242bd52942
Introduces unboxed (and arity-specialized) storage schemes for Atoms. It results in improvements both in memory consumption and runtime. Memory wise: instead of using an array, we now use object fields. We also enable unboxing. This cuts a good few pointers in an unboxed object. E.g. a quadruple of integers is now 64 bytes (4x8 bytes for long fields + 16 bytes for layout and constructor pointers + 16 bytes for a class header). It used to be 168 bytes (4x24 bytes for boxed Longs + 16 bytes for array header + 32 bytes for array contents + 8 bytes for constructor ptr + 16 bytes for class header), so we're saving 104 bytes a piece. In the least impressive scenarios (all-boxed fields) we're saving 8 bytes per object (saving 16 bytes for array header, using 8 bytes for the new layout field). In the most-benchmarked case (list of longs), we save 32 bytes per cons-cell. Time wise: All list-summing benchmarks observe a ~2x speedup. List generation benchmarks get ~25x speedups, probably both due to less GC activity and better allocation characteristics (only allocating one object per Cons, rather than Cons + Object[] for fields). The "map-reverse" family gets a neat 10x speedup (part of the work is reading, which is 2x faster, the other is allocating, which is now 25x faster, we end up with 10x when combined). |
||
|---|---|---|
| .. | ||
| src | ||
| README.md | ||
Enso Runtime
The Enso runtime is responsible for the actual execution of Enso code. This means that it encompasses the following functionality:
- Parsing: Taking Enso code as input and generating an AST that maintains a sophisticated set of information about the input.
- Desugaring: Reducing the user-facing Enso code into a simplified language
known as
Core. - Type Inference: Inferring the types of bindings in the user's code.
- Type Checking: Checking that the inferred and provided types for bindings match up across the codebase.
- Optimisation: Static optimisation processes to improve the performance of the user's program.
- Code Execution: Actually running the Enso code.
- Introspection Hooks: Providing hooks into the running code to allow the language server to inspect information about the code as it runs.
Truffle Nodes creation convention
All Truffle nodes that are expected to be created as part of ASTs should
implement a public, static build method for creating an instance. If the node
is DSL generated, the build method should delegate to the autogenerated
create method, so that nodes are always created with build. Such a
convention allows us to easily switch node back and forth between manual and DSL
generated implementations, without the need to change its clients.
The only exception are nodes that are never expected to be a part of an AST –
e.g. root nodes of builtin functions, for which an asFunction method should be
implemented instead.
This convention should be implemented for every node throughout this codebase – if you see one not obeying it – please fix it.