Hewwo there, I'm glad you are here ! This instruction will guide you to use MeowIce's Flags to make your Minecraft server run as it best ^^

My flags are only compatible and tested with:

• Java 24+,
• GraalVM JVM,
• Linux,
• Minecraft 1.20+

--add-modules=jdk.incubator.vector -XX:+UseG1GC -XX:MaxGCPauseMillis=200 -XX:+UnlockExperimentalVMOptions -XX:+UnlockDiagnosticVMOptions -XX:+DisableExplicitGC -XX:+AlwaysPreTouch -XX:G1NewSizePercent=28 -XX:G1MaxNewSizePercent=50 -XX:G1HeapRegionSize=16M -XX:G1ReservePercent=15 -XX:G1MixedGCCountTarget=3 -XX:InitiatingHeapOccupancyPercent=20 -XX:G1MixedGCLiveThresholdPercent=90 -XX:SurvivorRatio=32 -XX:G1HeapWastePercent=5 -XX:MaxTenuringThreshold=1 -XX:+PerfDisableSharedMem -XX:G1SATBBufferEnqueueingThresholdPercent=30 -XX:G1ConcMarkStepDurationMillis=5 -XX:G1RSetUpdatingPauseTimePercent=0 -XX:+UseNUMA -XX:-DontCompileHugeMethods -XX:MaxNodeLimit=240000 -XX:NodeLimitFudgeFactor=8000 -XX:ReservedCodeCacheSize=400M -XX:NonNMethodCodeHeapSize=12M -XX:ProfiledCodeHeapSize=194M -XX:NonProfiledCodeHeapSize=194M -XX:NmethodSweepActivity=1 -XX:+UseFastUnorderedTimeStamps -XX:+UseCriticalJavaThreadPriority -XX:AllocatePrefetchStyle=3 -XX:+AlwaysActAsServerClassMachine -XX:+UseTransparentHugePages -XX:LargePageSizeInBytes=2M -XX:+UseLargePages -XX:+EagerJVMCI -XX:+UseStringDeduplication -XX:+UseAES -XX:+UseAESIntrinsics -XX:+UseFMA -XX:+UseLoopPredicate -XX:+RangeCheckElimination -XX:+OptimizeStringConcat -XX:+UseCompressedOops -XX:+UseThreadPriorities -XX:+OmitStackTraceInFastThrow -XX:+RewriteBytecodes -XX:+RewriteFrequentPairs -XX:+UseFPUForSpilling -XX:+UseFastStosb -XX:+UseNewLongLShift -XX:+UseVectorCmov -XX:+UseXMMForArrayCopy -XX:+UseXmmI2D -XX:+UseXmmI2F -XX:+UseXmmLoadAndClearUpper -XX:+UseXmmRegToRegMoveAll -XX:+EliminateLocks -XX:+DoEscapeAnalysis -XX:+AlignVector -XX:+OptimizeFill -XX:+EnableVectorSupport -XX:+UseCharacterCompareIntrinsics -XX:+UseCopySignIntrinsic -XX:+UseVectorStubs -XX:UseAVX=2 -XX:UseSSE=4 -XX:+UseFastJNIAccessors -XX:+UseInlineCaches -XX:+SegmentedCodeCache -Djdk.nio.maxCachedBufferSize=262144 -Djdk.graal.UsePriorityInlining=true -Djdk.graal.Vectorization=true -Djdk.graal.OptDuplication=true -Djdk.graal.DetectInvertedLoopsAsCounted=true -Djdk.graal.LoopInversion=true -Djdk.graal.VectorizeHashes=true -Djdk.graal.EnterprisePartialUnroll=true -Djdk.graal.VectorizeSIMD=true -Djdk.graal.StripMineNonCountedLoops=true -Djdk.graal.SpeculativeGuardMovement=true -Djdk.graal.TuneInlinerExploration=1 -Djdk.graal.LoopRotation=true -Djdk.graal.CompilerConfiguration=enterprise

