A [early prototype of] scala library for making memory-optimized data structures.

General idea:

• pack all members of case-class like objects to a single ByteArray buffer
• use a single shared buffer for all elements in a sequence
• supports String, Float, Double, Int, Long types
• supports Option[T], Map[K,V], Seq[T]
• can do shapeless-based codec auto-derivation for custom case classes
• can be extended for custom types

    import io.findify.scalapacked.PackedSeq

    // A simple case class to pack
    case class HelloPacked(i: Int, l: Long, s: String)
    
    // codecs for default scala types
    import io.findify.scalapacked.codec._
    // shapeless-based codec auto-deriver for case classes
    import io.findify.scalapacked.codec.generic._
    
    // a sequence of 1k objects
    val list = PackedSeq((0 to 1000).map(i => HelloPacked(i, i.toLong, i.toString)): _*)
    
    // use it as a typical scala collection
    list.filter(_.i % 10 == 0)

    import io.findify.scalapacked.pool.MemoryPool
    import io.findify.scalapacked.types.Codec
    import io.findify.scalapacked.immutable.PackedMap
    
    // custom type codec
    implicit val byteCodec = new Codec[Byte] {
      // read byte from buffer
      override def read(buffer: MemoryPool, offset: Int): Byte = buffer.readByte(offset)
      // packed object size in buffer
      override def size(buffer: MemoryPool, offset: Int): Int = 1
      // object size
      override def size(item: Byte): Int = 1
      // write object to buffer and return it's offset
      override def write(value: Byte, buffer: MemoryPool): Int = buffer.writeByte(value)
    }
    
    // case class with cutsom type
    case class NestedFoo(b: Byte)
    case class RootFoo(n: NestedFoo)

    // codecs for default scala types
    import io.findify.scalapacked.codec._
    // shapeless-based codec auto-deriver for case classes
    import io.findify.scalapacked.codec.generic._

    // build a packed map
    val map = PackedMap( (0 to 100).map(i => s"key$i" -> RootFoo(NestedFoo(i.toByte))): _*)

    // use it as a regular one
    map.get("key75")    

RAM usage is 5x-10x lower compared to generic collections of case classes:

List[Int]         = 40 byte/item
PackedList[Int]   = 4  byte/item (10.02% of original)

List[String]      = 72 byte/item
PackedSeq[String] = 7  byte/item (10.97% of original)

Map[String,String]              = 167 byte/item
PackedMap[String,String]        = 29 byte/item (17.38% of original)

Map[String, ComplexClass]       = 222 byte/item
PackedMap[String, ComplexClass] = 45 byte/item (20.24% of original)

CPU usage depends on workload, but in general it is 3x-5x slower than scala native collections:

For Map vs PackedMap:

Benchmark                             (mapType)  Mode  Cnt       Score      Error  Units
PackedMapBenchmark.build1000                Map  avgt   10  117845.527 ± 1101.192  ns/op
PackedMapBenchmark.build1000          PackedMap  avgt   10  213333.266 ± 3285.303  ns/op
PackedMapBenchmark.lookupExisting           Map  avgt   10      32.682 ±    0.509  ns/op
PackedMapBenchmark.lookupExisting     PackedMap  avgt   10      80.360 ±    0.487  ns/op
PackedMapBenchmark.lookupNonExisting        Map  avgt   10      25.683 ±    0.309  ns/op
PackedMapBenchmark.lookupNonExisting  PackedMap  avgt   10      15.055 ±    0.383  ns/op

For List vs PackedSeq:

Benchmark                   (listType)  Mode  Cnt     Score     Error  Units
PackedSeqBenchmark.filter         List  avgt   10  4014.360 ±  49.048  ns/op
PackedSeqBenchmark.filter    PackedSeq  avgt   10  8465.392 ± 117.966  ns/op
PackedSeqBenchmark.foreach        List  avgt   10  2250.216 ±  10.079  ns/op
PackedSeqBenchmark.foreach   PackedSeq  avgt   10  5346.679 ±  35.374  ns/op
PackedSeqBenchmark.head           List  avgt   10     4.213 ±   0.060  ns/op
PackedSeqBenchmark.head      PackedSeq  avgt   10   211.507 ±   1.748  ns/op

The MIT License (MIT)

Copyright (c) 2017 Findify AB

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.