Usually it isn't about execution of programs, but analysis of their properties.
To do so you can use eg model checking, and you require a semantic definition for doing so.
I don't recall floating point stuff being hard to encode. See eg the flocq project.
Re:Semantics of C in K Framework
by tom_mellior on Monday December 14, 2020 @05:43AM
(#25415525)
What's wrong with CompCert's floating point model?
Does it fully encode all semantics? Since typically formal verifications have issues with floating point stuff (hence why we STILL don't have complete formal models of C even in 2019).
What is the performance like compared to traditional compilers?
And to me the most important, what things can we suddenly do with K that we can't easily do with a combination of other tools, since it seems that it can do a lot more but this isn't really explained all that well to an outsider IMO.