https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120313
Andrew Pinski <pinskia at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Target| |x86_64 Keywords| |ice-on-invalid-code --- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> --- __asm__("fmul add %0, %2, %0\nfmul add %1, %3, %1" :"=f" (a[j]),"=f" (b[j]) : "f" (b[j]), "f" (a[j])); is invalid inline-asm for x86: From https://gcc.gnu.org/onlinedocs/gcc-15.1.0/gcc/Extended-Asm.html#x86-Floating-Point-asm-Operands Output operands must specifically indicate which register an output appears in after an asm. â=fâ is not allowed: the operand constraints must select a class with a single register.