AppliedE TPS Fix

By
BOLT_M4G1C

Direct ProjectE EMC integration with Applied Energistics 2.

File Details
Download Install

appliede-0.14.5.jar

R
Jun 13, 2026
248.31 KB
219
1.20.1
Forge

File Name

appliede-0.14.5.jar

Supported Versions

1.20.1

Curse Maven Snippet

Forge

implementation fg.deobf("curse.maven:appliede-tps-fix-1571176:8241719")

Curse Maven does not yet support mods that have disabled 3rd party sharing

Learn more about Curse Maven

big changes!

changed icon

added fingerprinting to suppress unchanged updates in KnowledgeService, improving update efficiency for crafting providers

changed the places in KnowledgeService that affect nodes or temporary patterns (addNode, removeNode, addTemporaryPattern, removeTemporaryPattern) to call forceUpdatePatterns() instead of the thrott…

Replaced the synchronized LinkedHashMap keyStringCache with a fastutil Object2ObjectLinkedOpenHashMap<AEItemKey,String> and added LongAdder hit/miss counters; updated getKnownItems() to use a synch…

Added a new config option saveDebounceMillis to control the coalesced async save debounce.

Converted KEY_CACHE to fastutil Object2ObjectLinkedOpenHashMap<AEKey, byte[]>, with manual synchronization, MRU reinsertion on access, and manual LRU eviction on insert.

Replaced boxed LRU Map<Long, CacheEntry> with a primitive Long2ObjectLinkedOpenHashMap<CacheEntry> and added synchronized access + manual LRU eviction (reinsert on access, evict eldest when size ex…

Added getCachedEmcPrimitive(AEItemKey) returning OptionalLong in KnowledgeService, and updated EMCStorage.insertItem / extractItem to use it (avoids boxing Long when cached EMC exists).

Replaced providers HashMap with fastutil Object2ObjectOpenHashMap<UUID, Supplier<IKnowledgeProvider>> and adjusted usages (add/get/listing) to avoid relying on HashMap-specific helpers.

Cached inputs and outputs inside TransmutationPattern to avoid repeated allocations — precompute outputs always and precompute inputs when a cached EMC value is provided (tier patterns always preco…

Added an LRU keyStringCache (AEItemKey → String) in KnowledgeService and replaced the direct key.toString() call in getKnownItems() with a computeIfAbsent lookup to reduce allocations.

Replaced HashMap-based patternCache with fastutil Object2ObjectOpenHashMap<AEItemKey, TransmutationPattern> in KnowledgeService to reduce GC and speed lookups/iteration in hot paths.

Added a single-threaded ioExecutor in KnowledgeService and a debounced scheduleSaveEmcCacheToDisk() that snapshots the primitive persistedEmc and writes it asynchronously to disk, coalescing multip…

Added a hybrid warm queue to KnowledgeService: a ConcurrentLinkedQueue for requests, a background scheduled task that deduplicates/aggregates into a final queue, and per‑tick finalization that call…

Added KEY_CACHE hit/miss counters (LongAdders) to GridInventoryEMCEntry and increment them on cached writes and misses.

Made KEY_CACHE size configurable via new keyCacheMax config option in AppliedEConfig and used it in GridInventoryEMCEntry (KEY_CACHE_MAX = AppliedEConfig.CONFIG.getKeyCacheMax()).

Replaced persistedEmc Map<String,Long> with Object2LongOpenHashMap<string> (primitive long) in KnowledgeService, and updated load/save and accessors to use primitive methods.</string>

Added KEY_CACHE for serialized AEKey bytes in GridInventoryEMCEntry, plus the warmKey(AEKey) helper.