Currently benchmark

Deserialization micro benchamark with BenchmarkDotNet. Target object has 9 members, value are zero.

// [MessagePackObject][ProtoContract] and key, tags...
public class SerializerTarget
{
    public int MyProperty1 { get; set; }
    public int MyProperty2 { get; set; }
    public int MyProperty3 { get; set; }
    public int MyProperty4 { get; set; }
    public int MyProperty5 { get; set; }
    public int MyProperty6 { get; set; }
    public int MyProperty7 { get; set; }
    public int MyProperty8 { get; set; }
    public int MyProperty9 { get; set; }
}

[Benchmark(Baseline = true)]
public SerializerTarget IntKey()
{
    return MessagePackSerializer.Deserialize<SerializerTarget>(intKeyObject);
}

IntKey, StringKey, Typeless_IntKey, Typeless_StringKey are MessagePack for C# options. JsonNetString/JilString is deserialized from string. JsonNetByteArray/JilByteArray is deserialized from UTF8 byte[]. Deserialization is normally read from Stream. Thus, it will be restored from byte[] instead of string.

MessagePack for C# IntKey is fastest. StringKey is slower than IntKey because matching from the character string is required. All MessagePack for C# options achive zero memory allocation on deserialization process.

StringKey deserialization is not so fast

The difference with IntKey is the cost of looking up a dictionary.
Jil uses automata-based-member-name-lookups technique.
It is good, can be improve string key deserialization performance.

Here is initial design's performance of automata.

I will do inlining of IL code generation, making it faster.