RTCD: add quantizer functions

This commit continues the process of converting to the new RTCD
system.

Change-Id: Iba9df4c03a508e51c37201c621be43523fae87d9

vp8/common/reconintra4x4.h

@@ -12,6 +12,8 @@
 #ifndef __INC_RECONINTRA4x4_H
 #define __INC_RECONINTRA4x4_H
 
-extern void vp8_intra_prediction_down_copy(MACROBLOCKD *x);
+struct macroblockd;
+
+extern void vp8_intra_prediction_down_copy(struct macroblockd *x);
 
 #endif

vp8/common/rtcd_defs.sh

@@ -3,6 +3,10 @@ cat <<EOF
 struct blockd;
 struct macroblockd;
 struct loop_filter_info;
+
+/* Encoder forward decls */
+struct block;
+struct macroblock;
 EOF
 }
 forward_decls common_forward_decls
@@ -407,5 +411,31 @@ prototype void vp8_short_walsh4x4 "short *input, short *output, int pitch"
 specialize vp8_short_walsh4x4 mmx sse2 media neon
 vp8_short_walsh4x4_media=vp8_short_walsh4x4_armv6
 
+#
+# Quantizer
+#
+prototype void vp8_regular_quantize_b "struct block *, struct blockd *"
+specialize vp8_regular_quantize_b sse2 sse4_1
+vp8_regular_quantize_b_sse4_1=vp8_regular_quantize_b_sse4
+
+prototype void vp8_fast_quantize_b "struct block *, struct blockd *"
+specialize vp8_fast_quantize_b sse2 ssse3 media neon
+vp8_fast_quantize_b_media=vp8_fast_quantize_b_armv6
+
+prototype void vp8_regular_quantize_b_pair "struct block *b1, struct block *b2, struct blockd *d1, struct blockd *d2"
+# no asm yet
+
+prototype void vp8_fast_quantize_b_pair "struct block *b1, struct block *b2, struct blockd *d1, struct blockd *d2"
+specialize vp8_fast_quantize_b_pair neon
+
+prototype void vp8_quantize_mb "struct macroblock *"
+specialize vp8_quantize_mb neon
+
+prototype void vp8_quantize_mby "struct macroblock *"
+specialize vp8_quantize_mby neon
+
+prototype void vp8_quantize_mbuv "struct macroblock *"
+specialize vp8_quantize_mbuv neon
+
 # End of encoder only functions
 fi

vp8/encoder/arm/arm_csystemdependent.c

@@ -38,9 +38,6 @@ void vp8_arch_arm_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.encodemb.subb                  = vp8_subtract_b_armv6;
         cpi->rtcd.encodemb.submby                = vp8_subtract_mby_armv6;
         cpi->rtcd.encodemb.submbuv               = vp8_subtract_mbuv_armv6;
-
-        /*cpi->rtcd.quantize.quantb                = vp8_regular_quantize_b;*/
-        cpi->rtcd.quantize.fastquantb            = vp8_fast_quantize_b_armv6;
     }
 #endif /* HAVE_MEDIA */
 
@@ -53,11 +50,6 @@ void vp8_arch_arm_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.encodemb.subb                  = vp8_subtract_b_neon;
         cpi->rtcd.encodemb.submby                = vp8_subtract_mby_neon;
         cpi->rtcd.encodemb.submbuv               = vp8_subtract_mbuv_neon;
-
-        /*cpi->rtcd.quantize.quantb                = vp8_regular_quantize_b;
-        cpi->rtcd.quantize.quantb_pair           = vp8_regular_quantize_b_pair;*/
-        cpi->rtcd.quantize.fastquantb            = vp8_fast_quantize_b_neon;
-        cpi->rtcd.quantize.fastquantb_pair       = vp8_fast_quantize_b_pair_neon;
     }
 #endif /* HAVE_NEON */
 #endif /* CONFIG_RUNTIME_CPU_DETECT */

