...14 years?, I only know the 3 & 1/2 years old project. please share a link if you can, I would like to compare both codes. Thx!

 

They've been at it since 2011. They said in an interview that they didn't complete their product the first time around because their "only investor ran out of money" (right before the finish line ;)).

Convolution on GPU in plugin form has been demonstrated before, including by LiquidSonics (Reverberate LE, Filtrate LE), but there was no demonstrable advantage over CPU processing.

 

It's right there in the original post, but okay:
https://gpuimpulsereverb.de/

 

Thx a lot, I'll start to get inside this, anyway it's not the same project, nor same binaries released by Guerra but I'm very interested to take a deep look in all of this from different development ways.

Edit; I didn't see the 2º post of @Radio !, yes, I've seen several other unsuccesful projects before. I'll try a dive.

 

Last edited: Mar 25, 2025

I'll try to compile the binaries released by the GPU Audio team for NAM, if I can load 10 NAM instances in a project without glitches/bottlenecking I'll finally come back to really play & I'll left some ugly noise over here

 

A modern alternative would be the "Waves Renaissance Reverb" , the one and only with very low cpu and very high quality
Renaissance Reverb - 35 $ www.waves.com/plugins/renaissance-reverb

 

Renaissance is algoritmic reverb, not impulse and doesn't run on GPU.. it's high quality though

 

Yes, I know, it's supposed to be an alternative to the old "GPU Impulse Reverb VST" that @avenocturno wants to try. That reverb is probably too old and won't be updated. If your PC doesn't have enough power, it's a good alternative to save resources.

 

As far as I'm aware, there have been several attempts at GPU-accelerated audio processing and they all went, basically, vaporware fairly quickly due to several reasons; GPUs offer fast *parallel* processing which doesn't provide any real advantages for audio (except convolution and now AI), it's fiddly to set up and the interest was low for serious investment. I see this idea coming back in regular intervals and then dying out again, it's become cyclical. GPU audio revolution!!!!----no one cares-----crickets. Rinse and repeat.

Even Acustica had a go with it, with Nebula CUDA. Didn't last long.

 

Last edited: Mar 25, 2025

@DAWinci, finally someone who says it , it's probably better to buy a faster CPU and make music instead of wasting your time.

Spoiler: Here are some potential disadvantages of GPU audio:

Here are some potential disadvantages of GPU audio:

1. Complexity: Implementing GPU audio can be more complex than using traditional CPU audio processing. Developers must deal with specialized APIs and shader programming.

2. Compatibility: Not all audio plugins or applications support GPU audio, which can limit flexibility. Compatibility issues between different hardware and software configurations can also arise.

3. Resource Consumption: Although GPUs are optimized for parallel processing, using GPU audio can consume more power or overload the GPU in certain scenarios, especially when many effects are applied simultaneously.

4. Latency: In some cases, processing audio on the GPU can result in higher latency, which can be problematic in real-time applications.

5. Cost: High-performance GPUs can be expensive, and not everyone has access to the necessary hardware to reap the benefits of GPU audio.

Source:AI

 

Last edited: Mar 25, 2025

Although I'm interested to dive in all of this GPU audio stuff, I'm not specifically interested about reverbs, but in the logic of this kind of development.

& I don't know why it should be as a religious discussion about to support or not the topic when a team states/shows (in their video) that their binaries are capable to load more than five NAM instances through a 4090RTX GPU, something that isn't possible (today) through powerfull CPUs, accord to many owners/users of those CPUs. I guess hat the only way to speak with authority enough would be to try & see for ourselves, ain't it?

If I can reach what GPU Audio's team states, then will have incredible discussions about the disguisting digital audio vs. the preciouss real amp-heads and monstruous vintage cabinets (which I'll never can afford anyway) & so on...

 

Everyone has their own interests. I wrote earlier that young computer science students who have finished their studies would like to start their own company and use what they have learned in a forward-looking and innovative way. So, if you have a business idea and enough money, I think it's great when new products are created to make people's lives more pleasant.

I can live quite well without the GPU invention; I'd rather have products that are stable and user-friendly. In my opinion, there's still a lot of catching up to do in that regard. I was very happy to have a language translator that really made life and communication easier.

I wish you every success in realizing your dream.

 

I don't get why so complicated to try the product they are announcing.
a VST3 download would be much easier.

I downloaded all the requisites they mention in the Github instructions but I couldn't do it. Instructions are not clear or uncomplete.
I got Visual Studio, Cmaker, CUDA drivers, etc no luck.
Seems they announce a plugin for coders not for musicians.

 

Absolutely agreed, after 15 days of confussion I left it in a frustration ocean. Surely is that u've said, the silence about it is too loud