FabFilter

Sounds exciting.

I can imagine having a rack like Waves StudioRack, or BlueCat's PatchWork, or even Cantabile shifting the workload to graphics processing and returning the results to the host.

It wouldn't be hard to imagine a rack bridge or even new release of VST3G versions of existing plugins. Core processing could also benefit...

Unless proprietary battles derail the possibilities.

Don't get your hopes too high up, as GPU accelerated computation only shines on the stuff where there are multiple computations go at once in parallel, such as reverbs, delays etc., as they have a lot of reflections that do not strictly depend on one another.

EQ is a serial computation, Compressors are serial, thus a regular channel rack wouldn't really benefit from GPUs multithreaded architecture.

Would be nice to offload some of the spatial stuff from the CPUs though.

www.researchgate.net/publication/245586866_Audio_Signal_Processing_Using_Graphics_Processing_Units

There is plenty of research and work to show all the plugins can be fully optimized and more capable on the GPU. What that will entail is investing in more resources for Metal API on macOS, Nvidia based CUDA and HIP based AMD coding, to Intel's OneAPI.

From an economics standpoint most vendors don't want to invest. Then again if I just used RPN based EQ/Comp and found hardware based Reverbs/Delays that didn't cost $7k or what have you I'd be down to no software plugins.

On the Software plugins front all Zen 5 and beyond AMD CPUs are APUs [integrated SoC like Apple -- AMD actually was first here, even before Intel] with Xilinx [AMD subsidiary] Neural Engines. These will be announced starting 1Q 2024 but 4Q 2023 for EPYC based hardware.

Intel is moving this way as well. Within the next 18 months all major vendors for Windows/macOS will be SoC/unified memory backplane/Neural Engines [ML FPGAs] all on various 3D substrates.

This will probably be when Plugin vendors become interested as they can not worry about OEM external GPU issues, offloading and latency issues with roundtrips between the CPU/GPU never mind the Neural Engines. All sharing the same memory pool that directly interfaces across PCI will all the likes of Thunderbolt 4/5 Audio interfaces and more to reduce latency, across the board.

Still, nothing beats out Analog gear as it isn't sampled and thus zero latency, but much better than the present state. By Zen 6 and Intel similar offerings most people will forget about buying external GPUs unless they are content providers working on products like Maya, Blender, etc.

"QUOTE"
Current graphics processing units (GPUs) are massively parallel computing environments offering remarkable performance boosts in parallelizable tasks. Audio signal processing is a potential application area. Three different cases for GPU implementation are studied: additive synthesis, fast Fourier transforms (FFT), and time-domain convolution. For additive synthesis nearly two million sine waves were computable in real time on the GPU, giving a factor of 250-3000 performance gain over CPU implementation. Similarly, the GPU was able to perform an FFT eight times longer than the CPU version in the same time. For a stereo signal the GPU was able to compute a two-million-point FFT and inverse FFT in real time with an input buffer size of 1024 samples at a sampling rate of 48 kHz with 50% overlap. Finally the GPU could compute approximately 130 times longer FIR filters than the CPU in the same time. For stereo input and output, requiring four filters altogether, the GPU processing was able to implement FIR filters of length 376 000 taps in real time. The latency in all these tasks is tolerable, since the performance is nearly optimal with a hundred-sample buffer, which corresponds to a few milliseconds. In summary the results show that GPUs are highly useful for computationally intensive audio signal processing tasks.

we're on to you, GPU audio marketing team