vp8/encoder/arm/quantize_arm.c

@@ -9,9 +9,11 @@
  */
 
 
+#include "vpx_config.h"
+#include "vpx_rtcd.h"
+#include "vp8/encoder/block.h"
 #include <math.h>
 #include "vpx_mem/vpx_mem.h"
-
 #include "vp8/encoder/quantize.h"
 #include "vp8/common/entropy.h"
 

vp8/encoder/arm/quantize_arm.h

-/*
- *  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- *  Use of this source code is governed by a BSD-style license
- *  that can be found in the LICENSE file in the root of the source
- *  tree. An additional intellectual property rights grant can be found
- *  in the file PATENTS.  All contributing project authors may
- *  be found in the AUTHORS file in the root of the source tree.
- */
-
-
-#ifndef QUANTIZE_ARM_H
-#define QUANTIZE_ARM_H
-
-#if HAVE_MEDIA
-
-extern prototype_quantize_block(vp8_fast_quantize_b_armv6);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef  vp8_quantize_fastquantb
-#define vp8_quantize_fastquantb vp8_fast_quantize_b_armv6
-#endif
-
-#endif /* HAVE_MEDIA */
-
-
-#if HAVE_NEON
-
-extern prototype_quantize_block(vp8_fast_quantize_b_neon);
-extern prototype_quantize_block_pair(vp8_fast_quantize_b_pair_neon);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-#undef  vp8_quantize_fastquantb
-#define vp8_quantize_fastquantb vp8_fast_quantize_b_neon
-
-#undef  vp8_quantize_fastquantb_pair
-#define vp8_quantize_fastquantb_pair vp8_fast_quantize_b_pair_neon
-
-#undef vp8_quantize_mb
-#define vp8_quantize_mb vp8_quantize_mb_neon
-
-#undef vp8_quantize_mbuv
-#define vp8_quantize_mbuv vp8_quantize_mbuv_neon
-
-#undef vp8_quantize_mby
-#define vp8_quantize_mby vp8_quantize_mby_neon
-#endif
-
-#endif /* HAVE_NEON */
-
-#endif
-

vp8/encoder/block.h

@@ -25,7 +25,7 @@ typedef struct
     int offset;
 } search_site;
 
-typedef struct
+typedef struct block
 {
     // 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries
     short *src_diff;
@@ -57,7 +57,7 @@ typedef struct
     } bmi[16];
 } PARTITION_INFO;
 
-typedef struct
+typedef struct macroblock
 {
     DECLARE_ALIGNED(16, short, src_diff[400]);       // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y
     DECLARE_ALIGNED(16, short, coeff[400]);     // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y

vp8/encoder/encodeframe.c

@@ -1147,10 +1147,8 @@ int vp8cx_encode_inter_macroblock
         /* Are we using the fast quantizer for the mode selection? */
         if(cpi->sf.use_fastquant_for_pick)
         {
-            cpi->mb.quantize_b      = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                      fastquantb);
-            cpi->mb.quantize_b_pair = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                      fastquantb_pair);
+            cpi->mb.quantize_b      = vp8_fast_quantize_b;
+            cpi->mb.quantize_b_pair = vp8_fast_quantize_b_pair;
 
             /* the fast quantizer does not use zbin_extra, so
              * do not recalculate */
@@ -1162,10 +1160,8 @@ int vp8cx_encode_inter_macroblock
         /* switch back to the regular quantizer for the encode */
         if (cpi->sf.improved_quant)
         {
-            cpi->mb.quantize_b      = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                      quantb);
-            cpi->mb.quantize_b_pair = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                      quantb_pair);
+            cpi->mb.quantize_b      = vp8_regular_quantize_b;
+            cpi->mb.quantize_b_pair = vp8_regular_quantize_b_pair;
         }
 
