Benchmark Report
CPU quota
The JVM sees far fewer cores than the host advertises. Everything that parallelises is bounded by the container, not by the machine.
- host_threads = 32 cores
- usable_cores = 5 cores
- core_ratio = 6.4 ratio
Weakest test: Mob AI, 4.99 TPS on average. The 15 other tests hold at least 9.55 TPS.
Compared with other servers
The comparison with other servers uses reports scored with version 8.3.0 of the scoring rules. This report was scored with another version, so its score is not comparable.
What matters
Read from the measurements, never estimatedCPU quota
The JVM sees far fewer cores than the host advertises. Everything that parallelises is bounded by the container, not by the machine.
Likely cause: CPU quota
- host_threads = 32 cores
- usable_cores = 5 cores
- core_ratio = 6.4 ratio
Tick spikes
The worst ticks are several times the mean. Average TPS stays fine while players feel the stutter.
- mspt_avg = 573.9 ms
- mspt_p99 = 1,328.82 ms
- spike_ratio = 2.32 ratio
- sample_count = 41 count
Non-default server configuration
Compare with care against a server running default settings.
- spawn-limit.water-ambient 15 (default 20)
›+6 more findings
chunkLoadingThe worst ticks are several times the mean. Average TPS stays fine while players feel the stutter.
- mspt_avg = 14.1 ms
- mspt_p99 = 50.86 ms
- spike_ratio = 3.61 ratio
- sample_count = 102 count
networkThe worst ticks are several times the mean. Average TPS stays fine while players feel the stutter.
- mspt_avg = 74.6 ms
- mspt_p99 = 178.3 ms
- spike_ratio = 2.39 ratio
- sample_count = 137 count
memorynot measuredNo access latency was published, so the score assumed an unfavourable one. Publishing the measurement can only help.
- assumed_latency_ns = 180 ns
- reference_latency_ns = 120 ns
disknot measuredDirect IO was refused by the host, so the random part of the disk score rests on an assumption rather than a measurement.
- assumed_random_mbps = 90 MB/s
- reference_random_mbps = 150 MB/s
blockPhysicssingle coreOne game thread is already over its tick budget under the benchmark's own load. Players and redstone will push it further.
- mspt_avg = 43.1 ms
- mspt_target = 35 ms
- tps_avg = 19.74 tps
- tps_held_threshold = 18 tps
redstonesingle coreOne game thread is already over its tick budget under the benchmark's own load. Players and redstone will push it further.
- mspt_avg = 56 ms
- mspt_target = 35 ms
- tps_avg = 19.95 tps
- tps_held_threshold = 18 tps
A finding stays silent rather than attribute a cause without evidence. An absent cause is an answer, not an omission.
Test by test
Problems first — select a row for the full test- Mob AIEntities4.99UnstableCritical
- HopperMechanics9.55StableUnstable
- Network (NBT)Network13.98UnstableUnstable
- Chunk TickingExploration18.26UnstableUnstable
- Chunk LoadingExploration18.77UnstableUnstable
- Block PhysicsSingle-Core Stress19.74UnstableHealthy
- RedstoneSingle-Core Stress19.95UnstableHealthy
- Disk PerformanceMemory & Storage19.96StableHealthy
- Multi-Core CPUCPU19.97StableHealthy
- ExplosionMechanics19.98ModerateHealthy
- Lighting UpdateExploration19.98StableHealthy
- Memory PerformanceMemory & Storage19.98StableHealthy
- Bone Meal GrowthMechanics19.99StableHealthy
- Single-Core (tick-aligned)CPU19.99StableHealthy
- World SaveMechanics19.99StableHealthy
- Ticking Tile EntitiesMechanics20.00StableHealthy
Settings your measurements justify
Each setting below is triggered by a measurement from this report, and names the test to replay to find out whether it helped.
