GPU Driven | The Future Of DAW Processing Or A Dream ?

[Ah_Dziz]

So 1 ms latency at 96k. That's 2 ms at 48k. I wouldn't find that acceptable

Sounds OK to me - Apparently that's the same time it takes from a drummer hitting their snare to the sound hitting their eardrums

Not OK if you want to monitor with FX while recording live.

I’ve found that up to 256 samples is damn near unnoticeable for most people through their headphones at 48k. I’ve tested lots of different setups (just to see how far it can be pushed before anyone says anything or I notice a change in the performance) through the years with folks in and out of my studio. I can’t deal with much more than 128 for playing any keys that need accuracy myself, but i know it’s there. If you don’t tell people they don’t tend to notice much. I would be amazed if a 2 ms difference would be anything to anyone used to digital equipment if you didn’t mention it.

It's not enough when multitracking live vocals or guitars, for example. You need the tightest possible latency then.

I would imagine it could be a problem in certain scenarios. A full band laying down basic tracks may get thrown off. It just seems lower than the threshold where anyone I’ve worked with has ever mentioned the latency.
Anyway, my experience with messing with graphics card DSP stuff (including the first uad cards) was always that it would add two buffers to the signal, so that if you had a 32 sample buffer for instance it would jump to 96 samples when monitoring through the software and the extra DSP.

I did think that this would have picked up more momentum by now, but I suppose that lots of coders just don’t want to bother.

[generaldiomedes]

For the folks in these forums that are NOT multitracking live guitars or vocals (quite a few I imagine) 2 ms is nothing .. I would trade that for increased processing power any day. In fact, until I upgraded my laptop recently lack of processing power was the primary source of latency while tracking MIDI.

[v1o]

Liquidsonics made a couple free ones using CUDA, but it doesn't look like that made its way into any of their commercial products:
https://www.liquidsonics.com/software/reverberate-le/
https://www.liquidsonics.com/software/filtrate-le/

I was just about to mention them when I saw your post. I've been following this topic for about 6 years now, and other than their free stuff, LiquidSonics doesn't seem to have used CUDA anywhere else. (And neither has anyone else, from what I've seen.) Reverberate LE has a built-in latency of at least 512 samples, so a few more milliseconds of latency isn't really going to matter much.

According to their website, the zero latency CPU edition can be delayed up to 8192 samples, but otherwise offers "zero" latency. But the NVIDIA CUDA GPU edition can't do less than 512 samples of delay (up to 8192:

"Zero latency CPU edition (or up to 8192 samples for lower CPU usage)
Low latency GPU edition (512 – 8192 samples delay)"

I guess they determined that implementing CUDA processing in their commercial products wasn't worth the performance hit. They didn't even upgrade the CUDA-based, "LE" versions of their plug-ins to 64-bit, which is another sign that they don't think there's a future in CUDA processing for plug-ins.

It's an interesting concept that keeps raring its head on forums every so often. But from everything I've read about it, it will never be useful for plug-in processing or anything that's time-sensitive, apparently. While some may moan about it, I'm just glad I don't have to worry about getting a powerful graphics card now, just to run more plug-ins.

Steve

The latency can’t be any worse than using an external DSP card surely. GPUs should be used as a completely optional extra for offloading processing when it’s too taxing for the CPU. Maybe have a toggle switch to shift plugins from CPU to GPU processing.

[EvilDragon]

Yes it can be worse because GPU doesn't process things in the same way DSPs do.

[PeterP_swe]

I don't care about latency when mixing or experimenting with sound design. I'm happy to increase my buffers to 2048 then and wouldn't mind using GPU with lots of latency then either.

[wikter]

The fact is that openCL and Vulkan standards are pushing to change the game.
For mixing facilities, have the chance to increase the processing power by 10x or 30x depending on the solution by only spending 200-300 bucks is something not to overlook.
Of course, the way GPU operates separating processes in small tasks doesn't fit in the most common audio processing engines, and the fact that each task introduces around 32/512 samples lag limits the usage for recording or live playing... But not for mixing, arranging and tracking.
In video processing, a simple Nvidia Geforce GTX1050 Ti processes up to 15x faster a video than a Xeon E5-2670...

[jdnz]

n video processing, a simple Nvidia Geforce GTX1050 Ti processes up to 15x faster a video than a Xeon E5-2670...

and can be much higher - we moved our molecular dynamics back-end processor from cpu-computer (on a dual 10-core xeon) to gpu-compute (on a titan rtx) - 70x speed up!

[Andre van Kammen]

AUTO-ADMIN: Non-MP3, WAV, OGG, SoundCloud, YouTube, Vimeo, Twitter and Facebook links in this post have been protected automatically. Once the member reaches 5 posts the links will function as normal.

I've seen sound shaders (GPU based) on shadertoy and made some there myself like this one that generates music

https://www.shadertoy.com/view/XsfXWf (https://www.shadertoy.com/view/XsfXWf)

Last year I was wondering why still nobody has bult a synth using this technique and started to build my own. I would like it to become a full DAW over time where people can create their own instruments using shaders and others can use these instruments to make music, all web based. GPU's have plenty of power, in a recent test on my GTX1070 i could calculate arround 100.000 sound reflections in real time. I'm still working on it, currently it can play midi files from the https://bitmidi.com (https://bitmidi.com) website. It has generation shader and effect shader support and i'm working on a new pipeline. And it supports Midi 1.0 keyboards through webMIDI. You can check my progress at

https://shadersynth.com (https://shadersynth.com)

[MuzikFreq]

For gpu audio testing, Trying in 10 instances of Thorn with a heavy cpu patch. Can't wait for the flames

[jens]

For the folks in these forums that are NOT multitracking live guitars or vocals (quite a few I imagine) 2 ms is nothing .. I would trade that for increased processing power any day. In fact, until I upgraded my laptop recently lack of processing power was the primary source of latency while tracking MIDI.

I don't know what that even means - multitracking live guitars? Live? Like in a concert? Playing multiple guitars at once? That would be neat!

[wikter]

For gpu audio testing, Trying in 10 instances of Thorn with a heavy cpu patch. Can't wait for the flames

¿?

[syntonica]

This is doable, but really only with integrated GPUs with DMA. Apple's new M1, with its many coprocessors, is the perfect candidate and quite possibly, some of ARM Logic's processing may already be offloaded. I haven't really been keeping up with it all.

With discrete GPUs, there indeed is the latency involved with transferring the data back and forth. However, most things that the GPU can handle are already handled by SIMD.

It will be interesting to see what the future brings.