* ALL: new module API. Makes a few things a lot simpler, and we gain

[vlc] / plugins / imdct / ac3_imdct_sse.c

diff --git a/plugins/imdct/ac3_imdct_sse.c b/plugins/imdct/ac3_imdct_sse.c
index f55817611db99e8ccfb70d300682f16a984f2ded..06000970dc9e554dd9ac236e7d1c4ced4264cd31 100644 (file)
--- a/plugins/imdct/ac3_imdct_sse.c
+++ b/plugins/imdct/ac3_imdct_sse.c
@@ -2,7 +2,7 @@
  * ac3_imdct_sse.c: accelerated SSE ac3 DCT
  *****************************************************************************
  * Copyright (C) 1999, 2000 VideoLAN
- * $Id: ac3_imdct_sse.c,v 1.5 2001/07/26 20:00:33 reno Exp $
+ * $Id: ac3_imdct_sse.c,v 1.13 2002/07/31 20:56:51 sam Exp $
  *
  * Authors: Renaud Dartus <reno@videolan.org>
  *          Aaron Holtzman <aholtzma@engr.uvic.ca>
@@ -22,21 +22,13 @@
  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111, USA.
  *****************************************************************************/
 
-#define MODULE_NAME imdctsse
-#include "modules_inner.h"
-
 /*****************************************************************************
  * Preamble
  *****************************************************************************/
-#include "defs.h"
-
 #include <math.h>
 #include <stdio.h>
 
-#include "config.h"
-#include "common.h"
-#include "threads.h"
-#include "mtime.h"
+#include <vlc/vlc.h>
 
 #include "ac3_imdct.h"
 #include "ac3_imdct_common.h"
@@ -46,16 +38,18 @@
 #   define M_PI 3.14159265358979323846
 #endif
 
-void _M( fft_64p )  ( complex_t *x );
-void _M( fft_128p ) ( complex_t *a );
-
-static void imdct512_pre_ifft_twiddle_sse (const int *pmt, complex_t *buf, float *data, float *xcos_sin_sse);
-static void imdct512_post_ifft_twiddle_sse (complex_t *buf, float *xcos_sin_sse);
-static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *window_prt, float *delay_prt);
-static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, float *window_prt, float *delay_prt);
+void E_( fft_64p )  ( complex_t *x );
+void E_( fft_128p ) ( complex_t *a );
 
+static void imdct512_pre_ifft_twiddle_sse  ( const int *, complex_t *,
+                                             float *, float * );
+static void imdct512_post_ifft_twiddle_sse ( complex_t *, float * );
+static void imdct512_window_delay_sse      ( complex_t *, float *,
+                                             float *, float * );
+static void imdct512_window_delay_nol_sse  ( complex_t *, float *,
+                                             float *, float * );
 
-void _M( imdct_init ) (imdct_t * p_imdct)
+void E_( imdct_init ) (imdct_t * p_imdct)
 {
     int i;
     float scale = 181.019;
@@ -71,21 +65,23 @@ void _M( imdct_init ) (imdct_t * p_imdct)
     }
 }
 
-void _M( imdct_do_512 ) (imdct_t * p_imdct, float data[], float delay[])
+void E_( imdct_do_512 ) (imdct_t * p_imdct, float data[], float delay[])
 {
-    imdct512_pre_ifft_twiddle_sse (pm128, p_imdct->buf, data, p_imdct->xcos_sin_sse);
-    _M( fft_128p ) ( p_imdct->buf );
-    imdct512_post_ifft_twiddle_sse (p_imdct->buf, p_imdct->xcos_sin_sse);
-    imdct512_window_delay_sse (p_imdct->buf, data, window, delay);
+    imdct512_pre_ifft_twiddle_sse( pm128, p_imdct->buf, data,
+                                   p_imdct->xcos_sin_sse );
+    E_( fft_128p ) ( p_imdct->buf );
+    imdct512_post_ifft_twiddle_sse( p_imdct->buf, p_imdct->xcos_sin_sse );
+    imdct512_window_delay_sse( p_imdct->buf, data, window, delay );
 }
 