Switch the redstone implementation
misc.redstone-implementation: ALTERNATE_CURRENTMeasured : redstone.mspt.avg 56.0 ms (threshold 35.0 ms)
Redstone costs more per tick than the budget allows. Paper ships an alternative wire implementation that computes the same result without the repeated updates of the vanilla one.
Test to replay : redstone
Shrink the monster activation range
world-settings.default.entity-activation-range.monsters: 24Measured : mobAI.mspt.avg 573.9 ms (threshold 35.0 ms)
Mob AI costs more per tick than the budget allows. The activation range decides how far from a player a monster keeps thinking; shrinking it puts distant ones to sleep without removing them.
Test to replay : mobAI
Lower the monster spawn limit
spawn-limits.monsters: 50Currently : 70
Measured : mobAI.mspt.avg 573.9 ms (threshold 50.0 ms)
Mob AI is past the critical cost per tick, where shrinking the activation range is no longer enough. The spawn limit caps how many monsters exist per player at all, which is the only setting here that reduces the work rather than deferring it.
Test to replay : mobAI
Lower the simulation distance
simulation-distance: 8Currently : 10
Measured : chunkTicking.mspt.avg 51.8 ms (threshold 50.0 ms)
Chunk ticking is past the critical cost per tick. The simulation distance decides how many chunks tick around each player; lowering it a notch is the bluntest lever here and the one players notice, so it is offered only when the value in use is known.
Test to replay : chunkTicking
These settings trade a little gameplay fidelity for tick budget. Apply one at a time and replay the named test: the effect stays a hypothesis until you have measured it.
Briefing for a language model
Paste this into the assistant of your choice to have it rewritten in prose. It already carries every number, so the model has none left to invent.
{
"format": "voxelbench.tuning.v1",
"instruction": "Rewrite each recommendation below as plain prose for a server operator. Every number you are allowed to use is already in this document: the measured value, its threshold, and the proposed setting. Do not compute, estimate or predict any other number, and never state an expected gain. Keep every setting name and value exactly as written.",
"server": {
"type": "paper",
"minecraft": "1.21.11",
"java": "21.0.10"
},
"recommendations": [
{
"id": "paper.redstone-alternatif",
"family": "paper",
"measure": "redstone.mspt.avg",
"value": 56,
"unit": "ms",
"threshold": 35,
"control_test": "redstone",
"file": "config/paper-world-defaults.yml",
"setting": "misc.redstone-implementation",
"proposed_value": "ALTERNATE_CURRENT",
"current_value": null
},
{
"id": "spigot.portee-activation-monstres",
"family": "spigot",
"measure": "mobAI.mspt.avg",
"value": 573.9,
"unit": "ms",
"threshold": 35,
"control_test": "mobAI",
"file": "spigot.yml",
"setting": "world-settings.default.entity-activation-range.monsters",
"proposed_value": "24",
"current_value": null
},
{
"id": "bukkit.limite-monstres",
"family": "bukkit",
"measure": "mobAI.mspt.avg",
"value": 573.9,
"unit": "ms",
"threshold": 50,
"control_test": "mobAI",
"file": "bukkit.yml",
"setting": "spawn-limits.monsters",
"proposed_value": "50",
"current_value": "70"
},
{
"id": "serveur.distance-de-simulation",
"family": "serveur",
"measure": "chunkTicking.mspt.avg",
"value": 51.8,
"unit": "ms",
"threshold": 50,
"control_test": "chunkTicking",
"file": "server.properties",
"setting": "simulation-distance",
"proposed_value": "8",
"current_value": "10"
}
]
}Java start-up line
The options this server was started with, checked against its Java version. Corrections fix what is certainly wrong; leads to test are settings to measure, never applied for you.
Current line (4 arguments)
Values of system properties, agents and file paths are never shown: they appear as <redacted>.