--add-modules=jdk.incubator.vector -XX:+UseZGC -XX:-ZProactive -XX:SoftMaxHeapSize=$((Memory - 2048))M -XX:+UnlockExperimentalVMOptions -XX:+UnlockDiagnosticVMOptions -XX:+DisableExplicitGC -XX:+AlwaysPreTouch -XX:+PerfDisableSharedMem -XX:+UseNUMA -XX:-DontCompileHugeMethods -XX:MaxNodeLimit=240000 -XX:NodeLimitFudgeFactor=8000 -XX:ReservedCodeCacheSize=400M -XX:NonNMethodCodeHeapSize=12M -XX:ProfiledCodeHeapSize=194M -XX:NonProfiledCodeHeapSize=194M -XX:NmethodSweepActivity=1 -XX:+UseFastUnorderedTimeStamps -XX:+UseCriticalJavaThreadPriority -XX:AllocatePrefetchStyle=1 -XX:+AlwaysActAsServerClassMachine -XX:+UseTransparentHugePages -XX:LargePageSizeInBytes=2M -XX:+UseLargePages -XX:+EagerJVMCI -XX:+UseStringDeduplication -XX:+UseAES -XX:+UseAESIntrinsics -XX:+UseFMA -XX:+UseLoopPredicate -XX:+RangeCheckElimination -XX:+OptimizeStringConcat -XX:+UseCompressedOops -XX:+UseThreadPriorities -XX:+OmitStackTraceInFastThrow -XX:+RewriteBytecodes -XX:+RewriteFrequentPairs -XX:+UseFPUForSpilling -XX:+UseFastStosb -XX:+UseNewLongLShift -XX:+UseVectorCmov -XX:+UseXMMForArrayCopy -XX:+UseXmmI2D -XX:+UseXmmI2F -XX:+UseXmmLoadAndClearUpper -XX:+UseXmmRegToRegMoveAll -XX:+EliminateLocks -XX:+DoEscapeAnalysis -XX:+AlignVector -XX:+OptimizeFill -XX:+EnableVectorSupport -XX:+UseCharacterCompareIntrinsics -XX:+UseCopySignIntrinsic -XX:+UseVectorStubs -XX:UseAVX=2 -XX:UseSSE=4 -XX:+UseFastJNIAccessors -XX:+UseInlineCaches -XX:+SegmentedCodeCache -Djdk.nio.maxCachedBufferSize=262144 -Djdk.graal.UsePriorityInlining=true -Djdk.graal.Vectorization=true -Djdk.graal.OptDuplication=true -Djdk.graal.DetectInvertedLoopsAsCounted=true -Djdk.graal.LoopInversion=true -Djdk.graal.VectorizeHashes=true -Djdk.graal.EnterprisePartialUnroll=true -Djdk.graal.VectorizeSIMD=true -Djdk.graal.StripMineNonCountedLoops=true -Djdk.graal.SpeculativeGuardMovement=true -Djdk.graal.TuneInlinerExploration=1 -Djdk.graal.LoopRotation=true -Djdk.graal.CompilerConfiguration=enterprise

MeowIce's Flags aimed to blend Aikar's Flags optimization with new JVM features to further boost the overall performance, while allowing it to use advanced CPU instructions to accelerate some workloads.

Almost 99% of Minecraft servers use Aikar's Flags, and it is recommended by many people (some use nothing !). But why use other flags ?
The answer is: We have now developed to Minecraft 1.21.x and Java 24 (at the time of writing this), and Aikar's Flags was written mainly to optimize Minecraft's Garbage Collector and Java version at that time (Java 8 and Java 11, around 2018 - 2020). Later Java versions (from Java 17 and above) have added many optimization options for the JVM. So I wrote a separate flag to support these optimizations and integrate new features of Java 17+ while still keeping Aikar's flags mainly to optimize GC. There's no risk in trying my flags !

GraalVM with better JIT performance and more aggressive optimizations compared to traditional HotSpot C2 found in many JDKs (including OpenJDK and AdoptiumJDK (aka. Eclipse Temurin)). This change makes inlining, loop unrolling, escape analysis, and vectorization work better while maintaining lower CPU usage.
Furthermore, the GraalVM compiler compiles plugins and server code faster and smarter, adapting to actual usage and environment on the fly.

