When we say Voidcom uses about a third of the RAM of Discord, that’s a real claim — and we want it to stay a real claim, not a marketing flourish. This post explains exactly what those numbers mean, how we measure, and what the actual values look like on real hardware.
What the website currently shows
Two numbers appear in a few places on this site:
- ~170 MB idle RAM for Voidcom
- ~560 MB idle RAM for Discord
- A “~3× less memory” callout
These are observed values from running both apps side-by-side on the same Windows 11 machine. They are not synthetic benchmarks, they are not best-case cherry picks, and they are not measured under load — they’re what most people will see while the app is sitting in the tray on a normal day.
The actual measurements we’ve run
Snapshots taken on the same Windows 11 machine, same session, both apps signed in to a single guild:
| Run | State | Voidcom | Discord (sum of all processes) | Ratio |
|---|---|---|---|---|
| 1 | Active — clicking around the app | 210 MB | 703 MB | 3.3× |
| 2 | Idle — sitting on the friends list | 173 MB | 556 MB | 3.2× |
| 3 | Connected to a voice channel (both apps) | 219 MB | 620 MB | 2.8× |
| 4 | Voice + webcam enabled (both apps) | 278 MB | 666 MB | 2.4× |
| 5 | Both apps active in another session | 240 MB | 803 MB (8 processes) | 3.3× |
Voidcom runs as a single process; Discord runs as six to eight (one renderer per workspace plus helpers), and we sum them for a fair comparison. Voidcom’s memory footprint also drops back down quickly once you stop interacting — the active-vs-idle gap is real but transient.
Run 3 is arguably the most honest comparison, because voice is what people actually open these apps for. With both clients sitting in a voice channel, Voidcom is using just over a third the RAM of Discord. Run 4 is the heaviest realistic load — webcam capture and encode running on both sides — where Voidcom’s video FFI buffers push memory up faster than Discord’s already-large baseline; the ratio narrows to 2.4× but Voidcom is still the lighter app.
The headline numbers on the site round to ~170 MB / ~560 MB and a ~3× ratio, taken from the idle run — the conservative end of what we’ve actually observed across all five runs (range: 2.4× to 3.3×).
How we measure
The protocol is deliberately boring so it’s reproducible. If you want to repeat the measurement on your own machine, you should land in the same ballpark.
- Same machine, same session. Windows 11 Pro, NVMe SSD, ethernet. Both apps tested back-to-back, with no other browser tabs or background apps competing for memory.
- Cold start. Reboot Windows. Open the app. Don’t click anything else.
- Sign in, join one server with one channel. No call active, no video, no screen share. Microphone unplugged or muted.
- Wait 60 seconds. Idle. Don’t scroll, don’t type, don’t hover.
- Minimize to tray. Wait another 30 seconds.
- Read Task Manager → Details tab → “Memory (active private working set)” for the parent process plus all children. Discord spawns several renderer processes (typically six to eight); we sum them. Voidcom is a single process; that’s the value we use.
- Repeat the whole flow three times after fresh reboots and average.
Cold-start time follows the same protocol: stopwatch from double-click on the desktop icon to the first frame the user can interact with. Average of three runs after reboots.
Why Discord’s number is a range
Discord’s memory usage swings noticeably between sessions — from ~556 MB to over 700 MB on the same machine, doing the same thing. That’s the nature of an Electron app: each renderer process and each cached page can hold onto memory independently. Voidcom’s footprint is more stable because there’s only one process and it doesn’t cache web pages it isn’t showing.
We use the lower end of the Discord observations (~560 MB) on the homepage. We could honestly use 700 MB and call it 4×; we don’t, because the conservative end is the more defensible claim.
What’s changing before launch
We’re locking down a more rigorous set of measurements that will replace the current observed values:
- A documented hardware profile (CPU model, RAM size, drive model) so the numbers are reproducible.
- Raw Task Manager screenshots embedded directly in this post.
- Equivalent measurements during a 5-minute idle voice call (still no video) to capture the runtime memory floor, not just the open-and-do-nothing floor.
- Equivalent measurements with five guilds joined and 100+ unread channels, because that’s closer to what a real Discord power user looks like.
When those land, we’ll update this page and the homepage in lockstep. If Voidcom’s real number ends up higher than ~170 MB, we change the homepage number — we don’t change the homepage claim and quietly hope nobody compares.
Why we publish this at all
Most apps say “uses less RAM” and ask you to take their word for it. We’d rather give you the recipe so you can verify the claim on your own hardware. If you run the test and see something different, tell us — that’s a bug report we want to receive.