Cherry Audio Forums

Hi

I don't know if any of you are aware of this.

https://www.gpu.audio/

Looks like a very powerful way to use your PC and MAC's graphics card to offload the CPU.

Will Cherry Audio be supporting this at all, there is or will be an SDK for it. This would really
make all CA's synths even more powerful.

Personally, I hope CA continues to support a wide level of users, not just those who can afford to upgrade technology every time something new comes along. By the way, I looked at this company and even joined their sever before I gained some insights into where they are coming from.

Steve W wrote: ↑Mon Jun 06, 2022 6:19 pm Personally, I hope CA continues to support a wide level of users, not just those who can afford to upgrade technology every time something new comes along. By the way, I looked at this company and even joined their sever before I gained some insights into where they are coming from.

What did you discover...?

Yes what did you discover, from what I can tell, most PC users have a decent-ish graphics card already so it is really just using what you already have.

What I took away from glancing at their web site is that modules have to be specially written to use it. They are, after all, selling just such modules. No?

Steve W wrote: ↑Mon Jun 06, 2022 6:19 pm Personally, I hope CA continues to support a wide level of users, not just those who can afford to upgrade technology every time something new comes along.

tarkus1971 wrote: ↑Mon Jun 06, 2022 6:45 pm . . . most PC users have a decent-ish graphics card already . . . .

And what about those of us that don't have "decent-ish" (whatever that means)? As I noted, I left the sever, so I cannot cite specifics, but they all seem rather gung ho about things that will ultimately feed a more-is-better purchasing behavior which often escalates and makes older technology less desirable if not obsolete.

As I have stated before, VCV used to work on my PC. Then they increased the requirements with the need for higher levels of graphics. Several soft synths makers have done likewise.

I have an investment in Voltage Modular and a number of Cherry Audio instruments and if they buy into GPU Audio's technology there is a strong possibility that the GPU Audio Technology requirements would be similar to what happened when VCV upped their graphics services requirement.

Again, I applaud VM/CA for not having bought into the graphics technology escalation and I hope they continue to support a wide level of users--including those that don't have "decent-ish" graphics cards that will most likely need to be updated.

Also, If I remember correctly, even at the level it was several weeks or so ago, there were system tweaks required on then-current systems to take advantage of the the power of their technology. To me that also says they will continue to push and push their requirements and tweaks will probably give way to the need for the greater-than-decent equipment.

Thank you for asking. I hope I have made the rationale for my previous comments clear.

UrbanCyborg wrote: ↑Tue Jun 07, 2022 2:51 am What I took away from glancing at their web site is that modules have to be specially written to use it. They are, after all, selling just such modules. No?

Your comment reminds me: their income model (to the extent they shared it) also caused me concern as a customer of VM, Cherry Audio, and the amazing assortment of talented and dedicated third-party VM module makers.

In theory GPUs designed for gaming might be an underutilized resource on a computer doing just audio work but actually exploiting this seems far from straightforward.

GPUs are designed to do massively parallel manipulations in 2D rendering (so things like blitting, shading, antialiasing, texture mapping, bump mapping) and in 3D rendering (things like the vector and matrix math involved in transforming polygon vertex data, lighting, ray tracing and such like).

Although some of the mathematical operations that GPUs are good at could be applied in audio DSP it's not going to be simple and each GPU would need custom drivers.

Modern GPU design is trending toward more general purpose parallel computing but as far as I know it's all still a bit experimental. So not a project I would like to manage or invest in.

In an ideal world there would be mass market audio processing units but this ain't going to happen any time soon. For now the good old fashioned CPU is probably your best bet.

Sines: I'm fairly sure Sines is now offloading processing to the GPU.

I'm using a Laptop - 16 cores at 3.6khz with a separate Nvidia GPU

I load Sines
The CPU usage is minimal

The GPU however - is instantly through the roof - 4 voices - and the GPU is maxing out - and then the audio is breaking up horrendously

It looks like the GX80 is doing the same

The result of this - is that the extra gpu load heats up the Laptop - and the i9900 then starts to slow down because it is over-heating
and I can see the cores getting unnecessarily hot

This wouldn't be so bad on a workstation - as the graphics can be cooled properly

Terrible idea on a Laptop

These are the only 2 plugs that do this
And previously - I could easily get 16 voices from the same Laptop

Having unloaded the CA plugs - and waiting for the system to cool down
Task manager is now showing the GPU has returned to between 1% and 7% max

This is reproducible
I'll capture some video later.

Also - having just tested again - this is in stand alone mode only where the gpu overloads

Here’s Bad Hair Detective’s hilarious take on GPU Audio.

A word of warning if you haven’t encountered BHD a before. It’s extreme satire in the mould of Monty Python, Chris Morris, League of Gentlemen etc. Not for everyone.

https://youtu.be/abPvDdUibfI?si=fhjAEwWAmDwzCeOw