[FFmpeg-cvslog] avcodec/x86/vp3dsp: Remove obsolete MMX functions

[Andreas Rheinhardt]

ffmpeg | branch: master | Andreas Rheinhardt <andreas.rheinhardt at outlook.com> | Mon Jun 13 16:57:39 2022 +0200| [eefec0663406d7c2749a280f5244caaacb069c60] | committer: Andreas Rheinhardt

avcodec/x86/vp3dsp: Remove obsolete MMX functions

The only system which benefit from these are truely ancient
32bit x86s as all other systems use at least the SSE2 versions
(this includes all x64 cpus (which is why this code is restricted
to x86-32)).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt at outlook.com>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=eefec0663406d7c2749a280f5244caaacb069c60
---

 libavcodec/x86/vp3dsp.asm    | 62 --------------------------------------------
 libavcodec/x86/vp3dsp_init.c |  7 -----
 2 files changed, 69 deletions(-)

diff --git a/libavcodec/x86/vp3dsp.asm b/libavcodec/x86/vp3dsp.asm
index d88d5a1edf..005ecbc9a0 100644
--- a/libavcodec/x86/vp3dsp.asm
+++ b/libavcodec/x86/vp3dsp.asm
@@ -571,40 +571,25 @@ cglobal vp3_idct_put, 3, 4, 9
     mova          m1, [r2+mmsize*2+%%i]
     mova          m2, [r2+mmsize*4+%%i]
     mova          m3, [r2+mmsize*6+%%i]
-%if mmsize == 8
-    packsswb      m0, [r2+mmsize*8+%%i]
-    packsswb      m1, [r2+mmsize*10+%%i]
-    packsswb      m2, [r2+mmsize*12+%%i]
-    packsswb      m3, [r2+mmsize*14+%%i]
-%else
     packsswb      m0, [r2+mmsize*1+%%i]
     packsswb      m1, [r2+mmsize*3+%%i]
     packsswb      m2, [r2+mmsize*5+%%i]
     packsswb      m3, [r2+mmsize*7+%%i]
-%endif
     paddb         m0, m4
     paddb         m1, m4
     paddb         m2, m4
     paddb         m3, m4
     movq   [r0     ], m0
-%if mmsize == 8
-    movq   [r0+r1  ], m1
-    movq   [r0+r1*2], m2
-    movq   [r0+r3  ], m3
-%else
     movhps [r0+r1  ], m0
     movq   [r0+r1*2], m1
     movhps [r0+r3  ], m1
-%endif
 %if %%i == 0
     lea           r0, [r0+r1*4]
 %endif
-%if mmsize == 16
     movq   [r0     ], m2
     movhps [r0+r1  ], m2
     movq   [r0+r1*2], m3
     movhps [r0+r3  ], m3
-%endif
 %assign %%i %%i+8
 %endrep
 
@@ -621,7 +606,6 @@ cglobal vp3_idct_add, 3, 4, 9
 
     lea           r3, [r1*3]
     pxor          m4, m4
-%if mmsize == 16
 %assign %%i 0
 %rep 2
     movq          m0, [r0]
@@ -647,47 +631,6 @@ cglobal vp3_idct_add, 3, 4, 9
 %endif
 %assign %%i %%i+64
 %endrep
-%else
-%assign %%i 0
-%rep 2
-    movq          m0, [r0]
-    movq          m1, [r0+r1]
-    movq          m2, [r0+r1*2]
-    movq          m3, [r0+r3]
-    movq          m5, m0
-    movq          m6, m1
-    movq          m7, m2
-    punpcklbw     m0, m4
-    punpcklbw     m1, m4
-    punpcklbw     m2, m4
-    punpckhbw     m5, m4
-    punpckhbw     m6, m4
-    punpckhbw     m7, m4
-    paddsw        m0, [r2+ 0+%%i]
-    paddsw        m1, [r2+16+%%i]
-    paddsw        m2, [r2+32+%%i]
-    paddsw        m5, [r2+64+%%i]
-    paddsw        m6, [r2+80+%%i]
-    paddsw        m7, [r2+96+%%i]
-    packuswb      m0, m5
-    movq          m5, m3
-    punpcklbw     m3, m4
-    punpckhbw     m5, m4
-    packuswb      m1, m6
-    paddsw        m3, [r2+48+%%i]
-    paddsw        m5, [r2+112+%%i]
-    packuswb      m2, m7
-    packuswb      m3, m5
-    movq   [r0     ], m0
-    movq   [r0+r1  ], m1
-    movq   [r0+r1*2], m2
-    movq   [r0+r3  ], m3
-%if %%i == 0
-    lea           r0, [r0+r1*4]
-%endif
-%assign %%i %%i+8
-%endrep
-%endif
 %assign %%i 0
 %rep 128/mmsize
     mova    [r2+%%i], m4
@@ -696,11 +639,6 @@ cglobal vp3_idct_add, 3, 4, 9
     RET
 %endmacro
 
-%if ARCH_X86_32
-INIT_MMX mmx
-vp3_idct_funcs
-%endif
-
 INIT_XMM sse2
 vp3_idct_funcs
 
diff --git a/libavcodec/x86/vp3dsp_init.c b/libavcodec/x86/vp3dsp_init.c
index ba47e1c6cd..f54fa57b3e 100644
--- a/libavcodec/x86/vp3dsp_init.c
+++ b/libavcodec/x86/vp3dsp_init.c
@@ -26,9 +26,6 @@
 #include "libavcodec/avcodec.h"
 #include "libavcodec/vp3dsp.h"
 
-void ff_vp3_idct_put_mmx(uint8_t *dest, ptrdiff_t stride, int16_t *block);
-void ff_vp3_idct_add_mmx(uint8_t *dest, ptrdiff_t stride, int16_t *block);
-
 void ff_vp3_idct_put_sse2(uint8_t *dest, ptrdiff_t stride, int16_t *block);
 void ff_vp3_idct_add_sse2(uint8_t *dest, ptrdiff_t stride, int16_t *block);
 
@@ -49,10 +46,6 @@ av_cold void ff_vp3dsp_init_x86(VP3DSPContext *c, int flags)
 
     if (EXTERNAL_MMX(cpu_flags)) {
         c->put_no_rnd_pixels_l2 = ff_put_vp_no_rnd_pixels8_l2_mmx;
-#if ARCH_X86_32
-        c->idct_put  = ff_vp3_idct_put_mmx;
-        c->idct_add  = ff_vp3_idct_add_mmx;
-#endif
     }
 
     if (EXTERNAL_MMXEXT(cpu_flags)) {