the cabs

I think this is a win. Probably 99.99% of the time I want the stock SIC. The only time I don't is when I'm trying to blend up an amp that isn't part of the regular models, like the Tone King Imperial or the Mesa Lonestar Special. We have things that are close, but stirring in some different settings gets closer. I get tired of trying out a different speaker and then having to switch it, and too often I forget then wonder why it doesn't sound right.

 

[jellodog]

I get tired of trying out a different speaker and then having to switch it, and too often I forget then wonder why it doesn't sound right

Yep. Happens to me all. the. damn. time.

Most definitely a desirable improvement and one that I've been hoping for since I discovered what SICs were.

Incidentally, a Dyna-Cab discussion started here too:

Axe-FX III Dyna-Cabs

@2112 uploaded a video with a quick review of Fractal's new "Dyna-Cabs" available on the public betas for firwmare v22. Sounds really damn good to me, and i like how you can blend up to four of these in a single block. The final release of v22.00 is expected to ship with 35+ Dyna-cabs, all...

thegearforum.com

Although ironically, there is probably a little more technical detail in this thread right now.

 

[Greg Ferguson]

He was specific that they are just conventional IRs captured with a mic robot.

He was specific they're capturing multiple discrete samples, that are more closely stepped. If that was all they were, then they could be considered "conventional" but the number of IRs would be overwhelming.

Beyond that, there is special software used to process them. If they were conventional, Fractal wouldn't need special software to process the IRs, they already have that in place and so do the 3rd party IR vendors.

 

[Whizzinby]

Sorry for the OT post...

Dude, I just realized that I spend too much time on this forum. I was watching TV and they showed a snipped of Pulp Fiction. My first thought was why is Whizz so upset and yelling, he always seems like a chill guy?! After a few milliseconds of brain latency, I realized the mistake.

 

[KHAAAAAAANNNNNNN!!!!!!!!]

Again, no, they work fine. If you chose to change the Cab block's IRs to Dyna-Cab IRs then you'll need to do some adjusting, but the factory presets, and any based on them, will continue to work fine.

I know that. I think you misunderstood me. I was making a joke about how getting rid of legacy cabs would break presets (to the person saying they should get rid of them) . Not adding the Dyna-Cabs.

 

[Stone]

Bottom Line Select appropriate cab your desiring for XXX reason Dial in the Mics and Bang off to Toanz

 

[Greg Ferguson]

I’d assume at this point FAS knows if it’ll work in the FM3/FM9 or not and it’s just a matter of doing the work to make it happen.

I think it’s safe to say the BIG updates will always hit the III first, but we’ve already seen some mixing up of updates among the 3 units with the FM’s getting a pedal or amp (can’t remember) before the III did. So I’d imagine the smaller stuff that’s already been introduced to the system can go right to the FM’s, but bigger ticket items need to spend some time in the III before going out to the masses.

I just tested the CPU % difference between the "Legacy" type and "Dyna-Cab": ~3%, which was consistent with 1 or 3 slots used in either mode.

The FM9 uses its cores differently so it should be able to handle Dyna-Cab without problems.

The FM3 should be able to also, but at what CPU cost is hard to say because of the differences in the processors. I *suspect* that Fractal will find some ways to run it on the FM3, and let the user decide where they want to allocate the CPU, either as Legacy or Dyna-Cab.

The Dyna-Cab IRs sound really nice but I don't think I'd use it as often on my FM3 because giving up that extra CPU when the unit already sounds really good… do I want "really good" or "extra premo good"?

The changes in the firmware since Cygnus was released are mind-boggling.

 

[yeky83]

Memory and storage are completely different things as I'm sure you know. The device doesn't need all the IRs in memory at all times. I haven't seen anything to suggest that the Helix (or any other high-end modeler) has only 32 MB of storage. Storage is dirt cheap and has been for a long time now so that'd be surprising, but maybe you've taken one apart to see.

