X-Git-Url: https://git.piment-noir.org/?p=deb_ffmpeg.git;a=blobdiff_plain;f=ffmpeg%2Flibavcodec%2Fx86%2Fflacdsp_init.c;h=d04af4511d83741d9cf46661d3c8c7eb59358ea1;hp=ad88e5bbd1306adebcbb8d08c4af0c80b067714b;hb=f6fa7814ccfe3e76514b36cf04f5cd3cb657c8cf;hpb=2ba45a602cbfa7b771effba9b11bb4245c21bc00

diff --git a/ffmpeg/libavcodec/x86/flacdsp_init.c b/ffmpeg/libavcodec/x86/flacdsp_init.c
index ad88e5b..d04af45 100644
--- a/ffmpeg/libavcodec/x86/flacdsp_init.c
+++ b/ffmpeg/libavcodec/x86/flacdsp_init.c
@@ -29,21 +29,89 @@ void ff_flac_lpc_32_xop(int32_t *samples, const int coeffs[32], int order,
 
 void ff_flac_enc_lpc_16_sse4(int32_t *, const int32_t *, int, int, const int32_t *,int);
 
-av_cold void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt,
+#define DECORRELATE_FUNCS(fmt, opt)                                                      \
+void ff_flac_decorrelate_ls_##fmt##_##opt(uint8_t **out, int32_t **in, int channels,     \
+                                          int len, int shift);                           \
+void ff_flac_decorrelate_rs_##fmt##_##opt(uint8_t **out, int32_t **in, int channels,     \
+                                          int len, int shift);                           \
+void ff_flac_decorrelate_ms_##fmt##_##opt(uint8_t **out, int32_t **in, int channels,     \
+                                          int len, int shift);                           \
+void ff_flac_decorrelate_indep2_##fmt##_##opt(uint8_t **out, int32_t **in, int channels, \
+                                             int len, int shift);                        \
+void ff_flac_decorrelate_indep4_##fmt##_##opt(uint8_t **out, int32_t **in, int channels, \
+                                              int len, int shift);                       \
+void ff_flac_decorrelate_indep6_##fmt##_##opt(uint8_t **out, int32_t **in, int channels, \
+                                              int len, int shift);                       \
+void ff_flac_decorrelate_indep8_##fmt##_##opt(uint8_t **out, int32_t **in, int channels, \
+                                              int len, int shift)
+
+DECORRELATE_FUNCS(16, sse2);
+DECORRELATE_FUNCS(16,  avx);
+DECORRELATE_FUNCS(32, sse2);
+DECORRELATE_FUNCS(32,  avx);
+
+av_cold void ff_flacdsp_init_x86(FLACDSPContext *c, enum AVSampleFormat fmt, int channels,
                                  int bps)
 {
 #if HAVE_YASM
     int cpu_flags = av_get_cpu_flags();
 
+#if CONFIG_FLAC_DECODER
+    if (EXTERNAL_SSE2(cpu_flags)) {
+        if (fmt == AV_SAMPLE_FMT_S16) {
+            if (channels == 2)
+                c->decorrelate[0] = ff_flac_decorrelate_indep2_16_sse2;
+            else if (channels == 4)
+                c->decorrelate[0] = ff_flac_decorrelate_indep4_16_sse2;
+            else if (channels == 6)
+                c->decorrelate[0] = ff_flac_decorrelate_indep6_16_sse2;
+            else if (ARCH_X86_64 && channels == 8)
+                c->decorrelate[0] = ff_flac_decorrelate_indep8_16_sse2;
+            c->decorrelate[1] = ff_flac_decorrelate_ls_16_sse2;
+            c->decorrelate[2] = ff_flac_decorrelate_rs_16_sse2;
+            c->decorrelate[3] = ff_flac_decorrelate_ms_16_sse2;
+        } else if (fmt == AV_SAMPLE_FMT_S32) {
+            if (channels == 2)
+                c->decorrelate[0] = ff_flac_decorrelate_indep2_32_sse2;
+            else if (channels == 4)
+                c->decorrelate[0] = ff_flac_decorrelate_indep4_32_sse2;
+            else if (channels == 6)
+                c->decorrelate[0] = ff_flac_decorrelate_indep6_32_sse2;
+            else if (ARCH_X86_64 && channels == 8)
+                c->decorrelate[0] = ff_flac_decorrelate_indep8_32_sse2;
+            c->decorrelate[1] = ff_flac_decorrelate_ls_32_sse2;
+            c->decorrelate[2] = ff_flac_decorrelate_rs_32_sse2;
+            c->decorrelate[3] = ff_flac_decorrelate_ms_32_sse2;
+        }
+    }
     if (EXTERNAL_SSE4(cpu_flags)) {
-        if (bps > 16 && CONFIG_FLAC_DECODER)
+        if (bps > 16)
             c->lpc = ff_flac_lpc_32_sse4;
-        if (bps == 16 && CONFIG_FLAC_ENCODER && CONFIG_GPL)
-            c->lpc_encode = ff_flac_enc_lpc_16_sse4;
+    }
+    if (EXTERNAL_AVX(cpu_flags)) {
+        if (fmt == AV_SAMPLE_FMT_S16) {
+            if (ARCH_X86_64 && channels == 8)
+                c->decorrelate[0] = ff_flac_decorrelate_indep8_16_avx;
+        } else if (fmt == AV_SAMPLE_FMT_S32) {
+            if (channels == 4)
+                c->decorrelate[0] = ff_flac_decorrelate_indep4_32_avx;
+            else if (channels == 6)
+                c->decorrelate[0] = ff_flac_decorrelate_indep6_32_avx;
+            else if (ARCH_X86_64 && channels == 8)
+                c->decorrelate[0] = ff_flac_decorrelate_indep8_32_avx;
+        }
     }
     if (EXTERNAL_XOP(cpu_flags)) {
-        if (bps > 16 && CONFIG_FLAC_DECODER)
+        if (bps > 16)
             c->lpc = ff_flac_lpc_32_xop;
     }
 #endif
+
+#if CONFIG_FLAC_ENCODER
+    if (EXTERNAL_SSE4(cpu_flags)) {
+        if (CONFIG_GPL && bps == 16)
+            c->lpc_encode = ff_flac_enc_lpc_16_sse4;
+    }
+#endif
+#endif /* HAVE_YASM */
 }