http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48258
Summary: Add VSX support for float/double vector reductions & improve float insert/extract Product: gcc Version: 4.7.0 Status: UNCONFIRMED Severity: enhancement Priority: P3 Component: target AssignedTo: meiss...@gcc.gnu.org ReportedBy: meiss...@gcc.gnu.org Host: powerpc64-linux Target: powerpc64-linux Build: powerpc64-linux Right now, the powerpc compiler does not do support for float and double vector reductions. For float in particular, the compiler does the code, and then stores the vector in the stack to extract the bottom element. With VSX it is possible to do vector reductions in fewer instructions and to eliminate the store.