Latency of amp modelers

Based on what?

The FM3 and FM9 use the same CPUs as far as I am aware.
from the wiki…the fm9 if managed properly by the designers, is more powerful for the final sound:

CPU​

  • The processing power of the III is the highest, allowing the most effects and functionality in presets. The III Turbo offers even more CPU power.
  • On the FM3, the delays share a core DSP with amp modeling. Reverbs have an impact on overall CPU usage.
  • On the FM9, amp modeling, delays and reverbs each have their own dedicated core DSP. This enables ultra-high quality reverbs without a large impact on overall CPU usage.
 
from the wiki…the fm9 if managed properly by the designers, is more powerful for the final sound:

CPU​

  • The processing power of the III is the highest, allowing the most effects and functionality in presets. The III Turbo offers even more CPU power.
  • On the FM3, the delays share a core DSP with amp modeling. Reverbs have an impact on overall CPU usage.
  • On the FM9, amp modeling, delays and reverbs each have their own dedicated core DSP. This enables ultra-high quality reverbs without a large impact on overall CPU usage.
That says nothing about latency, though.

Processing power of the FM9 is 2X the FM3 because it uses 2 of same CPUs that the FM3 uses one of.
 
I swear at some point I read something from Fractal saying the FM9 is lower, but I could all just be in my mind :grimacing:
You could be right... I don't recall seeing anything called out, but might be I just missed it or forgot ;)
 
That says nothing about latency, though.

Processing power of the FM9 is 2X the FM3 because it uses 2 of same CPUs that the FM3 uses one of.
i know…we can agree to disagree…wrt to latency…i gotta believe parallel processing would have a direct benefit.
 
i know…we can agree to disagree…wrt to latency…i gotta believe parallel processing would have a direct benefit.
Possibly true but depends on what is being processed ;)

Anyway, I'm not saying it's not - only that I've not seen the information posted.
 
Possibly true but depends on what is being processed ;)

Anyway, I'm not saying it's not - only that I've not seen the information posted.
Oh, on another note concerning cores…the wiki impies 2 cores…but going back to the announcement it was proclaimed as 4 cores: …The FM9 completes our product family, standing proudly between Axe-Fx III and FM3 in terms of power, portability and price. It is our most powerful floor unit ever, with a four-core architecture able to run impressively complex presets.

https://forum.fractalaudio.com/threads/announcing-the-new-fm9-amp-modeler-fx-processor.176122/

edit: 4 cores, 2 chips
from clif:
…The III uses (1) dual-core Texas Instruments DSP. The FM3 uses (1) dual-core Analog Devices DSP. The FM9 uses (2) dual-core Analog Devices DSPs.

The TI DSPs are much more powerful than the Analog Devices DSPs per clock and run at around twice the clock speed as well. So one TI DSP core is about four times more powerful than one Analog Devices DSP core.

If we normalize processing power to the III it would be:
Axe-Fx III: 100%
FM9: 50%
FM3: 25%

So why not use the TI DSPs in everything? Power. The TI DSPs use more power and generate more heat requiring active cooling. They are also more complicated to use requiring dedicated clock generation units, multiple power supplies with specific sequencing requirements, etc., etc.
 
Last edited:
When you compare it to a pc, the latency changes as you run different sound interfaces with it, even though the main processor stays the same.
A better main processor might allow a lower buffer size, right, but with a bad sound interface you can't get decend results.
So I think this is not about single parts, it's the whole system and how everything fits together.
 
Cab blocks do not add any latency.

Could someone explain in layman's terms so my small brain can understand how this is possible? I believe Cliff is telling the truth, I just don't understand how it works that a 20 ms IR doesn't add any latency.
 
Latency is the gap before the signal.
The length of an IR determines how long a signal is "processed" by the IR, and has no effect on latency unless the IR itself contains silence at its start.
 
Back
Top Bottom