Brummel
65e280bb70
Adds --alloc=<gc|bump> to ail build/run. Bump path links a 256MB no-free arena C stub instead of libgc; IR is byte-identical except for the @GC_malloc → @bump_malloc symbol swap. Bench harness times two allocation-heavy workloads (list cons/sum and balanced tree build/walk) under both modes. Numbers (RUNS=5, median of 4): bench_list_sum gc 0.141s bump 0.048s +194% bench_tree_walk gc 0.103s bump 0.041s +151% Bucket: large. ~60% of runtime is Boehm on these workloads — upper bound for any realistic program. Both fixtures hold the heap fully live, so the cost we're seeing is Boehm's allocate path itself, not collection work; that fact narrows the design space for the GC discussion. - crates/ailang-codegen: AllocStrategy enum, three callsites and the IR header parameterised. - crates/ail/src/main.rs: --alloc flag plumbed; bump runtime located + compiled on demand. - runtime/bump.c: 256MB static arena, abort-on-overflow. - examples/bench_list_sum, bench_tree_walk: accumulator-form fixtures (textbook recursive sum was constructor-blocked). - bench/run.sh: harness with Python timing helper (Arch's /usr/bin/time isn't part of the base install). No language-level changes; default --alloc=gc, all 141 workspace tests green, all 5 IR snapshots unchanged, 11 prior fixtures produce identical stdout. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
79 lines
2.7 KiB
Plaintext
79 lines
2.7 KiB
Plaintext
; Bench fixture (Bench iter): balanced-tree build + sum.
|
|
;
|
|
; Allocates a balanced binary tree of `Node value left right` cells,
|
|
; then walks it summing the value field. Distinct from
|
|
; `bench_list_sum` in two ways:
|
|
; 1. Each tree node has an additional pointer field versus a list
|
|
; cell — 32-byte alloc instead of 24-byte. The branching shape
|
|
; means the recursion structure is genuinely tree-shaped: the
|
|
; build cannot be made tail-recursive without explicit
|
|
; continuation passing, so this fixture is constrained to depths
|
|
; where the recursion stack fits.
|
|
; 2. The traversal pattern hits two children per node, exercising
|
|
; the GC's mark-phase pointer-chasing heuristics differently from
|
|
; a plain linked-list walk.
|
|
;
|
|
; Depth picked: 20 -> 2^20 - 1 = 1_048_575 nodes -> 32 MB heap usage.
|
|
; Recursion depth in `build_tree` and `sum_tree` matches `depth`,
|
|
; which fits comfortably in the default 8 MB system stack.
|
|
;
|
|
; Hardcoded multi-call form: build/sum the same tree thrice for
|
|
; signal averaging. The depths are different per call so the GC has
|
|
; to deal with three independent live-set sizes.
|
|
;
|
|
; Expected stdout (one int per line):
|
|
; depth 16: 2^16 - 1 = 65535 nodes, sum = 65535
|
|
; depth 18: 2^18 - 1 = 262143 nodes, sum = 262143
|
|
; depth 20: 2^20 - 1 = 1048575 nodes, sum = 1048575
|
|
;
|
|
; (Each node stores literal `1`; sum is therefore node count.)
|
|
|
|
(module bench_tree_walk
|
|
|
|
(data Tree
|
|
(ctor Leaf)
|
|
(ctor Node (con Int) (con Tree) (con Tree)))
|
|
|
|
(fn build_tree
|
|
(doc "Balanced binary tree of given depth, every value = 1.")
|
|
(type
|
|
(fn-type
|
|
(params (con Int))
|
|
(ret (con Tree))))
|
|
(params depth)
|
|
(body
|
|
(if (app == depth 0)
|
|
(term-ctor Tree Leaf)
|
|
(term-ctor Tree Node
|
|
1
|
|
(app build_tree (app - depth 1))
|
|
(app build_tree (app - depth 1))))))
|
|
|
|
(fn sum_tree
|
|
(doc "Sum every Node value via match recursion. Constructor-blocked: not tail-recursive, but recursion depth = tree depth so fits.")
|
|
(type
|
|
(fn-type
|
|
(params (con Tree))
|
|
(ret (con Int))))
|
|
(params t)
|
|
(body
|
|
(match t
|
|
(case (pat-ctor Leaf) 0)
|
|
(case (pat-ctor Node v l r)
|
|
(app + v (app + (app sum_tree l) (app sum_tree r)))))))
|
|
|
|
(fn run_one
|
|
(doc "Build a tree of given depth, sum it, print the sum.")
|
|
(type (fn-type (params (con Int)) (ret (con Unit)) (effects IO)))
|
|
(params depth)
|
|
(body
|
|
(do io/print_int (app sum_tree (app build_tree depth)))))
|
|
|
|
(fn main
|
|
(type (fn-type (params) (ret (con Unit)) (effects IO)))
|
|
(params)
|
|
(body
|
|
(seq (app run_one 16)
|
|
(seq (app run_one 18)
|
|
(app run_one 20))))))
|