[Bug target/65832] Inefficient vector construction

rguenth at gcc dot gnu.org gcc-bugzilla@gcc.gnu.org
Tue Apr 28 08:55:00 GMT 2015


https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65832

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |uros at gcc dot gnu.org

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
Somehow we get into very weird initial RTL generated by expand...

(insn 12 11 13 (set (reg:V2DI 101)
        (mult:V2DI (zero_extend:V2DI (vec_select:V2SI (reg:V4SI 95 [
vect_cst_.7 ])
                    (parallel [
                            (const_int 0 [0])
                            (const_int 2 [0x2])
                        ])))
            (zero_extend:V2DI (vec_select:V2SI (mem/c:V4SI (reg/f:DI 94) [1
MEM[(unsigned int *)&in]+0 S16 A256])
                    (parallel [
                            (const_int 0 [0])
                            (const_int 2 [0x2])
                        ]))))) t.c:17 -1
     (nil))

(WTF!?  As if there were no integer vector multiply with SSE2 but only DImode
ones?!)



More information about the Gcc-bugs mailing list