x86inc: Set program_name outside of x86inc.asm

[ffmpeg] / libavutil / x86 / x86util.asm

diff --git a/libavutil/x86/x86util.asm b/libavutil/x86/x86util.asm
index ca0041acf7aafeace866d270bbe9edaf1585d5aa..cf11183e713b36a206033313dc991ef541c8a33f 100644 (file)
--- a/libavutil/x86/x86util.asm
+++ b/libavutil/x86/x86util.asm
@@ -23,6 +23,7 @@
 ;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 ;******************************************************************************
 
+%define program_name ff
 %define cpuflags_mmxext cpuflags_mmx2
 
 %include "libavutil/x86/x86inc.asm"
@@ -145,13 +146,21 @@
 %endif
 %endmacro
 
-; PABSW macros assume %1 != %2, while ABS1/2 macros work in-place
-%macro PABSW_MMX 2
+; PABSW macro assumes %1 != %2, while ABS1/2 macros work in-place
+%macro PABSW 2
+%if cpuflag(ssse3)
+    pabsw      %1, %2
+%elif cpuflag(mmxext)
+    pxor    %1, %1
+    psubw   %1, %2
+    pmaxsw  %1, %2
+%else
     pxor       %1, %1
     pcmpgtw    %1, %2
     pxor       %2, %1
     psubw      %2, %1
     SWAP       %1, %2
+%endif
 %endmacro
 
 %macro PSIGNW_MMX 2
@@ -159,16 +168,6 @@
     psubw      %1, %2
 %endmacro
 
-%macro PABSW_MMXEXT 2
-    pxor    %1, %1
-    psubw   %1, %2
-    pmaxsw  %1, %2
-%endmacro
-
-%macro PABSW_SSSE3 2
-    pabsw      %1, %2
-%endmacro
-
 %macro PSIGNW_SSSE3 2
     psignw     %1, %2
 %endmacro
@@ -282,7 +281,14 @@
 %endif
 %endmacro
 
-%macro PALIGNR_MMX 4-5 ; [dst,] src1, src2, imm, tmp
+%macro PALIGNR 4-5
+%if cpuflag(ssse3)
+%if %0==5
+    palignr %1, %2, %3, %4
+%else
+    palignr %1, %2, %3
+%endif
+%elif cpuflag(mmx) ; [dst,] src1, src2, imm, tmp
     %define %%dst %1
 %if %0==5
 %ifnidn %1, %2
@@ -301,13 +307,6 @@
     psrldq  %4, %3
 %endif
     por     %%dst, %4
-%endmacro
-
-%macro PALIGNR_SSSE3 4-5
-%if %0==5
-    palignr %1, %2, %3, %4
-%else
-    palignr %1, %2, %3
 %endif
 %endmacro
 
@@ -319,6 +318,18 @@
     %endif
 %endmacro
 
+%macro PSWAPD 2
+%if cpuflag(mmxext)
+    pshufw    %1, %2, q1032
+%elif cpuflag(3dnowext)
+    pswapd    %1, %2
+%elif cpuflag(3dnow)
+    movq      %1, %2
+    psrlq     %1, 32
+    punpckldq %1, %2
+%endif
+%endmacro
+
 %macro DEINTB 5 ; mask, reg1, mask, reg2, optional src to fill masks from
 %ifnum %5
     pand   m%3, m%5, m%4 ; src .. y6 .. y4