• --add-modules=jdk.incubator.vector Enables advanced vector computation capabilities for Pufferfish/Purpur 1.18+. Useful for optimizing chunk loading and rendering.
• -XX:+UseG1GC -XX:MaxGCPauseMillis=200 -XX:+UnlockExperimentalVMOptions -XX:+UnlockDiagnosticVMOptions -XX:+DisableExplicitGC -XX:+AlwaysPreTouch -XX:G1NewSizePercent=28 -XX:G1MaxNewSizePercent=50 -XX:G1HeapRegionSize=16M -XX:G1ReservePercent=15 -XX:G1MixedGCCountTarget=3 -XX:InitiatingHeapOccupancyPercent=20 -XX:G1MixedGCLiveThresholdPercent=90 -XX:SurvivorRatio=32 -XX:G1HeapWastePercent=5 -XX:MaxTenuringThreshold=1 -XX:+PerfDisableSharedMem -XX:G1SATBBufferEnqueueingThresholdPercent=30 -XX:G1ConcMarkStepDurationMillis=5 -XX:G1RSetUpdatingPauseTimePercent=0 These are very generic G1GC flags, based on Aikar's Flags.
• -XX:+UseNUMA Enables NUMA or (Non Uniform Memory Access) for supported architecture.
• -XX:-DontCompileHugeMethods Allows the JVM to compile larger methods to optimize performance.
• -XX:MaxNodeLimit=240000 -XX:NodeLimitFudgeFactor=8000 Sets the maximum number of nodes for the compiler's intermediate representation.
• -XX:ReservedCodeCacheSize=400M Sets the cache size for compiled codes. This will ensure the JIT have enough space for compiling.
• -XX:NonNMethodCodeHeapSize=12M Sets the code heap size for non-method codes.
• -XX:ProfiledCodeHeapSize=194M -XX:NonProfiledCodeHeapSize=194M Same as above but for profiled and non profiled methods.
• -XX:NmethodSweepActivity=1 Helps managing code cache more efficiently by keeping "cold" code in the cache for a longer time. There is no risk of "filling up" the code cache either, as cold code is more aggressively removed as it fills up.

• -XX:+UseFastUnorderedTimeStamps This option enables the usage of fast unordered timestamps for time-related operations in the JVM.
• -XX:+UseCriticalJavaThreadPriority This option assigns a higher priority to critical Java threads, which can improve responsiveness and reduce latency in critical parts of the application.
• -XX:AllocatePrefetchStyle=3 This will instruct JVM to use code style 3 for faster memory allocation. A value of 3 enables the most aggressive prefetching. More aggressive prefetching is generally useful on newer CPUs with large caches.
• -XX:+AlwaysActAsServerClassMachine Forces the JVM to act as if it's running on a server-class machine.
• -XX:+UseTransparentHugePages As it name said.
• -XX:LargePageSizeInBytes=2M Size of huge (large) pages.
• -XX:+UseLargePages Enables the usage of large pages for the JVM's memory allocation. Large pages can improve performance by reducing Translation Lookaside Buffer misses.
• -XX:+EagerJVMCI Enables the JVM Compiler Interface as early as needed, especially with GraalVM.
• -XX:+UseStringDeduplication Reduces memory ussage by deduping identical strings. Might reduce frequent GCs.
• -XX:+UseAES -XX:+UseAESIntrinsics Enables Hardware accelerated Advanced Encryption Standart intrinsics.
• -XX:+UseFMA Enables Fused Multiply Add instructions.
• -XX:+UseLoopPredicate Enables loop predicate optimization.
• -XX:+RangeCheckElimination Enables the elimination of range checks.
• -XX:+OptimizeStringConcat Enables optimization of string concatenation in the JVM, which can lead to improved performance when working with string concatenation operations.
• -XX:+UseCompressedOops Enables the use of compressed object pointers (Oops) which reduces memory usage on 64-bit systems.
• -XX:+UseThreadPriorities Enables the use of thread priorities.
• -XX:+OmitStackTraceInFastThrow Omits the stack trace for exceptions that are frequently thrown.
• -XX:+RewriteBytecodes -XX:+RewriteFrequentPairs Improves performance by actively rewriting byte codes and its pairs.
• -XX:+UseFPUForSpilling Provides a faster temporary storage for spills, so when we need just a few additional spill slots, we can avoid tripping to L1 cache-backed stack for this.
• -XX:+UseFastStosb Enables support for fast string operations.
• -XX:+UseNewLongLShift Use optimized bitwise shift left.
• -XX:+UseVectorCmov This option enables the usage of vectorized compare and move (cmov) instructions for better performance on CPUs that support these instructions.
• -XX:+UseXMMForArrayCopy -XX:+UseXmmI2D -XX:+UseXmmI2F -XX:+UseXmmRegToRegMoveAll Uses XMM registers for Array Copy, int2double, int2float, register2register move operations
• -XX:+UseXmmLoadAndClearUpper Same as above but to load and clear upper registers.
• -XX:+EliminateLocks Improves single-threaded performance.
• -XX:+DoEscapeAnalysis Escape analysis is a technique by which the Java Hotspot Server Compiler can analyze the scope of a new object's uses and decide whether to allocate it on the Java heap.
• -XX:+AlignVector Aligns vectors for optimization.
• -XX:+OptimizeFill Enables optimization of memory fill operations.
• -XX:+EnableVectorSupport Its name suggested.
• -XX:+UseCharacterCompareIntrinsics Enables intrinsification of java.lang.Character functions.
• -XX:+UseCopySignIntrinsic Enables intrinsification of Math.copySign.
• -XX:+UseVectorStubs Optimizes vector operations.
• -XX:UseAVX=2 -XX:UseSSE=4 Enable supported AVX, SSE instructions set on x86/x64 compatible CPUs.
• -XX:+UseFastJNIAccessors Use optimized versions of GetField.
• -XX:+UseInlineCaches Use Inline Caches for virtual calls
• -XX:+SegmentedCodeCache Enables the segmented code cache, which improves code cache management.

