Skip to content

Latest commit

 

History

History
 
 

README.md

Benchmark Suites

Three benchmark suites measure Dis VM and cross-language performance.

v1: JIT vs Interpreter (6 benchmarks)

Source: appl/cmd/jitbench.b Benchmarks: Integer Arithmetic, Array Access, Function Calls, Fibonacci, Sieve of Eratosthenes, Nested Loops

bash benchmarks/bench-jit.sh v1 3

v2: JIT vs Interpreter (26 benchmarks, 9 categories)

Source: appl/cmd/jitbench2.b Categories: Integer ALU, Branch & Control, Memory Access, Function Calls, Big (64-bit), Byte Ops, List Ops, Mixed Workloads, Type Conversions

bash benchmarks/bench-jit.sh v2 3

v3: Cross-Language Comparison

Sources: jitbench.c, jitbench.go, JITBench.java, plus Limbo v1 Contestants: C -O0, C -O2, Go, Java (HotSpot), Limbo JIT, Limbo Interpreter

Same 6 benchmarks as v1, ported to each language with matched parameters and 64-bit integer types.

bash benchmarks/run-comparison.sh 3

Gracefully skips contestants whose toolchains aren't available (no Go, no Java, no emulator — each is optional).

Results

Full per-platform data in docs/BENCHMARKS.md.