Quote from abstract: " For many applications, the best possible code is conditionally correct: the
optimized kernel is equal to the code that it replaces only
under certain preconditions on the kernel’s inputs. The main
technical challenge in producing conditionally correct opti-
mizations is in obtaining non-trivial and useful conditions
and proving conditional equivalence formally in the pres-
ence of loops. We combine abstract interpretation, decision
procedures, and testing to yield a verification strategy that
can address both of these problems. This approach yields
a superoptimizer for x86 that in our experiments produces
binaries that are often multiple times faster than those pro-
duced by production compilers"
This paper from the same project was also cool: https://raw.githubusercontent.com/StanfordPL/stoke/develop/d...
Quote from abstract: " For many applications, the best possible code is conditionally correct: the optimized kernel is equal to the code that it replaces only under certain preconditions on the kernel’s inputs. The main technical challenge in producing conditionally correct opti- mizations is in obtaining non-trivial and useful conditions and proving conditional equivalence formally in the pres- ence of loops. We combine abstract interpretation, decision procedures, and testing to yield a verification strategy that can address both of these problems. This approach yields a superoptimizer for x86 that in our experiments produces binaries that are often multiple times faster than those pro- duced by production compilers"