• -Djdk.nio.maxCachedBufferSize=262144 Sets the maximum size for cached direct buffers.
• -Djdk.graal.UsePriorityInlining=true Enables priority inlining in the Graal compiler.
• -Djdk.graal.Vectorization=true Enables vectorization in the Graal compiler.
• -Djdk.graal.OptDuplication=true Enables optimization duplication in the Graal compiler.
• -Djdk.graal.DetectInvertedLoopsAsCounted=true Enables detection of inverted loops as counted loops.
• -Djdk.graal.LoopInversion=true Enables loop inversion in the Graal compiler.
• -Djdk.graal.VectorizeHashes=true Enables vectorization of hash computations.
• -Djdk.graal.VectorizeSIMD=true Enables SIMD (Single Instruction - Multiple Data) vectorization.
• -Djdk.graal.StripMineNonCountedLoops=true Enables strip mining of non-counted loops.
• -Djdk.graal.SpeculativeGuardMovement=true Enables speculative guard movement in the Graal compiler.
• -Djdk.graal.TuneInlinerExploration=1 Tunes the inliner exploration parameter in the Graal compiler.
• -Djdk.graal.LoopRotation=true Enables loop rotation in the Graal compiler.
• -Djdk.graal.UseCountedLoopSafepoints=true Enables safepoint insertion in counted loops.
• -Djdk.graal.EnterprisePartialUnroll=true Enables partial unrolling optimizations in enterprise mode.
• -Djdk.graal.CompilerConfiguration=enterprise Enables Enterprise mode optimizations in the Graal compiler.

I have made a comparison with one of my customers' servers.

Aikar's Flags and Adoptium JDK:

MeowIce's Flags and GraalVM:

With only Aikar's flags and OpenJDK: With MeowIce's flags and GraalVM:

• 10700K @ 5.00 GHz OC
• RAM DDR4 96 GB @ 3000 MT/s
• OS Proxmox 8.4.1 (Running Ubuntu 24.04.2 LTS Pterodactyl Panel)

Version used to test is leaf-1.21.8-52.jar. Both servers had 6 GB of allocated RAM and 16 dedicated threads. For monitoring and measuring data UnifiedMetrics was used.

Java Docker Image for Pterodactyl used: ghcr.io/vanes430/java:graaljdk_24-debian from https://github.com/vanes430/java

For Aikar's i used

java -Xms6G -Xmx6G -XX:+AlwaysPreTouch -XX:+DisableExplicitGC -XX:+ParallelRefProcEnabled -XX:+PerfDisableSharedMem -XX:+UnlockExperimentalVMOptions -XX:+UseG1GC -XX:G1HeapRegionSize=8M -XX:G1HeapWastePercent=5 -XX:G1MaxNewSizePercent=40 -XX:G1MixedGCCountTarget=4 -XX:G1MixedGCLiveThresholdPercent=90 -XX:G1NewSizePercent=30 -XX:G1RSetUpdatingPauseTimePercent=5 -XX:G1ReservePercent=20 -XX:InitiatingHeapOccupancyPercent=15 -XX:MaxGCPauseMillis=200 -XX:MaxTenuringThreshold=1 -XX:SurvivorRatio=32 --add-modules=jdk.incubator.vector -DLeaf.enableFMA=true -Dusing.aikars.flags=https://mcflags.emc.gs -Daikars.new.flags=true -jar server.jar nogui

