Profile your FiveM server with resmon

Introduction

resmon (Resource Monitor) is built into FiveM client and server. It shows real-time CPU usage and RAM consumed by each resource. The #1 tool to understand why your server lags, why a script crashes, or why a resource consumes 200 MB when it should consume 20.

Prerequisites

• A FiveM server at VeryCloud
• A FiveM client (you in-game as admin)
• Access to F8 console

Step 1: Open resmon (client)

In-game, F8 console:

resmon

Or bind to a key:

bind keyboard f3 "resmon"

You get a floating window with resource list, CPU% and MEM.

Step 2: Read the indicators

Thresholds to know:

• CPU msec < 0.5ms: OK
• CPU msec 0.5–1ms: watch
• CPU msec > 1ms: optimization needed
• CPU msec > 3ms: that's what ruins your perf

Step 3: Identify problem resources

Sort resmon by decreasing CPU msec. Top 2-3 resources consume the most.

Often:

• Frameworks (ESX, QBCore): 0.5-1.5ms — normal as they're everywhere
• ox_lib or util: 0.1-0.3ms — OK
• A custom resource at 2-5ms → suspect, profile it
• Anticheat: 0.5-1.5ms — normal but monitor

Step 4: Fine profiling with ProfData

Beyond resmon, FiveM exposes a more advanced profiler:

profiler record 1000
# wait 10 seconds of normal gameplay
profiler view

Generates a dump viewable in Chrome DevTools (chrome://tracing).

profiler save mygame_profile.json

Detailed timeline analysis of hooks and events.

Step 5: Diagnose a memory leak

If a resource's memory climbs without ever going down:

1. Note Memory value at T0
2. Wait 30 minutes of normal gameplay
3. Compare

Doubled without functional reason → leak.

Classic causes:

• Unsubscribed event listeners (cumulative RegisterNetEvent)
• Lua tables growing without purging (logs[#logs+1] = ... never cleared)
• Uncleared timers (looping SetTimeout)

Step 6: Optimize a Lua resource

CPU reduction techniques:

Increase Wait in loops:

-- BAD: tick every frame
CreateThread(function()
    while true do
        Wait(0)
    end
end)

-- GOOD: every 500ms if non-critical
CreateThread(function()
    while true do
        Wait(500)
    end
end)

Avoid recomputes:

-- BAD: recompute every call
function GetNearbyPlayers()
    local players = {}
    for _, p in ipairs(GetActivePlayers()) do
        if #(GetEntityCoords(...) - ...) < 50 then
            table.insert(players, p)
        end
    end
    return players
end

-- GOOD: cache + periodic refresh
local _cache = {}
CreateThread(function()
    while true do
        _cache = ComputeNearbyPlayers()
        Wait(1000)
    end
end)
function GetNearbyPlayers() return _cache end

Targeted events instead of global broadcasts:

-- BAD: everyone receives
TriggerClientEvent('myevent', -1, data)

-- GOOD: only the concerned player
TriggerClientEvent('myevent', playerId, data)

Step 7: Optimize on natives side

Major consumers:

• GetEntityCoords() in tight loops → cache
• GetClosestVehicle() → batch less frequently
• DrawText3D() every frame without distance guard

-- Guard before heavy draw
local plyCoord = GetEntityCoords(PlayerPedId())
local targetCoord = GetEntityCoords(target)
if #(plyCoord - targetCoord) < 20 then
    DrawText3D(...)
end

Step 8: Reduce RAM usage

Lua-side:

• Clear unused tables: myTable = {}
• Occasional collectgarbage('collect') after big operations
• Don't store entities in tables — only IDs

Assets-side:

• Compress YTDs via OpenIV
• Delete unused stream resources

Step 9: Server-side profiling

FXServer console:

prof start
prof stop

Or txAdmin → System Logs → CPU server-side per resource.

A resource consuming >5% CPU server-side continuously is suspect.

Step 10: Scheduled restarts

Well-coded resources shouldn't need restarts. But many third-party scripts have micro-leaks. Schedule auto-restarts:

• Every 12h via Wisp Schedules (Action: Power → Restart)
• Announce in-game 5 min before via Discord webhook or broadcast
• Off-peak time (5 AM)

Troubleshooting

resmon shows nothing — you're not admin (sometimes required), F8 wrong binding (try ~ or ^). Huge CPU msec on a "idle" resource — infinite loop without Wait somewhere; memory leak forcing constant GC. Server lags but no resource >1ms — not FiveM the issue. Check hardware (Wisp graphs) or network latency.

Useful commands

resmon
profiler record 1000
profiler view
prof start
prof stop
restart <resource>
stop <resource>

Conclusion

resmon = perf swiss army knife. 10 minutes to identify hungry resources, a few hours to fix the worst bottlenecks, server goes from "lag at 50 players" to "stable at 100". Combined with scheduled restarts you have a clean operation. Advanced profiler (profiler record) when resmon isn't enough.

Going further: load testing with bots, external monitoring via custom Prometheus exporter, hot path optimization in C++ via natives.

Resources

• FiveM Docs — Console Commands
• FiveM Docs — Performance
• Cfx.re Forum — Optimization
• Chrome DevTools — Performance tab