Another Gapless Switching Thread... Solved! (Post #136)

Paulzx said:
I can sympathise with this. As a Helix user who really wants a Fractal, I will say we have been spoilt with snapshot instant switching and a lot of available switches. The fm9 addresses the switch issue of course but still isn't gapless. I think that's hard to get your head around as a Helix user, it seems like a backwards step. I suppose you can use the superior tone argument but most of us coming over from Helix are probably wondering why fractal can't just overcome that little achilles heel so it's not even up for discussion
Click to expand...
Testify my brother!!!
 
For mancave strumming
I don't "mind the gap" but following discussions here with interest being an Axfx routing geek - great preset Mr GlennO!
 
I am guessing this would work on FM9 as we have the 2 amp blocks and mixer , it’s a great idea
Do you guys notice it taking up a ton more CPU vs channels and the gap just curious
I will check it out later this week when home
 
My name is mud said:
Transferring a mp3 in a usb key don’t take the same amount of time than a .wav. Maybe the files are just bigger and need more ressource, as they are more details in the sound . Don’t know
Click to expand...
I’m going to geek out on you - sorry. In the world of computing, there is something called Moore’s Law. This law, which has held remarkably true over time since 1975 has stated that the number of transistors on computer chips has consistently doubled every 2 years. This means (roughly) computers get twice as fast every 2 years with no or only negligible price increases. The pricing isn’t technically part of Moore’s Law, but I digress. The Helix came out in 2015. This would mean an FM9 or even more so, an FM9 Turbo, should be roughly somewhere between 8 and 16 times more capable than the Helix.

This should theoretically mean that we should easily be able to have our cake and eat it too. Now if you apply that to an Axe-FX III MK2 Turbo, that number should theoretically be 16 to 32 times more powerful than the 2015 Helix, which is still using the same silicon today as when it was released.

I firmly believe it to be theoretically possible, just not yet reality to get gapless switching with no unnatural gyrations requiring ping ponging between amp blocks and burning two of only 4 controllers in order to do so. I’m guessing this would require a massive code overhaul in order to make that a reality, which is no minor task at all and a really serious ordeal. Like, a really, really serious ordeal.

I fully acknowledge that this fat, bald, out-of-shape armchair-quarterback may not know what the hell he’s talking about though. So there is that minor consideration…. But it must be true because I really want it to be, right? I do believe that’s how I want the world to work, so it must be so.

Geek-time story hour is over folks. Nothing more to see here. Move along, move along…
 
Randalljax said:
I am guessing this would work on FM9 as we have the 2 amp blocks and mixer , it’s a great idea
Do you guys notice it taking up a ton more CPU vs channels and the gap just curious
I will check it out later this week when home
Click to expand...
It does, I tried it. I can’t say I paid attention to the before and after cpu consumption though.
 
juliatk said:
I’m going to geek out on you - sorry. In the world of computing, there is something called Moore’s Law. This law, which has held remarkably true over time since 1975 has stated that the number of transistors on computer chips has consistently doubled every 2 years. This means (roughly) computers get twice as fast every 2 years with no or only negligible price increases. The pricing isn’t technically part of Moore’s Law, but I digress. The Helix came out in 2015. This would mean an FM9 or even more so, an FM9 Turbo, should be roughly somewhere between 8 and 16 times more capable than the Helix.

This should theoretically mean that we should easily be able to have our cake and eat it too. Now if you apply that to an Axe-FX III MK2 Turbo, that number should theoretically be 16 to 32 times more powerful than the 2015 Helix, which is still using the same silicon today as when it was released.

I firmly believe it to be theoretically possible, just not yet reality to get gapless switching with no unnatural gyrations requiring ping ponging between amp blocks and burning two of only 4 controllers in order to do so. I’m guessing this would require a massive code overhaul in order to make that a reality, which is no minor task at all and a really serious ordeal. Like, a really, really serious ordeal.

I fully acknowledge that this fat, bald, out-of-shape armchair-quarterback may not know what the hell he’s talking about though. So there is that minor consideration…. But it must be true because I really want it to be, right? I do believe that’s how I want the world to work, so it must be so.