And for MeowIce i used

java -Xms6G -Xmx6G --add-modules=jdk.incubator.vector -XX:+UseG1GC -XX:MaxGCPauseMillis=200 -XX:+UnlockExperimentalVMOptions -XX:+UnlockDiagnosticVMOptions -XX:+DisableExplicitGC -XX:+AlwaysPreTouch -XX:G1NewSizePercent=28 -XX:G1MaxNewSizePercent=50 -XX:G1HeapRegionSize=16M -XX:G1ReservePercent=15 -XX:G1MixedGCCountTarget=3 -XX:InitiatingHeapOccupancyPercent=20 -XX:G1MixedGCLiveThresholdPercent=90 -XX:SurvivorRatio=32 -XX:G1HeapWastePercent=5 -XX:MaxTenuringThreshold=1 -XX:+PerfDisableSharedMem -XX:G1SATBBufferEnqueueingThresholdPercent=30 -XX:G1ConcMarkStepDurationMillis=5 -XX:G1RSetUpdatingPauseTimePercent=0 -XX:+UseNUMA -XX:-DontCompileHugeMethods -XX:MaxNodeLimit=240000 -XX:NodeLimitFudgeFactor=8000 -XX:ReservedCodeCacheSize=400M -XX:NonNMethodCodeHeapSize=12M -XX:ProfiledCodeHeapSize=194M -XX:NonProfiledCodeHeapSize=194M -XX:NmethodSweepActivity=1 -XX:+UseFastUnorderedTimeStamps -XX:+UseCriticalJavaThreadPriority -XX:AllocatePrefetchStyle=3 -XX:+AlwaysActAsServerClassMachine -XX:+UseTransparentHugePages -XX:LargePageSizeInBytes=2M -XX:+UseLargePages -XX:+EagerJVMCI -XX:+UseStringDeduplication -XX:+UseAES -XX:+UseAESIntrinsics -XX:+UseFMA -XX:+UseLoopPredicate -XX:+RangeCheckElimination -XX:+OptimizeStringConcat -XX:+UseCompressedOops -XX:+UseThreadPriorities -XX:+OmitStackTraceInFastThrow -XX:+RewriteBytecodes -XX:+RewriteFrequentPairs -XX:+UseFPUForSpilling -XX:+UseFastStosb -XX:+UseNewLongLShift -XX:+UseVectorCmov -XX:+UseXMMForArrayCopy -XX:+UseXmmI2D -XX:+UseXmmI2F -XX:+UseXmmLoadAndClearUpper -XX:+UseXmmRegToRegMoveAll -XX:+EliminateLocks -XX:+DoEscapeAnalysis -XX:+AlignVector -XX:+OptimizeFill -XX:+EnableVectorSupport -XX:+UseCharacterCompareIntrinsics -XX:+UseCopySignIntrinsic -XX:+UseVectorStubs -XX:UseAVX=2 -XX:UseSSE=4 -XX:+UseFastJNIAccessors -XX:+UseInlineCaches -XX:+SegmentedCodeCache -Djdk.nio.maxCachedBufferSize=262144 -Djdk.graal.UsePriorityInlining=true -Djdk.graal.Vectorization=true -Djdk.graal.OptDuplication=true -Djdk.graal.DetectInvertedLoopsAsCounted=true -Djdk.graal.LoopInversion=true -Djdk.graal.VectorizeHashes=true -Djdk.graal.EnterprisePartialUnroll=true -Djdk.graal.VectorizeSIMD=true -Djdk.graal.StripMineNonCountedLoops=true -Djdk.graal.SpeculativeGuardMovement=true -Djdk.graal.TuneInlinerExploration=1 -Djdk.graal.LoopRotation=true -Djdk.graal.CompilerConfiguration=enterprise -DLeaf.enableFMA=true -jar server.jar nogui

So now with results

Aikar's Grafana https://snapshots.raintank.io/dashboard/snapshot/2PwtSHpB6w3BT2jStlYWAbCaKoGf2WJ1?orgId=0

MeowIce's Grafana https://snapshots.raintank.io/dashboard/snapshot/xPugteW5fyKu6Mpq4Zi8J0HgTRU4KEDF?orgId=0

Data showing all of quantile's

Aikar's: https://snapshots.raintank.io/dashboard/snapshot/KqirFpcpC4Smo6qrSjfKf6aQeee4KvJy?orgId=0

MeowIce's: https://snapshots.raintank.io/dashboard/snapshot/IpKM7ZpJOrwJTwErZucogu79Akzmx9JR?orgId=0

Some screenshots

Overall, I observed an average performance uplift of around 20 MSPT. However, the most crucial difference appeared once I reached 1k simulated players:

MeowIce’s flags consistently held around 30–35 MSPT

Aikar’s flags spiked to and held around 90–95 MSPT

I did my best to minimize variance in the testing process. Please disregard the initial segment of the data, as it isn’t relevant due to me forgetting to kill all entities (they were removed on Aikar’s setup from the star).

This oversight does not affect the main conclusion: MeowIce’s flags demonstrated a significant performance advantage over Aikar’s.

You can confirm these findings by reviewing the Grafana snapshots.

One final note: I only encountered a single G1 Old Gen event, which caused a minor spike. Aside from that, the performance was very stable and highly impressive.

• Java Flags Benchmarked
• Obydux GraalVM Flags
• Optimize MC Server on RedHat

• Dịch vụ tối ưu, fix lag máy chủ Minecraft