[Bug rtl-optimization/19680] sub-optimial register allocation with sse

tbptbp at gmail dot com gcc-bugzilla@gcc.gnu.org
Mon Jan 31 23:28:00 GMT 2005


------- Additional Comments From tbptbp at gmail dot com  2005-01-31 23:28 -------
Wow! We got a winner. 15.8 fps with -fno-gcse, inlining and only d-19680-3.

  402680:       66 0f 6f d1             movdqa %xmm1,%xmm2
..
  402688:       66 0f db 50 30          pand   0x30(%eax),%xmm2
  40268d:       66 0f 6e 41 28          movd   0x28(%ecx),%xmm0
  402692:       66 0f 70 c0 00          pshufd $0x0,%xmm0,%xmm0
  402697:       66 0f df c8             pandn  %xmm0,%xmm1
  40269b:       66 0f eb ca             por    %xmm2,%xmm1
  40269f:       0f 29 48 30             movaps %xmm1,0x30(%eax)
That's the final integer update. Perfect.

Want me to try that champ in conjunction with d-19680-1?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19680



More information about the Gcc-bugs mailing list