Expression evaluator

Tokenise then evaluate 10,000 arithmetic expressions (mixed +, -, *, /, parens).

Compile time · median (cold)

median of 7 runs

Native Gocompiled

108 ms46.0×

Pipitbytecode VM

2.35 msbaseline

CPython 3.14bytecode VM

712 µs0.30×

PyPy 7.3tracing JIT

513 µs0.22×

Ttengobytecode VM

395 µs0.17×

Sscriggobytecode VM

unsupported in scriggo

n/a

Mmvmbytecode VM

712 µs0.30×

YyaegiAST walker

573 µs0.24×

Runner	N	Compile	Runtime	P95	Stddev	RSS	vs pipit	Status
Native Gocompiled	7	108 ms	61.9 ms	62.7 ms	600 µs	1.62 GiB	46.0×	OK
Pipitbytecode VM	7	2.35 ms	1.21 s	1.24 s	13.3 ms	534 MiB	1.00×	OK
CPython 3.14bytecode VM	7	712 µs	1.31 s	1.32 s	7.01 ms	n/a	0.30×	OK
PyPy 7.3tracing JIT	7	513 µs	258 ms	286 ms	11.8 ms	n/a	0.22×	OK
tengobytecode VM	7	395 µs	4.19 s	4.24 s	26.9 ms	1.66 GiB	0.17×	OK
scriggobytecode VM	0	n/a	n/a	n/a	n/a	n/a	n/a	unsupported
mvmbytecode VM	7	712 µs	4.21 s	4.24 s	21.7 ms	1.68 GiB	0.30×	OK
yaegiAST walker	7	573 µs	9.86 s	9.93 s	39.1 ms	1.66 GiB	0.24×	OK

Workload & symmetry rules

Workload

For each of 10,000 randomly-generated infix expressions: tokenise with a byte-loop lexer, parse via shunting-yard, evaluate over an integer stack.

Source code