 
-void _M( imdct_do_512_nol ) (imdct_t * p_imdct, float data[], float delay[])
+void E_( imdct_do_512_nol ) (imdct_t * p_imdct, float data[], float delay[])
 {
-    imdct512_pre_ifft_twiddle_sse (pm128, p_imdct->buf, data, p_imdct->xcos_sin_sse);  
-    _M( fft_128p ) ( p_imdct->buf );
-    imdct512_post_ifft_twiddle_sse (p_imdct->buf, p_imdct->xcos_sin_sse);
-    imdct512_window_delay_nol_sse (p_imdct->buf, data, window, delay);
+    imdct512_pre_ifft_twiddle_sse( pm128, p_imdct->buf, data,
+                                   p_imdct->xcos_sin_sse );
+    E_( fft_128p ) ( p_imdct->buf );
+    imdct512_post_ifft_twiddle_sse( p_imdct->buf, p_imdct->xcos_sin_sse );
+    imdct512_window_delay_nol_sse( p_imdct->buf, data, window, delay );
 }
 
 static void imdct512_pre_ifft_twiddle_sse (const int *pmt, complex_t *buf, float *data, float *xcos_sin_sse)
@@ -107,7 +103,7 @@ static void imdct512_pre_ifft_twiddle_sse (const int *pmt, complex_t *buf, float
     "movl $64, -4(%%ebp)\n"
     
     ".align 16\n"
-".loop:\n"
+"0:\n"
     "movl  (%%eax), %%esi\n"
     "movl 4(%%eax), %%edi\n"
     "movss (%%ecx, %%esi, 8), %%xmm1\n" /* 2j */
@@ -139,7 +135,7 @@ static void imdct512_pre_ifft_twiddle_sse (const int *pmt, complex_t *buf, float
     
     "movaps  %%xmm0, -16(%%ebx)\n"
     "decl    -4(%%ebp)\n"
-    "jnz     .loop\n"
+    "jnz     0b\n"
 
     "popl %%esi\n"
     "popl %%edi\n"
@@ -163,7 +159,7 @@ static void imdct512_post_ifft_twiddle_sse (complex_t *buf, float *xcos_sin_sse)
     "movl  $32, %%ebx\n"               /* loop counter */
 
     ".align 16\n"
-".loop1:\n"
+"0:\n"
     "movaps (%%eax), %%xmm0\n"         /*  im1 | re1 | im0 | re0 */
 
     "movaps (%%ecx), %%xmm2\n"         /* -c | -s | -s | c */
@@ -214,7 +210,7 @@ static void imdct512_post_ifft_twiddle_sse (complex_t *buf, float *xcos_sin_sse)
     "addl $64, %%ecx\n"
     "addl $32, %%eax\n"
     "decl %%ebx\n"
-    "jnz .loop1\n"
+    "jnz 0b\n"
 
     "popl %%ebx\n"
     : "=a" (buf)
@@ -249,7 +245,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps %%xmm2, %%xmm0\n"      /* 0.0 | im1 | 0.0 | im0 */
     "movlhps %%xmm3, %%xmm1\n"      /* 0.0 | re1 | 0.0 | re0 */
 
-    "movups (%%edx), %%xmm4\n"      /* w3 | w2 | w1 | w0 */
+    "movaps (%%edx), %%xmm4\n"      /* w3 | w2 | w1 | w0 */
     "movaps (%%ecx), %%xmm5\n"      /* d3 | d2 | d1 | d0 */
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* re1 | 0.0 | re0 | 0.0 */
 
@@ -263,7 +259,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps   %%xmm3, %%xmm2\n"    /* 0.0 | re3 | 0.0 | re2 */
     "addps  %%xmm5, %%xmm0\n"
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* re3 | 0.0 | re2 | 0.0 */
-    "movups 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
+    "movaps 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
     "movaps 16(%%ecx), %%xmm5\n"    /* d7 | d6 | d5 | d4 */
     "subps  %%xmm2, %%xmm6\n"       /* -re3 | im3 | -re2 | im2 */
     "addl   $32, %%edx\n"
@@ -292,7 +288,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps %%xmm2, %%xmm0\n"  /* 0.0 | re1 | 0.0 | re0 */
     "movlhps %%xmm3, %%xmm1\n"  /* 0.0 | im1 | 0.0 | im1 */
 
