Memory optimizations (#505)

* Remove LocatedBlock overhead in LBL map

* Add new space-efficient block map, with thourough testing

* Drop ordering property, add full insertion test

* Add licenses

* Fix mocked platform conflicts

* Disable full block map testing for faster builds

* Re-implement BlockMap with fastutil maps

* Re-write chunk batching to be memory efficient

* Make MultiStageReorder use BlockMap

* Increase LBL load factor, fix long-pack limit detection

* Fix infinite loop in chunk batching

* Save memory in history by cleaning up MSR

* Re-implement LocatedBlockList in BlockMap

* Fix data race with BlockType lazy fields

* Make IDs ALWAYS present, only runtime-consistent. Use for memory efficiency in BlockMap

* Remap inner structure of BlockMap for smaller maps

* Remove containedBlocks fields, not very efficient

* Fix minor de-optimizing bug in stage reorder

* Make long packed y signed

* Add extended Y limit configuration option

* Add licenses

* Store 3 ints for unoptimized BV list

* Add final to BitMath

* Correct int-cast for long-packing
This commit is contained in:
Kenzie Togami
2019-08-12 05:06:40 -07:00
committed by Matthew Miller
parent ec5bc5a3b7
commit f472c20bfb
30 changed files with 2014 additions and 139 deletions

View File

@ -21,6 +21,7 @@ dependencies {
"compile"("com.google.code.findbugs:jsr305:1.3.9")
"compile"("com.google.code.gson:gson:2.8.0")
"compile"("org.slf4j:slf4j-api:1.7.26")
"compile"("it.unimi.dsi:fastutil:8.2.1")
"compileOnly"(project(":worldedit-libs:core:ap"))
"annotationProcessor"(project(":worldedit-libs:core:ap"))
@ -28,7 +29,6 @@ dependencies {
"annotationProcessor"("com.google.guava:guava:21.0")
"compileOnly"("com.google.auto.value:auto-value-annotations:${Versions.AUTO_VALUE}")
"annotationProcessor"("com.google.auto.value:auto-value:${Versions.AUTO_VALUE}")
"testCompile"("org.mockito:mockito-core:1.9.0-rc1")
}
tasks.withType<JavaCompile>().configureEach {