         /* restore cpi->zbin_mode_boost_enabled */

vp8/encoder/generic/csystemdependent.c

@@ -32,10 +32,6 @@ void vp8_cmachine_specific_config(VP8_COMP *cpi)
     cpi->rtcd.encodemb.submby                = vp8_subtract_mby_c;
     cpi->rtcd.encodemb.submbuv               = vp8_subtract_mbuv_c;
 
-    cpi->rtcd.quantize.quantb                = vp8_regular_quantize_b;
-    cpi->rtcd.quantize.quantb_pair           = vp8_regular_quantize_b_pair;
-    cpi->rtcd.quantize.fastquantb            = vp8_fast_quantize_b_c;
-    cpi->rtcd.quantize.fastquantb_pair       = vp8_fast_quantize_b_pair_c;
     cpi->rtcd.search.full_search             = vp8_full_search_sad;
     cpi->rtcd.search.refining_search         = vp8_refining_search_sad;
     cpi->rtcd.search.diamond_search          = vp8_diamond_search_sad;

vp8/encoder/onyx_if.c

@@ -990,17 +990,13 @@ void vp8_set_speed_features(VP8_COMP *cpi)
 
     if (cpi->sf.improved_quant)
     {
-        cpi->mb.quantize_b      = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                  quantb);
-        cpi->mb.quantize_b_pair = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                  quantb_pair);
+        cpi->mb.quantize_b      = vp8_regular_quantize_b;
+        cpi->mb.quantize_b_pair = vp8_regular_quantize_b_pair;
     }
     else
     {
-        cpi->mb.quantize_b      = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                  fastquantb);
-        cpi->mb.quantize_b_pair = QUANTIZE_INVOKE(&cpi->rtcd.quantize,
-                                                  fastquantb_pair);
+        cpi->mb.quantize_b      = vp8_fast_quantize_b;
+        cpi->mb.quantize_b_pair = vp8_fast_quantize_b_pair;
     }
     if (cpi->sf.improved_quant != last_improved_quant)
         vp8cx_init_quantizer(cpi);

vp8/encoder/onyx_int.h

@@ -224,7 +224,6 @@ typedef struct
 typedef struct VP8_ENCODER_RTCD
 {
     vp8_encodemb_rtcd_vtable_t  encodemb;
-    vp8_quantize_rtcd_vtable_t  quantize;
     vp8_search_rtcd_vtable_t    search;
     vp8_temporal_rtcd_vtable_t  temporal;
 } VP8_ENCODER_RTCD;

vp8/encoder/quantize.c

@@ -103,7 +103,7 @@ void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d)
 #endif
 
 #ifdef EXACT_QUANT
-void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d)
+void vp8_regular_quantize_b_c(BLOCK *b, BLOCKD *d)
 {
     int i, rc, eob;
     int zbin;
@@ -156,7 +156,7 @@ void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d)
 }
 
 /* Perform regular quantization, with unbiased rounding and no zero bin. */
-void vp8_strict_quantize_b(BLOCK *b, BLOCKD *d)
+void vp8_strict_quantize_b_c(BLOCK *b, BLOCKD *d)
 {
     int i;
     int rc;
@@ -215,7 +215,7 @@ void vp8_strict_quantize_b(BLOCK *b, BLOCKD *d)
 
 #else
 
-void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d)
+void vp8_regular_quantize_b_c(BLOCK *b, BLOCKD *d)
 {
     int i, rc, eob;
     int zbin;

