Arm based Macs (platform support)

[pdxindy]

Do devs actually have to buy new Apple hardware or is there some kind of emulator/virtual machine available?

theres a dev kit. they announced it at the WWDC. mac mini already on ARM chips. devs have to buy it but i have no idea how much it is. seems like it'd be cheap.

$500

[Urs]

Not only that, but there's also an application process - starter kits are in limited supply.

Don't expect much information from devs who got it though. Apple prototypes are subject to air tight non-disclosure agreements.

[Urs]

(good thing we've been developing our first hardware product on ARM architecture though)

[EvilDragon]

look to NI for long port times.

NI is looking at the ARM ports very seriously.

[lunardigs]

Hopefully this will be a boon to Raspberry Pi like devices.

In a similar move, Apple's push for Class Compliance benefited Linux, big time.
Also, the move away from Flash!

[machinesworking]

look to NI for long port times.

NI is looking at the ARM ports very seriously.

I'm still waiting on resizable GUIs and VST3. I've been with NI long enough to know it takes them some time, but they get it right when it comes. For instance, VST3, IMO a lot of really bad ports from people, only about half of the VST3 plug ins I own are as stable as the VST2 version.

[EvilDragon]

Resizeable GUIs are a huge problem because all old products are tied to NI's homebrew graphical toolkit from 20 years ago that didn't keep up with the times. More recent products are all done with Qt so they don't have such issues. This is gonna take a lot longer than ARM or VST3 support (both of which are on the table).

[sascha]

Hopefully this will be a boon to Raspberry Pi like devices.

Hehe, the new Reverb for Uhbik 2 was initially designed on a Pi3 running Raspbian Jessie, although it was just the naked dsp and doesn't mean a thing for the product line

[lunardigs]

Hopefully this will be a boon to Raspberry Pi like devices.

Hehe, the new Reverb for Uhbik 2 was initially designed on a Pi3 running Raspbian Jessie, although it was just the naked dsp and doesn't mean a thing for the product line

Hmm, that's interesting. Why did you design it on a Pi3?

Pi3s are great, btw. I've actually made gifts out of them: Kodi, plus a wireless mini-keyboard.

Speaking of DSP & ARM though, I've had a MOD DUO since 2017. It's got an A20 (dual core ARM A7 @ 1.0GHz), plus 1GB RAM.
It can produce really nice reverb tails! You can build just about any combo of delay, reverb and phase modulation you like and it might approach ~75% cpu, including an amp sim.
Yet, it's definitely under-powered when it comes to demanding stuff. For instance, the Calf Multiband Limiter ( https://calf-studio-gear.org/doc/Multib ... miter.html ) makes it puke. You might be able to use it alone, but good luck. There's also a JCM800 amp sim ( https://pedalboards.moddevices.com/plug ... gwMHByZV8= ) which sounds great, but slams it.

Of course, the A7 is pretty dated at this point. I believe their next gen stuff will use the iMX8, in quad core (A53 @ 1.3GHz) ... Btw, MOD is in Berlin. Maybe you could become DSP chums? They've ported hundreds plugins to ARM so far ( https://pedalboards.moddevices.com/plugins ); a strategic alliance might be possible. They also know hardware, which is something U-he has gotten into.
Also, I'd love to see a U-he offering in their plugin store, like Runciter perhaps.

[sascha]

It was just out of curiosity, I needed a good & natural sounding reverb for a home project (building an edrum module), but eventually used it as a breadboard. Reverb design is special, and can be mentally taxing. It's maybe 20% coding and the rest is endless listening and tweaking, or maybe trashing an idea and start over again. And many nights to sleep over it. When we decided we wanted a new mode in Uhbik-A, it was already quite far, and the code was easy to transplant.

[Urs]

Likewise, I had been curious about running our latest filter designs on an STM32F4. CVilization could do it, but it's reaching limits. But then, a voice of Diva easily takes more than 10% of a 1,8GHz CPU, so I wouldn't expect it to run a single voice on a 180MHz one.

[EvilDragon]

Of course, x86 instruction set is not the same thing as ARM instruction set, it's CISC vs RISC, so Hz for Hz it's not nearly the same thing. But yeah it's still 10x difference in frequency...

[blakflag]

This was the first thing I thought of when I heard the Arm news. Once VST ports begin to be finished, then all we'd need is a generic linux VST host for Arm, and some clever person will do that I am sure. Then we can begin to run embedded VSTs as super deluxe standalone hardware FX. I'd kill to run Presswerk or Bazille in a eurorack slot, even if only a few parameters had hardware mapped knobs. (with hopefully some way to see the whole GUI on demand, like a tablet connection or something)

[Urs]

I'd kill to run Presswerk or Bazille in a eurorack slot

You'd need an ARM A8+. If you look at modules that do this, they drain like 700+mA. That's like a third of the power of a decent case (to be fair, the Percussa module also takes as much space).

The only good option power/mA currently is the Cortex M range, such as the F4 or even the H7. But they don't support the NEON instruction set. They will however support the upcoming "HELIUM" in a few years, which will not be bianry compatible to NEON thus requires extra development time.

[blakflag]

Thanks Urs, it sounds like it's still on the bleeding edge of feasibility.. but this possible day is getting closer though as technology marches on. I think having a full synth of U-He quality would be worth 700mA to many people, especially if it was an open platform that you could load with whatever software you wanted. If I was a hardware engineer I'd get right on that.

edit: it also occurs to me that hardware platforms evolve. I've been in the PC world for 30 years. If anyone told me in 1990 that a *video card* could eat 800+ watts of power.. I'd laugh and ask what they were smoking. It might be that what we think of as a eurorack today will not bear much resemblence to what we have in 10 years.