Bob Nystrom
5d8ecad5a6
Add a new benchmark that measures the performance of calling foreign methods through the Wren C embedding API, specifically testing argument passing and return values with a tight loop of 10 million calls. Refactor test API classes from generic "Api" to specific names like "Call", "Returns", "Value", and "ForeignClass" to avoid naming conflicts with the new benchmark module. Update the benchmark runner to use the test_api executable for API-specific benchmarks and remove the special API test argument handling from the test runner.
The benchmarks in here attempt to faithfully implement the exact same algorithm in a few different languages. We're using Lua, Python, and Ruby for comparison here because those are all in Wren's ballpark: dynamically-typed, object-oriented, bytecode-compiled.
A bit about each benchmark:
binary_trees
This benchmark stresses object creation and garbage collection. It builds a few big, deeply nested binaries and then traverses them.
fib
This is just a simple naïve Fibonacci number calculator. It was the first benchmark I wrote when Wren supported little more than function calls and arithmetic. It isn't particularly representative of real-world code, but it does stress function call and arithmetic.
for
This microbenchmark just tests the performance of for loops. Not too useful, but i used it when implementing for in Wren to make sure it wasn't too far off the mark.
method_call
This is the most useful benchmark: it tests dynamic dispatch and polymorphism. You'll note that the main iteration loop is unrolled in all of the implementations. This is to ensure that the loop overhead itself doesn't dwarf the method call time.