Fission Benchmark

Compiler × opt variants

Break down semantic pass rates by compiler_variant (e.g. gcc -O0, gcc-m32 -O2). Useful for opt cliffs and ISA width (m32) triage.

← Overview · Per-function →

Statusofficial · stale · 36dCorpusdevFunctions36Decompilers2Contractrelease-baseline-v1 · 216 subjectsMeasured8/20/2026, 2:10:22 AM UTCBuildv0.2.2 @ c9d82f27

Cross-variant semantic means

Extension pivot — how each decompiler behaves under compiler and opt changes. Ranking still uses overall semantic on Overview.

Extension — semantic pass rate by compiler × optimization. Not a ranking axis; use to spot opt-level regressions. Each variant uses a shared subject denominator across decompilers.

Decompiler	Variant	Compiler	Opt	Observed / shared	Shared semantic mean	Perfect
fission	`clang -O0`	clang	-O0	36 / 36	73.2%	22
fission	`clang -O2`	clang	-O2	33 / 36	37.1%	8
fission	`gcc -O0`	gcc	-O0	36 / 36	79.5%	23
fission	`gcc -O2`	gcc	-O2	36 / 36	62.6%	15
fission	`gcc-m32 -O0`	gcc-m32	-O0	36 / 36	73.0%	22
fission	`gcc-m32 -O2`	gcc-m32	-O2	36 / 36	59.1%	15
ghidra	`clang -O0`	clang	-O0	36 / 36	75.0%	27
ghidra	`clang -O2`	clang	-O2	36 / 36	80.6%	29
ghidra	`gcc -O0`	gcc	-O0	36 / 36	0.0%	0
ghidra	`gcc -O2`	gcc	-O2	35 / 36	91.7%	33
ghidra	`gcc-m32 -O0`	gcc-m32	-O0	36 / 36	94.4%	34
ghidra	`gcc-m32 -O2`	gcc-m32	-O2	36 / 36	94.4%	34