One of the bottlenecks when using computers to create music is the CPU (central processing unit), the chip that does all the processing. Most musicians don’t reach the limits of their machines, but if you like to use lots of effects plugins (specialized apps that run inside the main app), you’ll eventually bring your machine to a hot-and-bothered crawl. Now, one plugin maker is running its software on your computer’s GPU (graphics processing unit or graphic processor), which can allow much more complex effects and/or more effects simultaneously. So why haven’t we seen this before? “Because it is offloading the processing from the CPU to the GPU, it will free up the CPU, which In turn gives us the opportunity for opening more tracks, instruments, and effects panels,” music producer Nikhil Koparde told Lifewire via email. “Not all DAWs & plugins currently support this, so this will take some time to catch up within the music industry.”

Graphic Content

To visualize the difference between a CPU and a GPU, let’s try a chef metaphor. A CPU is one skilled chef that can work very, very fast, whereas a GPU is a room of hundreds of chefs, all of whom work slower and have fewer skills, but combined, they can prep food way faster.  The CPU in your computer can do anything, but a GPU is optimized for this kind of parallel processing by using many computing cores simultaneously. This makes it incredibly fast for certain tasks, like shifting graphics or computing Bitcoins. “GPUs have many more cores than CPUs, which means they can handle more tasks simultaneously. In addition, GPUs have access to more memory than CPUs, allowing them to store more data and processing instructions,” Jeroen van Gils, managing director at tech company Lifi.co told Lifewire via email. So why not just use GPUs for everything? Because they work best when a task can be divided into blocks that can be executed simultaneously, like all those chefs chopping onions. And that’s the problem that GPU Audio is solving with its plugins. 

GPU FTW

GPU Audio’s plugin suite is in open beta, so you can try it out for yourself if you have a Windows machine. The Beta Suite offers classic effects—chorus, flanger, and phaser (which are—let’s be honest—variations on the same effect). But it’s the fact that they are running audio effects on graphics chips that is the big news. By their nature, GPUs are highly optimized for the tasks they perform—massive parallel processing of computer graphics. They are not suited to audio.  “The issue historically has been that GPU pipelines are built for 120 frames a second, not 48,000,” adds electronic musician Dymaxion on the Elektronauts forum.  “I’m 99% sure the reason GPU audio isn’t widespread yet is because of high latency. It’s probably doable on modern hardware, though,” electronic musician skinpop said in the same forum thread.  But GPU Audio says it has solved this problem, with a one-millisecond buffer when used as a plugin and 150 microseconds when used in custom-built software. For comparison, if you are sitting ten feet away from a speaker, it takes just under nine milliseconds for the sound to reach you through the air.  The other problem is that Digital Audio Workstation (DAW) software like Logic Pro and Ableton Live, which are used to host audio plugins, do not yet support GPU audio. But that could easily change if this technology takes off.  Finally, for most people, there may not be much point in even bothering with GPU audio processing. Music hardware, like drum machines and samplers, already have a long-established equivalent of the GPU: the DSP, or digital signal processor, is a chip optimized for audio processing and is just fine. And computer GPUs themselves are already more than powerful enough to handle most musical demands. Returning to the chef metaphor, it’s like you have the world’s most athletic, least-alcoholic chef working in a lazy village restaurant on a Tuesday evening. You just don’t need that room of onion-chopping assistants.  That’s not to say GPU audio won’t be a thing in the future. It’s just that now, it is both too young and not significantly better than what we already have.