[PR] avcodec/h264qpel: Deduplicate >8 bit fpel functions (PR #24119)

PR #24119 opened by mkver
URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/24119
Patch URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/24119.patch

Saves 18760B of .text here.



>From 14d4c89606c8e1afa0d0f1dba472b687f8d80341 Mon Sep 17 00:00:00 2001
From: Andreas Rheinhardt <[email protected]>
Date: Thu, 13 Aug 2026 03:02:46 +0200
Subject: [PATCH] avcodec/h264qpel: Deduplicate >8 bit fpel functions

Saves 18760B of .text here.

Signed-off-by: Andreas Rheinhardt <[email protected]>
---
 libavcodec/h264qpel.c          | 32 +++++++++++++++++++-------------
 libavcodec/h264qpel_template.c | 30 ++++++++++++++++++++----------
 libavcodec/snow.c              |  2 +-
 3 files changed, 40 insertions(+), 24 deletions(-)

diff --git a/libavcodec/h264qpel.c b/libavcodec/h264qpel.c
index c64d35b73d..cf49c35f33 100644
--- a/libavcodec/h264qpel.c
+++ b/libavcodec/h264qpel.c
@@ -53,7 +53,6 @@ av_cold void ff_h264qpel_init(H264QpelContext *c, int bit_depth)
 #define FUNCC(f, depth) f ## _ ## depth ## _c
 
 #define dspfunc2(PFX, IDX, NUM, depth)                                  \
-    c->PFX ## _pixels_tab[IDX][ 0] = FUNCC(PFX ## NUM ## _mc00, depth); \
     c->PFX ## _pixels_tab[IDX][ 1] = FUNCC(PFX ## NUM ## _mc10, depth); \
     c->PFX ## _pixels_tab[IDX][ 2] = FUNCC(PFX ## NUM ## _mc20, depth); \
     c->PFX ## _pixels_tab[IDX][ 3] = FUNCC(PFX ## NUM ## _mc30, depth); \
@@ -70,29 +69,36 @@ av_cold void ff_h264qpel_init(H264QpelContext *c, int bit_depth)
     c->PFX ## _pixels_tab[IDX][14] = FUNCC(PFX ## NUM ## _mc23, depth); \
     c->PFX ## _pixels_tab[IDX][15] = FUNCC(PFX ## NUM ## _mc33, depth)
 
-#define SET_QPEL(depth)                         \
-    dspfunc2(put_h264_qpel, 0, 16, depth);      \
-    dspfunc2(put_h264_qpel, 1,  8, depth);      \
-    dspfunc2(put_h264_qpel, 2,  4, depth);      \
-    dspfunc2(avg_h264_qpel, 0, 16, depth);      \
-    dspfunc2(avg_h264_qpel, 1,  8, depth);      \
-    dspfunc2(avg_h264_qpel, 2,  4, depth)
+#define SET_QPEL(M, depth)               \
+    M(put_h264_qpel, 0, 16, depth);      \
+    M(put_h264_qpel, 1,  8, depth);      \
+    M(put_h264_qpel, 2,  4, depth);      \
+    M(avg_h264_qpel, 0, 16, depth);      \
+    M(avg_h264_qpel, 1,  8, depth);      \
+    M(avg_h264_qpel, 2,  4, depth)
+
+#define FPEL(PFX, IDX, NUM, depth) c->PFX ## _pixels_tab[IDX][0] = FUNCC(PFX ## NUM ## _mc00, depth);
+    if (bit_depth <= 8) {
+        SET_QPEL(FPEL, 8);
+    } else {
+        SET_QPEL(FPEL, 16);
+    }
 
     switch (bit_depth) {
     default:
-        SET_QPEL(8);
+        SET_QPEL(dspfunc2, 8);
         break;
     case 9:
-        SET_QPEL(9);
+        SET_QPEL(dspfunc2, 9);
         break;
     case 10:
-        SET_QPEL(10);
+        SET_QPEL(dspfunc2, 10);
         break;
     case 12:
-        SET_QPEL(12);
+        SET_QPEL(dspfunc2, 12);
         break;
     case 14:
-        SET_QPEL(14);
+        SET_QPEL(dspfunc2, 14);
         break;
     }
 