If you're getting 28 MB from the download size, that is going to be very highly compressed (lossless compression, to be clear for those who don't understand file compression) since the IRs are so similar - compression relies on redundancy in the data. Maybe they're interpolating, I don't know, but it seems like you're basing this theory on some shaky assumptions

You don't think you can compress files? Cliff, you should really go take some undergrad engineering courses if that's the case.

So your idea is.... that any time a user changes some speaker settings, the device somehow madly real-time decompresses and/or transfers IRs from storage to memory in time for real-time audio processing?

Sure...but from an implementation perspective, Fractal does something different. For example, I've got a York Audio Mix IR with a size of 72 KB as a wav file, same IR in sysex format is 11 KB (EDIT: in fact, I think all sysex IRs are 11KB). Maybe that's just compression, maybe it's something else, but it's definitely something proprietary.

It's probably just IR length differences. Downloading from York directly probably gets you a little longer IR file.

 

[DLC86]

I just tested the CPU % difference between the "Legacy" type and "Dyna-Cab": ~3%, which was consistent with 1 or 3 slots used in either mode.

The FM9 uses its cores differently so it should be able to handle Dyna-Cab without problems.

The FM3 should be able to also, but at what CPU cost is hard to say because of the differences in the processors. I *suspect* that Fractal will find some ways to run it on the FM3, and let the user decide where they want to allocate the CPU, either as Legacy or Dyna-Cab.

The Dyna-Cab IRs sound really nice but I don't think I'd use it as often on my FM3 because giving up that extra CPU when the unit already sounds really good… do I want "really good" or "extra premo good"?

The changes in the firmware since Cygnus was released are mind-boggling.

My guess is that dyna-cabs will actually use less cpu on the fm3/fm9 than ultrares, basically the same as normal-res.
I base this assumption on what I've observed on all Fractal devices I've owned. Let me explain:

Originally came the ax8 which was the first fractal device to run IRs on an accelerator and not on the main core of the dsp, and this allowed us to load/run IRs in three different formats: normal-res (1024 samples), high-res (2048) and ultrares (~170 ms but same file size as 2048 samples).
On the ax8 the first two formats had exactly the same impact on the cpu, which was around 3% iirc and no matter if the cab block was set to mono (1 IR) or stereo (2 IRs).
Ultrares on the other hand increased the cpu usage to 6-7% for a single IR (still iirc).
This leads me to think that those accelerators are able to run two 2048 samples IRs but not the Ultrares processing which is offloaded to the main cpu.

As we know, IRs on the fm3 and fm9 run on an accelerator as well, but on these devices high-res support has been ditched for some reason (maybe just to simplify things and avoid confusion, even though it's still available on the axe fx III), although the same rules seem to apply: minimal impact on the cpu with 1024 samples IRs and no change between mono and stereo, while ultrares increases cpu usage significantly.

So in conclusion, these new dyna-cabs, being made with 2048 samples IRs, will hopefully run entirely on the accelerator and cpu usage will be the same as normal-res IRs, or maybe just a bit more for the interpolation.

 

[Will Chen]

It's probably just IR length differences. Downloading from York directly probably gets you a little longer IR file.

I looked at about 30 sysex IRs and all of them were 11kb even if the original file was shorter. So it's a proprietary format that somehow allows them to recreate it in the CAB block (or they're doing some very different than I thought).

 

[JiveTurkey]

So your idea is.... that any time a user changes some speaker settings, the device somehow madly real-time decompresses and/or transfers IRs from storage to memory in time for real-time audio processing?

It's probably just IR length differences. Downloading from York directly probably gets you a little longer IR file.

 

[DLC86]

I looked at about 30 sysex IRs and all of them were 11kb even if the original file was shorter. So it's a proprietary format that somehow allows them to recreate it in the CAB block (or they're doing some very different than I thought).

That's probably the size of a single IR slot in the memory