Which rule of MISRA C:2012 matches rule 13.3 of MISRA C: 2004

[misra-c]

Floating-point arithmetic is inherently implementation-defined, which is covered by:

Directive 1.1 Any implementation-defned behaviour on which the output of the
program depends shall be documented and understood

The https://misra.org.uk/forum/viewtopic.php?t=1256 document on MISRA-C:2004 to MISRA C:2012 mappings states

Focussing on one aspect of floating-point arithmetic (as in MISRA C:2004 Rule 13.3) attached too much importance to that aspect and might give users a false sense of confidence.