CekState record optimization + profiling: 1.5x speedup, root cause found

Transpiler (transpiler.sx): detects CEK state dict literals (5 fields:
control/env/kont/phase/value) and emits CekState OCaml record instead
of Dict(Hashtbl). Eliminates 200K Hashtbl allocations per page.

Bootstrapper: skip stdlib.sx (functions already registered as OCaml
primitives). Only transpile evaluator.sx.

Runtime: get_val handles CekState with direct field access. type_of
returns "dict" for CekState (backward compat).

Profiling results (root cause of slowness):
  Pure eval: OCaml 1.6x FASTER than Python (expected)
  Aser: OCaml 28x SLOWER than Python (unexpected!)

Root cause: Python has a native optimized aser. OCaml runs the SX
adapter-sx.sx through the CEK machine — each aserCall is ~50 CEK
steps with closures, scope operations, string building.

Fix needed: native OCaml aser (like Python's), not SX adapter
through CEK machine.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-03-19 17:40:34 +00:00
parent 8707f21ca2
commit c16142d14c
6 changed files with 64 additions and 16 deletions

View File

@@ -94,9 +94,10 @@ def compile_spec_to_ml(spec_dir: str | None = None) -> str:
trampoline(eval_expr(expr, env))
# Spec files to transpile (in dependency order)
# stdlib.sx functions are already registered as OCaml primitives —
# only the evaluator needs transpilation.
sx_files = [
("evaluator.sx", "evaluator (frames + eval + CEK)"),
("stdlib.sx", "stdlib (library functions from former primitives)"),
]
parts = [PREAMBLE]
@@ -111,8 +112,14 @@ def compile_spec_to_ml(spec_dir: str | None = None) -> str:
src = f.read()
defines = extract_defines(src)
# Skip defines provided by preamble or fixups
skip = {"trampoline"}
# Skip defines provided by preamble, fixups, or already-registered primitives
# Skip: preamble-provided, math primitives, and stdlib functions
# that use loop/named-let (transpiler can't handle those yet)
skip = {"trampoline", "ceil", "floor", "round", "abs", "min", "max",
"debug-log", "debug_log", "range", "chunk-every", "zip-pairs",
"string-contains?", "starts-with?", "ends-with?",
"string-replace", "trim", "split", "index-of",
"pad-left", "pad-right", "char-at", "substring"}
defines = [(n, e) for n, e in defines if n not in skip]
# Deduplicate — keep last definition for each name (CEK overrides tree-walk)