Geek-time story hour is over folks. Nothing more to see here. Move along, move along…
Click to expand...
Only the maker knows if he can. That’s just a supposition. The gap thing is on the table since the release of the product. He already reduce the gap in some previous firmware. if it were so easy for him we would already have a gapless thing, but it seems more complicated than that.
I end up simplifying my presets and the gap is ridiculous now. But yeah at first I was whining too when I was using a different amp per scene …. Even if the helix is not bad and got his qualities, that’s not the same category of product.
 
juliatk said:
I’m going to geek out on you - sorry. In the world of computing, there is something called Moore’s Law. This law, which has held remarkably true over time since 1975 has stated that the number of transistors on computer chips has consistently doubled every 2 years. This means (roughly) computers get twice as fast every 2 years with no or only negligible price increases. The pricing isn’t technically part of Moore’s Law, but I digress. The Helix came out in 2015. This would mean an FM9 or even more so, an FM9 Turbo, should be roughly somewhere between 8 and 16 times more capable than the Helix.

This should theoretically mean that we should easily be able to have our cake and eat it too. Now if you apply that to an Axe-FX III MK2 Turbo, that number should theoretically be 16 to 32 times more powerful than the 2015 Helix, which is still using the same silicon today as when it was released.

I firmly believe it to be theoretically possible, just not yet reality to get gapless switching with no unnatural gyrations requiring ping ponging between amp blocks and burning two of only 4 controllers in order to do so. I’m guessing this would require a massive code overhaul in order to make that a reality, which is no minor task at all and a really serious ordeal. Like, a really, really serious ordeal.

I fully acknowledge that this fat, bald, out-of-shape armchair-quarterback may not know what the hell he’s talking about though. So there is that minor consideration…. But it must be true because I really want it to be, right? I do believe that’s how I want the world to work, so it must be so.

Geek-time story hour is over folks. Nothing more to see here. Move along, move along…
Click to expand...
Moored law is dead. If you disagree, please show me a chip made recently that has double the transistors of the version 2 years before it. I’ll wait.
 
trixdropd said:
Moored law is dead. If you disagree, please show me a chip made recently that has double the transistors of the version 2 years before it. I’ll wait.
Click to expand...
If you stick to the strict definition of Moore’s law being a doubling of transistors every two years, I believe you may be correct. However, the important part is the doubling of the processing capability every two years. Chip manufacturers are using different methods to continue doubling the horsepower without doubling the transistor counts now. Tri-gate’s 3D transistors and IBM’s new 2 nanometer chips continue to push forward. Apple silicon is a good example in the “available now” in the real world space. But here’s a good chart to show performance over time.

https://www.cpubenchmark.net/year-on-year.html
 
Horse power is the easy part - programming such features is where things probably get interesting - in a case like this, likely involving the need to have all amp parameter value permiatations + combinations loaded in memory at all times and available to "switch to" at all times with no gap. The depth of complexity of Axefx amp models would seem to make this more challenging compared to other modellers, but not impossible I guess.
 
Last edited:
Randalljax said:
I am guessing this would work on FM9 as we have the 2 amp blocks and mixer , it’s a great idea
Do you guys notice it taking up a ton more CPU vs channels and the gap just curious
I will check it out later this week when home
Click to expand...
Amp modeling runs on a dedicated CPU core. There should be no apparent difference in CPU use between running 1 or 2 Amp blocks.
 
@marshall2553 Moore's Law is indeed an observation and it is a broken pattern today. It was once useful as a rough predictor of future transistor density, but it was always, always an oversimplification of potential system performance. In that it was little help even in its glory days of ever rising Intel share prices.

"Horsepower", even a lot more of it, won't always solve a problem. An incalculable amount money has been wasted buying new hardware when the limiting factor was the code, the network, or the person at the keyboard. Moore's Law was never a measure of efficiency and in that area we have chips that are doing amazing things today on very little power, thanks to smart phones, purpose built apps and great economies of scale.

As I said before, I don't second guess Cliff and co. The code I've written is not real time in its use which is indeed a special devotion in all fields - music, medical, driving, etc.
 
Last edited:
juliatk said:
I’m going to geek out on you - sorry. In the world of computing, there is something called Moore’s Law. This law, which has held remarkably true over time since 1975 has stated that the number of transistors on computer chips has consistently doubled every 2 years. This means (roughly) computers get twice as fast every 2 years with no or only negligible price increases. The pricing isn’t technically part of Moore’s Law, but I digress. The Helix came out in 2015. This would mean an FM9 or even more so, an FM9 Turbo, should be roughly somewhere between 8 and 16 times more capable than the Helix.