vp8/encoder/quantize.h

@@ -12,81 +12,12 @@
 #ifndef __INC_QUANTIZE_H
 #define __INC_QUANTIZE_H
 
-#include "block.h"
-
-#define prototype_quantize_block(sym) \
-    void (sym)(BLOCK *b,BLOCKD *d)
-
-#define prototype_quantize_block_pair(sym) \
-    void (sym)(BLOCK *b1, BLOCK *b2, BLOCKD *d1, BLOCKD *d2)
-
-#define prototype_quantize_mb(sym) \
-    void (sym)(MACROBLOCK *x)
-
-#if ARCH_X86 || ARCH_X86_64
-#include "x86/quantize_x86.h"
-#endif
-
-#if ARCH_ARM
-#include "arm/quantize_arm.h"
-#endif
-
-#ifndef vp8_quantize_quantb
-#define vp8_quantize_quantb vp8_regular_quantize_b
-#endif
-extern prototype_quantize_block(vp8_quantize_quantb);
-
-#ifndef vp8_quantize_quantb_pair
-#define vp8_quantize_quantb_pair vp8_regular_quantize_b_pair
-#endif
-extern prototype_quantize_block_pair(vp8_quantize_quantb_pair);
-
-#ifndef vp8_quantize_fastquantb
-#define vp8_quantize_fastquantb vp8_fast_quantize_b_c
-#endif
-extern prototype_quantize_block(vp8_quantize_fastquantb);
-
-#ifndef vp8_quantize_fastquantb_pair
-#define vp8_quantize_fastquantb_pair vp8_fast_quantize_b_pair_c
-#endif
-extern prototype_quantize_block_pair(vp8_quantize_fastquantb_pair);
-
-typedef struct
-{
-    prototype_quantize_block(*quantb);
-    prototype_quantize_block_pair(*quantb_pair);
-    prototype_quantize_block(*fastquantb);
-    prototype_quantize_block_pair(*fastquantb_pair);
-} vp8_quantize_rtcd_vtable_t;
-
-#ifndef vp8_quantize_mb
-#define vp8_quantize_mb vp8_quantize_mb_c
-#endif
-extern prototype_quantize_mb(vp8_quantize_mb);
-
-#ifndef vp8_quantize_mbuv
-#define vp8_quantize_mbuv vp8_quantize_mbuv_c
-#endif
-extern prototype_quantize_mb(vp8_quantize_mbuv);
-
-#ifndef vp8_quantize_mby
-#define vp8_quantize_mby vp8_quantize_mby_c
-#endif
-extern prototype_quantize_mb(vp8_quantize_mby);
-
-#if CONFIG_RUNTIME_CPU_DETECT
-#define QUANTIZE_INVOKE(ctx,fn) (ctx)->fn
-#else
-#define QUANTIZE_INVOKE(ctx,fn) vp8_quantize_##fn
-#endif
-
-extern void vp8_strict_quantize_b(BLOCK *b,BLOCKD *d);
-
 struct VP8_COMP;
+struct macroblock;
 extern void vp8_set_quantizer(struct VP8_COMP *cpi, int Q);
 extern void vp8cx_frame_init_quantizer(struct VP8_COMP *cpi);
-extern void vp8_update_zbin_extra(struct VP8_COMP *cpi, MACROBLOCK *x);
-extern void vp8cx_mb_init_quantizer(struct VP8_COMP *cpi, MACROBLOCK *x, int ok_to_skip);
+extern void vp8_update_zbin_extra(struct VP8_COMP *cpi, struct macroblock *x);
+extern void vp8cx_mb_init_quantizer(struct VP8_COMP *cpi, struct macroblock *x, int ok_to_skip);
 extern void vp8cx_init_quantizer(struct VP8_COMP *cpi);
 
