Non-determenism on this slide refers to allowing compiler to rearrange/reassociate/optimize floating point expression in the way that it may change numeric value of the result (but still being valid result in terms of satisfying the same mathematical formula in the source code).
Note, compiled binary will produce the same value. Different versions of compiler may produce binaries, which produce different numeric values (when --opt=fast-maths is used). Dmitry. On Tue, Feb 25, 2020 at 10:59 PM David Nadaski <[email protected]> wrote: > I've seen a slide from SIGGRAPH 2019 whose page 45 says "to increase > floating point precision/determinism, don't use --opt=fast-maths and do use > --opt=disable-fma" > My question is if doing as the slides suggests makes programs compiled w/ > ispc completely deterministic. > > -- > You received this message because you are subscribed to the Google Groups > "Intel SPMD Program Compiler Users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/ispc-users/c4b16788-b476-4f9a-b22b-37d22997e839%40googlegroups.com > <https://groups.google.com/d/msgid/ispc-users/c4b16788-b476-4f9a-b22b-37d22997e839%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "Intel SPMD Program Compiler Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/ispc-users/CACRFwugwahL-5BxWtW2NFqNpL-MnLyzS%3D0LAGrR1JjdbMUAd0g%40mail.gmail.com.