-Xms128M -Xmx16000M -Dterminal.jline=false -Dterminal.ansi=true
To correct
- DangerHeap too close to the container's limit
-Xmx16000MThe heap may use 16000 MiB of the container's 16800 MiB, which leaves 800 MiB for everything else (code, threads, network buffers). Under load the server can be killed without any Java error. The corrected line keeps 2520 MiB outside the heap.
Corrected line to copy
- Changed
-Xmx16000M-Xmx14280Mleaves 2520 MiB outside the heap.
java -Xms128M -Xmx14280M -jar server.jar noguiReplace server.jar with your server's jar. Paste the line where your server is started: the start-up parameters (or JVM arguments) of your hosting panel, or your start.sh / start.bat. Only the corrections above differ from your current line.
Your current line also carries 2 other arguments (system properties, agents): keep them between java and -jar. Their values are not shown here.
To test
Performance settings to try one at a time, then measure with a benchmark before and after. None of them is applied in the corrected line, and none comes with a promised gain.
Settings documented by PaperMC for Minecraft servers: they size G1's young generation for the game's very high allocation rate and start collections earlier. Initial and maximum heap at the same size, with AlwaysPreTouch, commit the whole heap at start-up: proposed only because that heap already stays under the memory limit.
-Xms14280M -XX:+UseG1GC -XX:+ParallelRefProcEnabled -XX:MaxGCPauseMillis=200 -XX:+UnlockExperimentalVMOptions -XX:+DisableExplicitGC -XX:+AlwaysPreTouch -XX:G1NewSizePercent=30 -XX:G1MaxNewSizePercent=40 -XX:G1HeapRegionSize=8M -XX:G1ReservePercent=20 -XX:G1HeapWastePercent=5 -XX:G1MixedGCCountTarget=4 -XX:InitiatingHeapOccupancyPercent=15 -XX:G1MixedGCLiveThresholdPercent=90 -XX:G1RSetUpdatingPauseTimePercent=5 -XX:SurvivorRatio=32 -XX:+PerfDisableSharedMem -XX:MaxTenuringThreshold=1No measurement of this run points to it: it stays a lead to measure.
sourceA collector with very short pauses for large heaps, at the cost of more CPU and memory. It replaces the current collector: remove the current collector option and its own tuning options, then measure. On Java 21 and 22 the generational mode needs -XX:+ZGenerational.
-XX:+UseZGC -XX:+ZGenerationalNo measurement of this run points to it: it stays a lead to measure.
sourceRe-run and compare
- Apply the corrected line where your server starts (panel start-up parameters, start.sh).
- Restart the server.
- Run a new benchmark:
/bench start - Compare the new report with this one.
This report is the before; add the new report to the comparison once it is in. A single run on each side says little: the gap between two runs of an unchanged server is noise.
Every measurement
Each test in full. Tests in trouble are open; the others fold to their summary line.
Single-Core Stress
RedstoneHealthy19.95 TPS56.0 ms47.0s
Single-core stress via intensive redstone circuit updates.
- Heavy garbage collection61 collections, 1539ms total
Percentiles and stability
Raw data (16 fields)
Block PhysicsHealthy19.74 TPS43.1 ms25.6s
Single-core stress via falling-block physics.
- Heavy garbage collection8 collections, 1058ms total
Percentiles and stability
Raw data (22 fields)
Exploration
Chunk LoadingUnstable18.77 TPS14.1 ms31.8s
Chunk generation and loading speed across dispersed zones.
- Unfiltered samples usedUsing unfiltered samples (filter rate: 55%)
- Heavy garbage collection17 collections, 3517ms total
- The run did not hold steadyTPS avg=18.77<19, TPS P1=10.00<15, TPS StdDev=3.10>2, TPS min=10.00 (53% of avg)
Percentiles and stability
Raw data (13 fields)
Chunk TickingUnstable18.26 TPS51.8 ms45.0s
Server performance when ticking many loaded chunks with random tick updates.
