Axe 3, you say? TigerSHARCs are phased out without replacement, it seems

Status
Not open for further replies.
Make a design with four of the new chips?

As far as I understand, that would increase latency.

I don't know if there are DSP alternatives to TigerSHARCs. It's not necessary to use DSPs, of course, I hear that modern FPGAs are fast as well. No idea if they're fast enough for this application.

In either case, this means that Axe FX II will be phased out as well as soon as whatever stock of TigerSHARCs left runs out. This could take a while of course.
 
So that being said, were OK????
I'm not sure which ones Fractal uses.

For a while, we're fine, yes. Less than one year to purchase all the future stock isn't much time though given that you need to spend like $600 for a pair, so were talking about millions of dollars, I guess. And whatever model comes to replace current Axe will need to have a different architecture.

The ADSP-SC58x & ADSP-2158x SHARC Processors will provide suitable replacement products for many applications using the ADSP-TS20xS products today

The TS201S which, I think, what is used in Axe II, is like 2.5 times as fast as any of these.
 
Cliff et. al will rewrite their entire code base to work nicely with the new processors, and they must have already done a lot of that to make the AX8 and FX8 possible.

Well, AX8, FX8 and, to some extent, Helix show that there's no replacement for processing power, you still need it no matter how much you optimize code. A lot of what's happening is just math operations.

Regarding latency, I'm not sure that it will go up, because it's the same amount of computation, but spread out over a different combination of resources.

Cliff said it would increase latency. I personally don't know why this happens.

Anyway, I wouldn't worry

I'm not worrying, I'm curious about what this means. In a way it's a good thing because even if Fractal had a vicious plan to rest on the laurels now they can't do it and simply MUST bring that Axe 3. :)
 
But there are other factors at play that might not make it so bad, and this is the new hardware accelerators that are on the new chips, and I assume that Cliff et. al are rewriting their stuff to use them

Those hardware accelerators don't seem to help much though as the devices that use them are WAY less functional.
 
I thought that the topic was that consumer electronics will be built with cheap DSP chips, like it or not

I meant not this topic but the topic where Cliff replied that adding DSPs (even fast ones) will increase latency so that's not the way to go.

Consumer electronics has always been built with cheap chips, but we aren't talking about Zoom or Boss here, right? Or even the feature limited Fractal products such as AX8 or FX8. Currently, limiting processing power leads to simple results - limited features or worse sound. And I fail to see how that's a good thing. Outside of consumer electronics where nobody cares about how things sound, if course.
 
Well you know Metric Halo makes their interfaces "future proof", meaning that they will allow you to swap out boards. They're completely redesigning their boards and front/back panels with new whatever chips that replace the Sharcs. Something like 5 times as powerful and as much in more memory. You have to pay for the boards but you can either replace them yourself, with excellent instructions, or you can send them back for a fee.

An idea?
 
Well, this is certainly interesting. I know the TigerSHARC has been part of the AxeFX's DNA from day one. I'm sure Cliff and crew are prepared for this eventuality. I look forward to seeing what they have in store.

I'm just a web programmer, which is another real compared to what Fractal does, so I can't say how easy it would be to use another company's processor in the AxeFX. I do know the Helix is powered by a Freescale DSP which is supposed to be a beast, so I suppose there are alternatives.
 
I can't say how easy it would be to use another company's processor in the AxeFX. I do know the Helix is powered by a Freescale DSP which is supposed to be a beast, so I suppose there are alternatives.

I'm no expert in embedded systems and have very superficial knowledge of these things myself, but my understanding (which may be totally wrong) is this.

1. TigerSHARC was the most powerful DSP available. Maybe something from Texas Instruments was also as powerful, I'm not sure, but the DSPs used in other products, including Helix, are weaker.
2. Axe FX relies on a lot of horsepower both for modeling and effects. The amp models are very complex, all those knobs cost something in terms of CPU cycles, and effects are very CPU consuming as well. You can hear the difference between Axe and Helix reverbs even on YouTube demos if you pay attention.
3. You can't just fit Axe FX into something considerably weaker. Maybe some things can be optimized but a lot of what's going on inside is pure math, it needs power. So making an Axe with that Freescale DSP is probably not an option.
4. The whole super high power DSP thing seems to be dead now, nobody is developing anything. Sales were rather poor, these chips haven't been updated for ages, production lines no longer exist or are being dismantled.
5. For some reason increasing the number of DSPs is problematic as it increases latency. Maybe there are solutions, I'm not sure.
6. So it may not be a matter of switching to another company's alternative chip, but changing the architecture quite dramatically. It seems that users of high power DSPs are switching to these new FPGAs, which are totally different beasts as far as programming is concerned.
7. Axe FX was a weird device in that the cost of the DSPs was outrageous to begin with. Nobody else does it this way, AFAIK. I think those FPGAs aren't cheap(er). And development costs are higher.

So, from my layman's perspective Cliff and Co. have one hell of a choice to make. And it will be interesting to see what comes out of it for us users.
 
maybe FAS can team up with UAD and buy the schematic for the chip....and make it themselves........

I'm afraid the costs would be prohibitive. It appears that even AD cannot maintain production, and they sell to a much wider customer base than just audio. Those DSPs are primarily made for industrial and military purposes, I think.
 
As long as I don't read "hey we developped an Ipad app that does the same as the axefx" :D

Oh you will read it. :) I'm sure there will be a lot of newcomers who will do iPad apps and cheap modelers. Bias FX is one. There already are some. You have Amplifire. I think folks like Two Notes are considering it as well. The old guys like Boss or Zoom or what not won't disappear either. And there will be iPad apps. :)

Who knows what Fractal will do. But unless Cliff has a hundred million dollars to spare I'd bet Fractal will try to remain in the high end. Mass production of cheap stuff is a different business.
 
Yea, I'm sure Cliff and the team are looking at their options and will develop something that doesn't compromise quality.
 
It may be wishful thinking on my part, but I wouldn't be surprised if Fractal had stockpiled enough processors for a few years of production.
 
Status
Not open for further replies.
Back
Top Bottom