• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

New low-cost DSP platform in development

AVX-512 can handle 8 64-bit floats in one instruction. And then we have AVX, AVX2 and SSE2. All can handle 64-bit floats.
Ah OK, looks I'm not up to date anymore or more likely, I simply mixed up things :-(
 
This is not a solution, nobody actually wants to use a VM just to run some junk proprietary software, and if you had an iPad, you'd still be out of luck

Nope, since you can still only use the effects SigmaStudio gives to you, but my DSP platform will let you write and run anything
But you said no VST, so 99% of "plug and play" algorithms aren't usable on your platform
 
An iPad will run a Windows on Arm VM just fine though ;)
It would need to be a Windows on x86(_64) VM though. SigmaStudio won't run without the programmer driver being loaded, even though it's not necessary if using a networked programmer. And unless something's changed it doesn't provide an ARM version of the driver. If they removed the driver requirement it should run under WINE.
 
Progress update: Runtime code execution is now working!

At least, it is on x86 because I'm too lazy to set up one of my RISC-V machines for development right now, so this is with some manual machine code magic. You wouldn't believe just how long it took to get to this point.
Screenshot from 2024-11-17 00-16-13.png

Next steps: Actually try it on RISC-V and get more language features working.
 
I should have asked this earlier, but what inputs/outputs would be good to have?
 
I should have asked this earlier, but what inputs/outputs would be good to have?

It's really up to your target market. I would say: Toslink, SPDIF, AES/EBU, USB are all mandatory. Optional: HDMI and analog inputs (RCA and XLR).

Then you need to decide if you want your customers to be able to take measurements through this device. This makes it MUCH more complex. Not only do you need an XLR input with 48V Phantom Power and an ADC, but you also need to write an ASIO driver for Windows.
 
I should have asked this earlier, but what inputs/outputs would be good to have?
Is it easy to add boards with your preferred connectivity?

For me it's a twist on the straight wire with gain idea, it's a spdif wire with DSP. USB hosts should be able to do their own DSP, so I'm not sure about needing to support USB.
 
Is it easy to add boards with your preferred connectivity?
For me it's a twist on the straight wire with gain idea, it's a spdif wire with DSP. USB hosts should be able to do their own DSP, so I'm not sure about needing to support USB.
A pure spdif thing would cost almost nothing ($1 for the spdif chip, $1 for the optical ports, a couple dollars for capacitors/resistors) so that's an obvious choice. As far as other boards go, they probably wouldn't be too much more but they would require new case designs.
 
Last edited:
A pure

A pure spdif thing would cost almost nothing ($1 for the spdif chip, $1 for the optical ports, a couple dollars for capacitors/resistors) so that's an obvious choice. As far as other boards go, they probably wouldn't be too much more but they would require new case designs.
I didn't realise it would have a case, I was thinking it was just a board, and the user would find a case, that's even better.
 
I should have asked this earlier, but what inputs/outputs would be good to have?
Multichannel PCM input through HDMI would be awesome, and toslink and USB-C. For output eight RCA would be nice. Maybe even do it modular so people can pick and choose what and how many inputs and outputs they'd have. But now I guess it starts getting a bit less low-cost ;)
 
May I suggest a major shortcut?

Why not contribute to CamilaDSP, and make it run on RISC-V? A port should be relatively simple, and if you want more DSP flexibility, you could add more and more flexible building blocks. Web interfaces are also already available and most of the heavy lifting is already done. Major thing to port would probably be the super fast (I)FFT library for Rust.
 
May I suggest a major shortcut?

Why not contribute to CamilaDSP, and make it run on RISC-V? A port should be relatively simple, and if you want more DSP flexibility, you could add more and more flexible building blocks. Web interfaces are also already available and most of the heavy lifting is already done. Major thing to port would probably be the super fast (I)FFT library for Rust.
Unfortunately, this is off the table for now because CamillaDSP is just for audio. I have bigger plans than audio because DSP is used in way more things than that. However, what wouldn't be off the table is adding compat for a custom DSP effect within CamillaDSP, which shouldn't be too bad to implement.
 
Multichannel PCM input through HDMI would be awesome, and toslink and USB-C. For output eight RCA would be nice. Maybe even do it modular so people can pick and choose what and how many inputs and outputs they'd have. But now I guess it starts getting a bit less low-cost ;)
Looks like HDMI audio is like SPDIF, so that's just a couple dollars for a chip capable of converting it. Then you can get a single 8 channel DAC from ESS or 4x 2 channel ones, which is just a few more dollars. The supporting components are just a couple dollars more. So, it looks like such a setup might not be so expensive after all!
 
Looks like HDMI audio is like SPDIF, so that's just a couple dollars for a chip capable of converting it. Then you can get a single 8 channel DAC from ESS or 4x 2 channel ones, which is just a few more dollars. The supporting components are just a couple dollars more. So, it looks like such a setup might not be so expensive after all!
Whet is like SPDIF is ARC. Multichannel LPCM over hdmi is not that straightforward to diy unfortunately.
 
Last edited:
Whet is like SPDIF is ARC. Multichannel LPCM over hdmi is not that straightforward to diy unfortunately.
Looks like a standard SPDIF chip (like the DIR9001) can take in a multichannel input and pass it through, so we can just do that multichannel processing in software.
 
There's also the discovery and control over CEC which I gather is the cause of many compatibility problems.
 
There's also the discovery and control over CEC which I gather is the cause of many compatibility problems.
Is CEC itself or the lack of it is the cause of these compatibility problems? I found an implementation of it for Arduino, so doing the same on the Duo board shouldn't be too hard/expensive. https://github.com/floe/CEC
 
Looks like a standard SPDIF chip (like the DIR9001) can take in a multichannel input and pass it through, so we can just do that multichannel processing in software.
Audio over HDMI - what includes uncompressed multichannel LPCM - is transmitted embedded with the video signal via various TDMS differential pairs, there is no way you can interface those with a standard SPDIF chip.
Again, what you might get is ARC, that gives you decent 2 channel spdif audio or compressed multichannel, that can be good enough, but that any cheapo extractor also gives you. But for uncompressed multichannel audio, your only options are eARC or regular audio over HDMI, both of them far too complex, or are blocked by license fees, or require dedicated hardware, or in the worst case all of them together.
But maybe there are other ways, I honestly have no clue. Good luck!
 
Last edited:
SPDIF chips will, of course, pass through compressed DTS or DD content. Your receiving end will need to support this, though, which may not be as easy, because normally a PC-like thing is used to output that bitstream, not input it. And if it works, you'll need to hook up the necessary codecs to make it into actual multichannel audio.
 
Back
Top Bottom