- Unfiltered samples usedUsing unfiltered samples (filter rate: 18%)
- Heavy garbage collection18 collections, 1388ms total
- The run did not hold steadyTPS avg=18.26<19, TPS P1=10.00<15, TPS StdDev=2.62>2, TPS min=10.00 (55% of avg)
Percentiles and stability
Raw data (16 fields)
Lighting UpdateHealthy19.98 TPS7.1 ms25.3s
Lighting recomputation throughput on block place / break.
- Heavy garbage collection4 collections, 382ms total
Percentiles and stability
Raw data (11 fields)
Entities
Mob AICritical4.99 TPS573.9 ms32.4s
Complex mob AI behaviours (villagers, iron golems, hostile mobs).
- Unfiltered samples usedUsing unfiltered samples (filter rate: 17%)
- Heavy garbage collection12 collections, 1698ms total
- The run did not hold steadyTPS avg=4.99<19, TPS P1=4.92<15
Percentiles and stability
Raw data (28 fields)
Mechanics
HopperUnstable9.55 TPS10.2 ms41.0s
Intensive hopper chain item transfers.
- Unfiltered samples usedUsing unfiltered samples (filter rate: 5%)
- Heavy garbage collection17 collections, 1798ms total
- The run did not hold steadyTPS avg=9.55<19, TPS P1=7.00<15, TPS min=6.50 (68% of avg)
Percentiles and stability
Raw data (23 fields)
ExplosionHealthy19.98 TPS12.9 ms17.7s
Massive TNT explosions and block destruction.
- Heavy garbage collection6 collections, 973ms total
Percentiles and stability
Raw data (64 fields)
Ticking Tile EntitiesHealthy20.00 TPS6.6 ms37.3s
Many ticking tile entities (furnaces, hoppers, etc.).
- Heavy garbage collection3 collections, 1026ms total
Percentiles and stability
Raw data (9 fields)
World SaveHealthy19.99 TPS5.1 ms4.9s
World save speed and TPS impact during save operations.
- Heavy garbage collection3 collections, 796ms total
Percentiles and stability
Raw data (7 fields)
Bone Meal GrowthHealthy19.99 TPS6.4 ms23.6s
Rapid crop and tree growth via bone meal.
- Heavy garbage collection1 collections, 336ms total
Percentiles and stability
Raw data (13 fields)
CPU
Single-Core (tick-aligned)Healthy19.99 TPS7.4 ms10.0s
Single-thread compute throughput measured on the main thread under the tick scheduler.
- Heavy garbage collection3 collections, 863ms total
Percentiles and stability
Raw data (8 fields)
Multi-Core CPUHealthy19.97 TPS5.4 ms34.4s
Parallel computation throughput across worker threads.
- Heavy garbage collection1 collections, 291ms total
Percentiles and stability
Raw data (9 fields)
Memory & Storage
Memory PerformanceHealthy19.98 TPS6.7 ms128.6s
RAM bandwidth and latency.
JVM measurements — do not reflect actual bare-metal performance.
- Heavy garbage collection13 collections, 1062ms total
Percentiles and stability
Raw data (12 fields)
Disk PerformanceHealthy19.96 TPS7.3 ms76.8s
Sequential and random 4K read/write performance.
JVM measurements — do not reflect actual bare-metal performance.
- Heavy garbage collection43 collections, 967ms total
Percentiles and stability
Raw data (7 fields)
Network
Network (NBT)Unstable13.98 TPS74.6 ms29.5s
Serialisation throughput — proxy for the rate at which the server can ship world data to clients.
- Unfiltered samples usedUsing unfiltered samples (filter rate: 8%)
- Heavy garbage collection39 collections, 524ms total
- The run did not hold steadyTPS avg=13.98<19, TPS P1=4.97<15, TPS StdDev=3.16>2, TPS min=4.90 (35% of avg)
Percentiles and stability
Raw data (11 fields)
›Machine and serverAMD Ryzen 9 5950X 16-Core Processor · Paper · 1.21.11
Score Breakdown
Detailed sub-category scores