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

Status
Not open for further replies.
That TI chip seems to have been available for 5 years or so, is a fraction of TigerSHARC's cost, yet it hasn't set the world on fire for some reason. Wonder why.

I wonder about this, too. I kinda skimmed over the white paper, and it seems like it may be a viable replacement. I'm sure it'll be a major PITA to adapt the existing code to the new platform.
 
@FractalAudio, are there any DSPs out there where you can easily reuse your fundamental architecture? Porting assembly code is tricky enough without having to rethink the whole approach. Any thoughts of using C/C++ for some parts next go around?
 
Porting assembly code is tricky enough without having to rethink the whole approach

Guess those hardware/firmware/DSP engineer openings in late winter/spring were there for a reason. :)

It's still a DSP though, so if these are related, we have one less uncertainty to figure out. Or not. :)
 
Porting assembly code is tricky enough without having to rethink the whole approach. Any thoughts of using C/C++ for some parts next go around?

Do we know what language it's written in? I would assume it's already in C/C++. I mean, I love assembler, but nobody has written their whole system in assembler since the 80's.
 
^ for one, the core of KolibriOS (kernel and drivers) is written entirely in x86 assembler [FASM].

IIRC, CC has previously written that the code is 'C' in a C++/OOP style... presumably with intrinsics and inline assembly.
 
Well people didn't think that axe2 algo's could be ran on anything less, yet with axe8 it is, and with nobody complaining about it or even mentioning it for that matter.
 
Well people didn't think that axe2 algo's could be ran on anything less, yet with axe8 it is, and with nobody complaining about it or even mentioning it for that matter.

Well it's not like Axe2 is running on something less in the case of AX8. More like half of Axe2. AX8 is very limited compared to its elder sibling. AX8 users may be happy with it for what it is, but I wouldn't be.
 
Also, a convolution reverb computes a weighted sum on a each cycle. Plenty of parallelism there -- if you can do a lot of multiply-adds at once, you don't need a fast clock.
Parallel processing is one of the areas where an FPGA really shines. That and a big reduction in 'glue' components. They have a learning curve though.
 
Well people didn't think that axe2 algo's could be ran on anything less, yet with axe8 it is, and with nobody complaining about it or even mentioning it for that matter.
You haven't been around that neck of the woods much, have you? AX8 owners are building massive boards adding Big Sky, H9 and other assorted strymon and eventide devices, along with drive pedals because the reverbs and drives are the biggest power hogs in the AX8.

(And I'll concede, also a bit because Strymon's have yummy presets. But if the AX8 could easily do them, I think they wouldn't add them to those massive boards.)
 
Last edited:
You haven't been around that neck of the woods much, have you? AX8 owners are building massive boards adding Big Sky, H9 and other assorted strymon and eventide devices, along with drive pedals because the reverbs and drives are the biggest power hogs in the AX8.

(And I'll conceide, also a bit because Strymon's have yummy presets. But if the AX8 could easily do them, I think they wouldn't add them to those massive boards.)

I have not but what I was stating was when the 2 came about it was said it needed all this power on one cpu just for the modeling, but maybe that was 2 amps at once, but still how did they get the algo's to work on the lesser cpu without sacrificing something?
 
I have not but what I was stating was when the 2 came about it was said it needed all this power on one cpu just for the modeling, but maybe that was 2 amps at once, but still how did they get the algo's to work on the lesser cpu without sacrificing something?
Just one amp on one cpu and the cab block on something called a hardware accelerator or something like that...?!? I don't know how that works. My limited knowledge of DSP processing is satisfied with not knowing more.

I guess it works. Sort of. You want to get into the why part? Good luck, then.
 
I have not but what I was stating was when the 2 came about it was said it needed all this power on one cpu just for the modeling, but maybe that was 2 amps at once, but still how did they get the algo's to work on the lesser cpu without sacrificing something?
They sacrificed TWO AMP blocks running at the same time. The AX8 uses approximately half the DSP power for AMP modeling.

For those panicked about supplies: Fractal doesn't buy directly from Analog Devices*. Very few companies do, really. Unless you're dealing in HUGE volumes, you buy from a VAR and VARs carry stock. Usually a fair bit of it. So while AD may be winding down production, stock supplies aren't going to deplete very fast, especially at the VAR warehouses. By the time they do wind down, we'll be on to whatever hardware revision Cliff dreams up next.

* Note: I don't know specifically here. I'm making a highly educated guess based on what I think Fractal's output happens to be and how semi-conductor sales work in my experience in the FPGA industry. This is outside speculation on my part.
 
Last edited:
I have a II so I don't care what cpu they use for anything really lol.. shut up and play yer guitar.... but better check with Ahmet on that one, probably trademarked.
 
I suppose it's slightly disconcerting since we don't know how this will eventually play out. I have to imagine that the fairly large user base would encourage a chip maker somewhere to produce a new line that they could market to all the various modeling and other audio application companies. Maybe I'm overestimating the numbers compared to the scale of demand for chips in general. (probably). I hope my Axe doesn't die anytime soon, although I have given serious thought to having an analog rig again, as just another (simpler) option and now it might also serve as a backup. My sincere hope is this discontinuation will spur development of a new technology or generation of audio chips and we'll get an AxeFx3. Which I will promptly purchase and never learn to fully utilize, just like the II.
 
Status
Not open for further replies.
Back
Top Bottom