Why I Run AudioGridder On Three Old PCs Instead of One Big Rig

I’ve been in Cubase for more than 25 years now — long enough that my relationship with the software has gone through several distinct hardware eras. Each one was built around whatever machine could keep up with the last one straining. Every generation of new gear inherited a heavier session than the one before it, for a simple reason: I never really stopped adding plugins.

I’ve been buying and trying plugins for two and a half decades now, keeping the ones that earned a permanent spot in my sessions. That adds up to a genuinely large collection at this point. A meaningful chunk of it is also the resource-hungry kind: convolution reverbs sampling real rooms, orchestral libraries streaming gigabytes of samples per instrument, saturation and analog-modeling plugins doing real-time circuit simulation instead of simple math. None of that was a problem back when a session might use five or six plugins total. It’s a very different problem when a real mix leans on fifty.

Every so often someone asks why I haven’t just built one monster machine to carry all of that. Something with enough cores and RAM to chew through a full mix session without breaking a sweat. The honest answer is: I did the math, and it made more sense to keep my old workstations working for a living instead of retiring them.

Repurposing Instead of Retiring

My main studio PC handles the interactive side of a mix — Cubase itself, the UI, anything that needs to feel instant. But the heavy lifting gets handed off over the network to a small fleet of older machines I’d otherwise have sold or shelved. That’s the convolution reverbs, the multi-instance orchestral libraries, the plugins that make your CPU meter climb just looking at them. An i9-9900K with an RTX 4080, an i9-10900 with an RTX 5060 Ti, and a newer Ryzen mini PC all sit headless, doing nothing but crunching audio through AudioGridder while my main rig stays light and responsive.

What AudioGridder Actually Does

AudioGridder is the piece that makes this work. It’s an open-source plugin host that lets you run VST instances on a remote machine, streaming the audio back over your LAN with surprisingly low latency — as if the plugin were running locally. Once you’ve got a 2.5GbE network in place, the round-trip is fast enough that you stop thinking about which machine is doing the work.

This only works as smoothly as it does because of the network underneath it. Every one of these machines is connected with an actual cable, not Wi-Fi. A wired connection is faster and far more consistent than a wireless one, and audio has zero tolerance for the tiny hiccups Wi-Fi occasionally introduces. If you’re curious what that home network actually looks like, that’s a separate story.

The practical upside is real during mixing and mastering, when the plugin count on a session balloons. Instead of freezing tracks or bouncing stems just to keep the session usable, I spread the load and keep everything live and editable right up until the final print. It also means none of that older hardware is dead weight — it’s earning its keep every time I open a project.

Where This Actually Pays Off

A few places this setup earns its keep in practice, running over a wired 2.5GbE network with a solid, low-latency connection between every node.

Loading Full Orchestral Templates

Loading a full orchestral template is the clearest case. Something like a deep sample library can eat tens of gigabytes of RAM and dozens of CPU threads just sitting there loaded. That’s before a single note plays. Offload that entire template to one dedicated node, and my main PC never has to carry that weight at all. It just sees the audio coming back.

Convolution Reverb

Convolution reverb is another good example. It’s genuinely one of the most CPU-hungry categories of plugin that exists. It’s mathematically processing audio against a real recorded space, instead of faking the effect with a simpler algorithm. Running two or three convolution reverbs on a busy mix used to mean choosing between a great-sounding room and a responsive session. Now it just means picking which node handles it.

Real-Time Metering During Mastering

During a mastering pass, I often want several loudness meters, analyzers, and processing plugins running at once, so I can actually watch what a chain is doing in real time. Spreading that instrumentation across nodes means none of it competes with the main audio engine for CPU time. The meters stay accurate and playback stays glitch-free. That matters when the judgment call you’re making depends on seeing and hearing the same thing at the same instant.

Once a project’s actually finished, it needs somewhere safe to live too, which is a whole separate story about the NAS setup backing all of this up.

Other Ways to Solve This

AudioGridder isn’t the only way to handle a session that’s outgrown one computer, and it’s worth knowing what else is out there. Some companies sell dedicated hardware for this exact problem — physical add-in cards or external boxes packed with their own processing power, built specifically to run that company’s own plugins. That hardware works well, but it only runs plugins made for it, and the hardware itself costs real money on top of what you already own. Other studios rent processing power from cloud services instead, paying a monthly fee to borrow someone else’s powerful computer over the internet.

AudioGridder takes a different path. It’s free, it’s open-source, and it works with the plugins you already own rather than locking you into a specific format. Instead of paying for new hardware or an ongoing subscription, it puts machines you already have back to work.

If you’re staring at an aging PC wondering whether to sell it or build around it, this is the case for the second option.

(If you ever want help putting something like this together for your own setup — network design, repurposing older hardware, the whole thing — that’s also literally an example of what I do at Raindog Solutions.)