Light or Performance?
Every simpl server is one of two types that differ in how CPU is shared, which decides which one fits your game.
Most games should start on Light and move to Performance if the game can’t be capped or can’t tolerate a slowdown.
Light
From $0.010/hrYour server shares a CPU core with other servers on the same machine, and you always get a guaranteed baseline slice of that core, plus burst credit you can borrow when you need extra for a moment, which refills when you use less than your baseline and drains when you use more.
Use Light if:- Your game loop is capped, or can be
- It's a light, low-CPU game like a card game, turn-based game, or lobby
- It sometimes needs a burst of extra CPU, but is fine the rest of the time
Performance
From $0.072/hrYour server gets a full CPU core that is never shared, so there’s no baseline and burst split, and you have 100% of that core, 100% of the time, whether you’re using 5% of it or all of it.
Use Performance if:- Your game loop can't be capped, or capping it would hurt the game
- It's a heavy, CPU-driven game that needs a full core consistently
- It can't tolerate a slowdown, even briefly
Every Performance size includes at least 2 GB of RAM per core.
Why tick rate matters
Light is built for servers that spike now and then and idle the rest of the time, like a lobby waiting for players or a turn-based game between moves, since those rarely touch their burst credit and so run indefinitely.
An uncapped real-time loop works differently, because it tries to use as much CPU as it can every frame, which on Light means constant bursting, and since burst credit isn’t unlimited, the server slows down once it runs out while players are connected.
Cap your tick or frame rate.
Set it to a fixed, modest rate, for example 30 or 60 ticks per second, instead of “as fast as possible.” A capped loop uses CPU in short, predictable bursts, which is exactly the pattern Light is built for, and most games that struggle on Light are fixed this way.
What we’ve tested
We ran real servers on Light sizes and measured what happened.
Stable indefinitely, with no throttling and no burst credit drain.
The same server was throttled to between 55 and 85%, because an uncapped loop asked for more than the size provides, spent its credit, and got slowed down.
With 2 clients connected it drained about 3 seconds of burst credit per minute, and measured throttling began around the 16 minute mark of continuous play.
A partner's live production card game runs on our smallest Light size without issue.
The takeaway: engine alone doesn’t decide whether Light works, since tick rate and game type do. A capped Unity or Unreal server stays low and steady, a lightweight, low-state game can run on an even smaller size, and an uncapped server on any engine will eventually hit the same wall, just on a different timeline.
How to tell if you’re being throttled
When a Light server uses up its burst credit and falls back to its baseline CPU, Simpl flags it in real time with a Slowed down badge next to the server on your dashboard.
What triggers it: sustained CPU use above your baseline after burst credit has hit zero.
Make sure your game loop isn't running uncapped, since capping a 60 Hz or 30 Hz loop usually brings CPU use back under your baseline, and burst credit recharges while the server stays live.
If your game truly needs continuous, unthrottled access to a full core, such as high-frequency physics or complex AI pathfinding, switch the server to a Performance size.
Not sure? Start on Light.
Cap your tick rate and watch your dashboard, and if you see Slowed down, cap first and then move to Performance if it keeps happening.