-    "movups (%%edx), %%xmm4\n"  /* w3 | w2 | w1 | w0 */
+    "movaps (%%edx), %%xmm4\n"  /* w3 | w2 | w1 | w0 */
     "movaps (%%ecx), %%xmm5\n"  /* d3 | d2 | d1 | d0 */
 
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* im1 | 0.0 | im0 | 0.0 */
@@ -305,7 +301,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps %%xmm3, %%xmm2\n"      /* 0.0 | im3 | 0.0 | im2 */
     "mulps   %%xmm4, %%xmm0\n"
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* im3 | 0.0 | im2 | 0.0 */
-    "movups 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
+    "movaps 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
     "addl   $32, %%esi\n"
     "subps  %%xmm2, %%xmm6\n"       /* -im3 | re3 | -im2 | re2 */
     "addps  %%xmm5, %%xmm0\n"
@@ -336,7 +332,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps %%xmm2, %%xmm0\n"      /* 0.0 | re1 | 0.0 | re0 */
     "movlhps %%xmm3, %%xmm1\n"      /* 0.0 | im1 | 0.0 | im0 */
 
-    "movups -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
+    "movaps -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* im1 | 0.0 | im0 | 0.0 */
     "movss  16(%%esi), %%xmm6\n"    /* re2 */
     "movss  24(%%esi), %%xmm7\n"    /* re3 */
@@ -347,7 +343,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps %%xmm7, %%xmm6\n"      /* 0.0 | re3 | 0.0 | re2 */
     "movlhps %%xmm3, %%xmm2\n"      /* 0.0 | im3 | 0.0 | im2 */
     "mulps   %%xmm4, %%xmm0\n"
-    "movups (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
+    "movaps (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* im3 | 0.0 | im2 | 0.0 */
     "movaps %%xmm0, (%%ecx)\n"
     "addl   $32, %%esi\n"
@@ -373,7 +369,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps %%xmm2, %%xmm0\n"      /* 0.0 | im1 | 0.0 | im0 */
     "movlhps %%xmm3, %%xmm1\n"      /* 0.0 | re1 | 0.0 | re0 */
 
-    "movups -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
+    "movaps -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* re1 | 0.0 | re0 | 0.0 */
     "movss  16(%%esi), %%xmm6\n"    /* im2 */
     "movss  24(%%esi), %%xmm7\n"    /* im3 */
@@ -384,7 +380,7 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     "movlhps %%xmm7, %%xmm6\n"      /* 0.0 | im3 | 0.0 | im2 */
     "movlhps %%xmm3, %%xmm2\n"      /* 0.0 | re3 | 0.0 | re2 */
     "mulps   %%xmm4, %%xmm1\n"
-    "movups (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
+    "movaps (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* re3 | 0.0 | re2 | 0.0 */
     "movaps %%xmm1, (%%ecx)\n"
     "addl   $32, %%esi\n"
@@ -408,7 +404,8 @@ static void imdct512_window_delay_sse (complex_t *buf, float *data_ptr, float *w
     
 }
 
-static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, float *window_prt, float *delay_prt)
+static void imdct512_window_delay_nol_sse( complex_t *buf, float *data_ptr,
+                                           float *window_prt, float *delay_prt )
 {
     __asm__ __volatile__ (
     ".align 16\n"
@@ -436,7 +433,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm2, %%xmm0\n"      /* 0.0 | im1 | 0.0 | im0 */
     "movlhps %%xmm3, %%xmm1\n"      /* 0.0 | re1 | 0.0 | re0 */
 
-    "movups (%%edx), %%xmm4\n"      /* w3 | w2 | w1 | w0 */
+    "movaps (%%edx), %%xmm4\n"      /* w3 | w2 | w1 | w0 */
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* re1 | 0.0 | re0 | 0.0 */
 