This should theoretically mean that we should easily be able to have our cake and eat it too. Now if you apply that to an Axe-FX III MK2 Turbo, that number should theoretically be 16 to 32 times more powerful than the 2015 Helix, which is still using the same silicon today as when it was released.

I firmly believe it to be theoretically possible, just not yet reality to get gapless switching with no unnatural gyrations requiring ping ponging between amp blocks and burning two of only 4 controllers in order to do so. I’m guessing this would require a massive code overhaul in order to make that a reality, which is no minor task at all and a really serious ordeal. Like, a really, really serious ordeal.

I fully acknowledge that this fat, bald, out-of-shape armchair-quarterback may not know what the hell he’s talking about though. So there is that minor consideration…. But it must be true because I really want it to be, right? I do believe that’s how I want the world to work, so it must be so.

Geek-time story hour is over folks. Nothing more to see here. Move along, move along…
Click to expand...
But the key here is this: you don't know how the processing power inside the Fractal (or the Helix) is being used ;)

I'd be willing to bet the Amp modeling in the Fractal is significantly more CPU intensive than the Helix, but that's a guess.

However, there's no need to guess on processing power. The DSP chips being used in all the Fractal products is known. I believe the DSP used in Helix has also been discussed here before.

These aren't general purpose consumer processors, they are specialized Digital Signal Processor chips.
 
Geoff L said:
No your alright, you seem set on being a moaning minnie. I'll place you on ignore :)
Click to expand...
It is a legit question especially for someone coming over from Helix. Quite a few of us are interested in ways to work around it and the solutions suggested have been great. If you are not interested just stay out of it!
 
My name is mud said:
You have 2 solutions

Or you accept the limitation of the product and you simplified your way or thinking, while having an amp for clean, and another amp for dist (with whatever effects for them both)

Or you change amp at every riff and you will have a gap.

In the beginning I was like you, annoyed by that gap as the others modelers don’t have it. But they don’t sound as good as the axe.

I still have a mini gap when I switch to my lead scene, even with same amp block on both, even without changing channel between 2 scenes , just adding an eq and a delay and changing the reverb type

This and the gate are the only 2 defaults of the unit. One is a hardware limitation, and the other is a code

But I prefer to have a better sound with a mini gap. Maybe the axe 4 will solve that.
Click to expand...
I use scene controllers and have no gaps. Works great!
 
Randalljax said:
I am guessing this would work on FM9 as we have the 2 amp blocks and mixer , it’s a great idea
Do you guys notice it taking up a ton more CPU vs channels and the gap just curious
Click to expand...

I measured it, and the technique I demonstrated above to do gapless amp block channel switching doesn't add anything to the cpu load. That's what you would expect since the amp blocks run on a dedicated core and don't contribute to the overall cpu load.
 
unix-guy said:
But the key here is this: you don't know how the processing power inside the Fractal (or the Helix) is being used ;)

I'd be willing to bet the Amp modeling in the Fractal is significantly more CPU intensive than the Helix, but that's a guess.

However, there's no need to guess on processing power. The DSP chips being used in all the Fractal products is known. I believe the DSP used in Helix has also been discussed here before.

These aren't general purpose consumer processors, they are specialized Digital Signal Processor chips.
Click to expand...
Yeah, I tried researching the DSP chips used in Helix vs Fractal products to come up with some kind of objective comparison and came up with nothing useful.
 
juliatk said:
Yeah, I tried researching the DSP chips used in Helix vs Fractal products to come up with some kind of objective comparison and came up with nothing useful.
Click to expand...
Here are some relevant links with info on the Fractal DSPs:

https://forum.fractalaudio.com/thre...-amp-modeler-fx-processor.176122/post-2138126

https://forum.fractalaudio.com/threads/fm3-vs-quad-cortex.169722/post-2040335

https://forum.fractalaudio.com/threads/fm3-and-moores-law.163627/post-1961639

Searching for "DSP" posted by user "FractalAudio" provides quite a bit of info.

Here's some info about Helix:

https://forum.fractalaudio.com/thre...hout-replacement-it-seems.115085/post-1376109

https://forum.fractalaudio.com/thre...p-chip-dedicated-to-amp-s.103359/post-1237843
 
You must log in or register to reply here.
Back
Top Bottom