I've seen ppl building Nebula servers ONLY to mix with Nebula instances on each track, eating CPU AND GPU for breakfast.
But how many ppl ?
Probably less than 10 on the whole planet ...

As i said earlier, CONVENIENCE over sound or power.
Most ppl mix/compose ... on 1 computer.

And if they need more power, they don't use "external" computer or GPU or whatever ... they simply buy another computer.
Most of the time, they will freeze ... or buy less CPU intensive plugins.

I don't click on the "GPU for audio" future ... i already done it for YEARS and seen nothing coming.
Some coders try, and they all leave fast.

 

This is what a lazy-ass like yours truly wanted to say but didn't.

 

Although I agree about most that you've wrote, is not just for guitar &/or bass amps demand, take in count Neutron Speakers plus a countless number of other kind of hardware profiles (eg.; tube compressors, consoles, tape machines & so on). All of this stuff could let us arrive to a moment in wich the sample & modeling worlds turns into an easyly affordable unique instance.

 

2º time you say this, & I agree about the happenned fails, but I wonder why you firmly believe that nothing changes along the years given the proofs of the contrary since the beggining of digital audio 'till now. Tell me.

 

I have been patiently waiting for something like this. We have discussed gpu dsp for plugins before here. And there have been enough legit attempts on the subject. Like Nebula & Nvidia GeForce. But all these attempts were not fully utilizing the gpu as the main dsp source. Nebula's GPU's DSP main prob i found in my tests was high latency with all the caveats this brings.
This one though showing Mir3 from VSL is quite substantial, to say the least. I am familiar with VSL's convo reverb, i have worked with the previous version in another studio and to my ears it is the absolute best convo reverb but also the most absurdly expensive. The newer Mir3 Pro 3D unlimited version and the whole Rooms bundle will set you back ~2000 euros! You will also need Vienna Power House which is the actual utility that outsources processing to the GPU, another 165.
So yeah this one's for the pros that can afford it. For Windows, VSL lists minimum requirements for Power House as Nvidia or AMD with 8gb ram.
Recommended: Nvidia RTX3060 12 GB VRAM, AMD Rx6700XT 12 GB VRAM.
For Mac, any Apple Silicon Mac with 16gb ram minimum, 32 gb recommended.
This opens a lot of doors for further advancement. To me, GPU audio future seems good already. I am sure as we speak that some of my friends will jump on this right away. And boy do i want Mir Pro 3D hehe. Surely not for hobbyists though.
Cheers
PS: I am very excited about this, as VSL make excellent professional products with great support and i am certain they just wouldn't stick their rep on dodgy technology. Also waiting to see what Analog Modelling will do with the same tech.

 

Last edited: Feb 17, 2025

@avenocturno
I didn't said "nothing change".
Of course, things more FAST !

I said "GPU audio" NEVER worked (i mean ... in the long run).
Read the thread i posted above ;D

I just don"t see any difference between THIS attempt and others

 

Would be great, I hope all of that keeps on the Atkinson's path, or at least more or less affordable

 

Last edited: Feb 18, 2025

Thx a lot for this, I'm so busy that couldn't know!

 

Has anyone compiled the GPU-accelerated VST3 of model and impulse loader?
If so, could you kindly share it with us codetarted people?

 

hahahaahaaa, I'm codetarded too 4 sure, but I'll start to study about it as soon as I can. Not really sure about nothing related to your question..., but I guess that the compilation task should be strictly related to the hardware that each of us has, I mean, I don't know if a generic pc installer (universal hardware compatible) is possible. Apple is much more simple about that, u don't have hundreds of Apple hardware builds over there... but, I repeat, I'm absolutely ignorant in this new (for me) field.

 

Last edited: Mar 23, 2025

gpuRIR is a free and open-source Python library for Room Impulse Response (RIR) simulation using the Image Source Method (ISM) with GPU acceleration. It can compute the RIRs between several source and receivers positions in parallel using CUDA GPUs. It is approximately 100 times faster than CPU implementations

https://github.com/DavidDiazGuerra/gpuRIR

 

But that´s not VST3 plugin like in the video shown

 

Nils Schneider - GPU Impulse Reverb ---> www.kvraudio.com/product/gpu-impulse-reverb-by-nils-schneider

GPU Impulse Reverb is a convolution reverb plug-in that does the full reverb computation on a NVidia Geforce / ATI Radeon HD graphics adapter or any other OpenCL compatible device and has a CPU consumption of nearly 0%. The plug-in requires an OpenCL capable device and recent drivers.

 

Actually the guys at Analogxai are exploring incorporating this technology into their Genesis plugin that does analog gear emulations.

 

It says open source here .. https://www.neuralampmodeler.com/the-code

 

So, if I try to do this example here -> gpuaudio-sdk/installation/main.md at main · gpuaudio/gpuaudio-sdk · GitHub
Will it compile GPU-NAM VST3 for me..
I´m asking this because I have no clue what I´m supposed to do and I don´t wanna waste time my time

 

Hello@Legotron, i ask the AI:

If you follow the instructions provided in thegpuaudio-sdk/installation/main.mdfile on GitHub, it should guide you through the process of compiling the GPU-NAM VST3 plugin. However, the success of the compilation depends on a few factors:

1. Prerequisites: Make sure you have all the necessary tools and dependencies installed as mentioned in the documentation. This often includes a compatible version of a C++ compiler, CMake, and any specific libraries required by the SDK.

2. Follow Instructions Carefully: Pay close attention to each step in the installation guide. It usually includes commands to run in your terminal, configuration settings, and any environment variables you may need to set.

3. Check for Errors: If you encounter any errors during the compilation process, the documentation may provide troubleshooting tips, or you can check the issues section of the GitHub repository for similar problems faced by others.

4. Community Support: If you're still unsure or run into issues, consider reaching out to the community or the maintainers of the project. They can provide guidance and help you troubleshoot any problems.

If you followthe steps carefully and ensure you have the right environment set up, you should be able to compile the GPU-NAM VST3 successfully. Good luck!

 

Last edited: Mar 24, 2025

Maybe just a little late to the party tho,
Released some 14 years ago, 2011