     "movss  16(%%esi), %%xmm6\n"    /* im2 */
@@ -448,7 +445,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm7, %%xmm6\n"      /* 0.0 | im3 | 0.0 | im2 */
     "movlhps %%xmm3, %%xmm2\n"      /* 0.0 | re3 | 0.0 | re2 */
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* re3 | 0.0 | re2 | 0.0 */
-    "movups 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
+    "movaps 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
     "subps  %%xmm2, %%xmm6\n"       /* -re3 | im3 | -re2 | im2 */
     "addl   $32, %%edx\n"
     "movaps %%xmm0, (%%eax)\n"
@@ -474,7 +471,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm2, %%xmm0\n"      /* 0.0 | re1 | 0.0 | re0 */
     "movlhps %%xmm3, %%xmm1\n"      /* 0.0 | im1 | 0.0 | im1 */
     
-    "movups (%%edx), %%xmm4\n"      /* w3 | w2 | w1 | w0 */
+    "movaps (%%edx), %%xmm4\n"      /* w3 | w2 | w1 | w0 */
 
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* im1 | 0.0 | im0 | 0.0 */
     "movss  16(%%esi), %%xmm6\n"    /* re2 */
@@ -486,7 +483,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm3, %%xmm2\n"      /* 0.0 | im3 | 0.0 | im2 */
     "mulps   %%xmm4, %%xmm0\n"
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* im3 | 0.0 | im2 | 0.0 */
-    "movups 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
+    "movaps 16(%%edx), %%xmm4\n"    /* w7 | w6 | w5 | w4 */
     "addl   $32, %%esi\n"
     "subps  %%xmm2, %%xmm6\n"       /* -im3 | re3 | -im2 | re2 */
     "mulps  %%xmm4, %%xmm6\n"
@@ -501,7 +498,6 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "leal 512(%%ebp), %%esi\n"  /* buf[64].re */
     "leal 508(%%ebp), %%edi\n"  /* buf[63].im */
     "movl $16, %%ebx\n"         /* loop count */
-    "addl  $-1024, %%ecx\n"  /* delay */
     
     ".align 16\n"
 ".first_128_delays:\n"
@@ -513,7 +509,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm2, %%xmm0\n"  /* 0.0 | re1 | 0.0 | re0 */
     "movlhps %%xmm3, %%xmm1\n"  /* 0.0 | im1 | 0.0 | im0 */
 
-    "movups -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
+    "movaps -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* im1 | 0.0 | im0 | 0.0 */
     "movss  16(%%esi), %%xmm6\n"    /* re2 */
     "movss  24(%%esi), %%xmm7\n"    /* re3 */
@@ -524,7 +520,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm7, %%xmm6\n"      /* 0.0 | re3 | 0.0 | re2 */
     "movlhps %%xmm3, %%xmm2\n"      /* 0.0 | im3 | 0.0 | im2 */
     "mulps   %%xmm4, %%xmm0\n"
-    "movups (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
+    "movaps (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* im3 | 0.0 | im2 | 0.0 */
     "movaps %%xmm0, (%%ecx)\n"
     "addl   $32, %%esi\n"
@@ -550,7 +546,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm2, %%xmm0\n"  /* 0.0 | im1 | 0.0 | im0 */
     "movlhps %%xmm3, %%xmm1\n"  /* 0.0 | re1 | 0.0 | re0 */
 
-    "movups -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
+    "movaps -16(%%edx), %%xmm4\n"   /* w3 | w2 | w1 | w0 */
     "shufps $0xb1, %%xmm1, %%xmm1\n"/* re1 | 0.0 | re0 | 0.0 */
     "movss  16(%%esi), %%xmm6\n"    /* im2 */
     "movss  24(%%esi), %%xmm7\n"    /* im3 */
@@ -561,7 +557,7 @@ static void imdct512_window_delay_nol_sse (complex_t *buf, float *data_ptr, floa
     "movlhps %%xmm7, %%xmm6\n"      /* 0.0 | im3 | 0.0 | im2 */
     "movlhps %%xmm3, %%xmm2\n"      /* 0.0 | re3 | 0.0 | re2 */
     "mulps   %%xmm4, %%xmm1\n"
-    "movups (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
+    "movaps (%%edx), %%xmm5\n"      /* w7 | w6 | w5 | w4 */
     "shufps $0xb1, %%xmm2, %%xmm2\n"/* re3 | 0.0 | re2 | 0.0 */
     "movaps %%xmm1, (%%ecx)\n"
     "addl   $32, %%esi\n"