 #endif

vp8/encoder/x86/quantize_x86.h

-/*
- *  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
- *
- *  Use of this source code is governed by a BSD-style license and patent
- *  grant that can be found in the LICENSE file in the root of the source
- *  tree. All contributing project authors may be found in the AUTHORS
- *  file in the root of the source tree.
- */
-
-#ifndef QUANTIZE_X86_H
-#define QUANTIZE_X86_H
-
-
-/* Note:
- *
- * This platform is commonly built for runtime CPU detection. If you modify
- * any of the function mappings present in this file, be sure to also update
- * them in the function pointer initialization code
- */
-#if HAVE_MMX
-
-#endif /* HAVE_MMX */
-
-
-#if HAVE_SSE2
-extern prototype_quantize_block(vp8_regular_quantize_b_sse2);
-extern prototype_quantize_block(vp8_fast_quantize_b_sse2);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_quantize_quantb
-#define vp8_quantize_quantb vp8_regular_quantize_b_sse2
-
-#undef vp8_quantize_fastquantb
-#define vp8_quantize_fastquantb vp8_fast_quantize_b_sse2
-
-#endif /* !CONFIG_RUNTIME_CPU_DETECT */
-
-#endif /* HAVE_SSE2 */
-
-
-#if HAVE_SSSE3
-extern prototype_quantize_block(vp8_fast_quantize_b_ssse3);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_quantize_fastquantb
-#define vp8_quantize_fastquantb vp8_fast_quantize_b_ssse3
-
-#endif /* !CONFIG_RUNTIME_CPU_DETECT */
-
-#endif /* HAVE_SSSE3 */
-
-
-#if HAVE_SSE4_1
-extern prototype_quantize_block(vp8_regular_quantize_b_sse4);
-
-#if !CONFIG_RUNTIME_CPU_DETECT
-
-#undef vp8_quantize_quantb
-#define vp8_quantize_quantb vp8_regular_quantize_b_sse4
-
-#endif /* !CONFIG_RUNTIME_CPU_DETECT */
-
-#endif /* HAVE_SSE4_1 */
-
-#endif /* QUANTIZE_X86_H */

vp8/encoder/x86/x86_csystemdependent.c

@@ -10,6 +10,7 @@
 
 
 #include "vpx_config.h"
+#include "vpx_rtcd.h"
 #include "vpx_ports/x86.h"
 #include "vp8/encoder/variance.h"
 #include "vp8/encoder/onyx_int.h"
@@ -147,10 +148,6 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.encodemb.subb                  = vp8_subtract_b_sse2;
         cpi->rtcd.encodemb.submby                = vp8_subtract_mby_sse2;
         cpi->rtcd.encodemb.submbuv               = vp8_subtract_mbuv_sse2;
-
-        cpi->rtcd.quantize.quantb                = vp8_regular_quantize_b_sse2;
-        cpi->rtcd.quantize.fastquantb            = vp8_fast_quantize_b_sse2;
-
 #if !(CONFIG_REALTIME_ONLY)
         cpi->rtcd.temporal.apply                 = vp8_temporal_filter_apply_sse2;
 #endif
@@ -167,21 +164,10 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi)
     }
 #endif
 
-#if HAVE_SSSE3
-    if (flags & HAS_SSSE3)
-    {
-        cpi->rtcd.quantize.fastquantb            = vp8_fast_quantize_b_ssse3;
-    }
-#endif
-
-
-
 #if HAVE_SSE4_1
     if (flags & HAS_SSE4_1)
     {
         cpi->rtcd.search.full_search             = vp8_full_search_sadx8;
-
-        cpi->rtcd.quantize.quantb                = vp8_regular_quantize_b_sse4;
     }
 #endif
 

vp8/vp8cx.mk

@@ -95,7 +95,6 @@ endif
 
 VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/encodemb_x86.h
 VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/mcomp_x86.h
-VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_x86.h
 VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/temporal_filter_x86.h
 VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/x86_csystemdependent.c
 VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/variance_mmx.c

vp8/vp8cx_arm.mk

@@ -18,7 +18,6 @@ VP8_CX_SRCS-$(ARCH_ARM)  += encoder/arm/arm_csystemdependent.c
 VP8_CX_SRCS-$(ARCH_ARM)  += encoder/arm/dct_arm.c
 VP8_CX_SRCS-$(ARCH_ARM)  += encoder/arm/encodemb_arm.h
 VP8_CX_SRCS-$(ARCH_ARM)  += encoder/arm/quantize_arm.c
-VP8_CX_SRCS-$(ARCH_ARM)  += encoder/arm/quantize_arm.h
 VP8_CX_SRCS-$(ARCH_ARM)  += encoder/arm/variance_arm.c
 
 #File list for edsp