diff --git a/libavcodec/h264qpel_template.c b/libavcodec/h264qpel_template.c
index 33df05475e..c3bc09fd7b 100644
--- a/libavcodec/h264qpel_template.c
+++ b/libavcodec/h264qpel_template.c
@@ -304,12 +304,13 @@ static void FUNC(OPNAME ## h264_qpel16_hv_lowpass)(uint8_t *dst, pixeltmp *tmp,
     FUNC(OPNAME ## h264_qpel8_hv_lowpass)(dst+8*sizeof(pixel), tmp+8, src+8*sizeof(pixel), dstStride, tmpStride, srcStride);\
 }\
 
-#define H264_MC(OPNAME, NAME, SIZE) \
-static void FUNCC(OPNAME ## NAME ## _qpel ## SIZE ## _mc00)(uint8_t *dst, const uint8_t *restrict src, ptrdiff_t stride)\
+#define H264_FPEL(OPNAME, NAME, SIZE) \
+static void FUNCC2(OPNAME ## NAME ## _qpel ## SIZE ## _mc00)(uint8_t *dst, const uint8_t *restrict src, ptrdiff_t stride)\
 {\
     FUNCC(OPNAME ## pixels ## SIZE)(dst, src, stride, SIZE);\
-}\
-\
+}
+
+#define H264_MC(OPNAME, NAME, SIZE) \
 static void FUNCC(OPNAME ## NAME ## _qpel ## SIZE ## _mc10)(uint8_t *dst, const uint8_t *restrict src, ptrdiff_t stride)\
 {\
     uint8_t half[SIZE*SIZE*sizeof(pixel)];\
@@ -463,15 +464,24 @@ static void FUNCC(OPNAME ## NAME ## _qpel ## SIZE ## _mc32)(uint8_t *dst, const
 #define op2_avg(a, b)  a = (((a)+CLIP(((b) + 512)>>10)+1)>>1)
 #define op2_put(a, b)  a = CLIP(((b) + 512)>>10)
 
+#undef H264_QPEL
+#if BIT_DEPTH == 8 || BIT_DEPTH == 9
+#define H264_QPEL(OPNAME, NAME, SIZE) \
+    H264_MC(OPNAME, NAME, SIZE)       \
+    H264_FPEL(OPNAME, NAME, SIZE)
+#else
+#define H264_QPEL(OPNAME, NAME, SIZE) H264_MC(OPNAME, NAME, SIZE)
+#endif
+
 #ifndef SNOW
 H264_LOWPASS(put_       , op_put, op2_put)
 H264_LOWPASS(avg_       , op_avg, op2_avg)
-H264_MC(put_, h264, 4)
-H264_MC(put_, h264, 8)
-H264_MC(put_, h264, 16)
-H264_MC(avg_, h264, 4)
-H264_MC(avg_, h264, 8)
-H264_MC(avg_, h264, 16)
+H264_QPEL(put_, h264, 4)
+H264_QPEL(put_, h264, 8)
+H264_QPEL(put_, h264, 16)
+H264_QPEL(avg_, h264, 4)
+H264_QPEL(avg_, h264, 8)
+H264_QPEL(avg_, h264, 16)
 #endif
 
 #undef op_avg
diff --git a/libavcodec/snow.c b/libavcodec/snow.c
index 38570645b7..ecacf0a751 100644
--- a/libavcodec/snow.c
+++ b/libavcodec/snow.c
@@ -90,7 +90,7 @@ static void put_snow_qpel2_hv_lowpass_8(uint8_t *dst, pixeltmp *tmp, const uint8
     }
 }
 
-H264_MC(put_, snow, 2)
+H264_QPEL(put_, snow, 2)
 
 static av_cold void init_qpel(SnowContext *const s)
 {
-- 
2.52.0

_______________________________________________
ffmpeg-devel mailing list -- [email protected]
To unsubscribe send an email to [email protected]