From: Luca Barbato <lu_zero@gentoo.org>
Date: Fri, 7 Apr 2017 13:50:12 +0000 (+0000)
Subject: ppc: vertical intrapred 16x16 and 32x32
X-Git-Tag: v1.7.0~549^2~4
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=7ad1faa6f8ed5472864155dca5a7a01b787f0365;p=libvpx

ppc: vertical intrapred 16x16 and 32x32

Change-Id: Ic80f3c050cfbe7697e81a311b4edaaa597b85cab
---

diff --git a/test/test_intra_pred_speed.cc b/test/test_intra_pred_speed.cc
index 17dde1a52..9ae05b02f 100644
--- a/test/test_intra_pred_speed.cc
+++ b/test/test_intra_pred_speed.cc
@@ -312,6 +312,16 @@ INTRA_PRED_TEST(MSA, TestIntraPred32, vpx_dc_predictor_32x32_msa,
                 vpx_tm_predictor_32x32_msa)
 #endif  // HAVE_MSA
 
+#if HAVE_VSX
+INTRA_PRED_TEST(VSX, TestIntraPred16, NULL, NULL, NULL, NULL,
+                vpx_v_predictor_16x16_vsx, NULL, NULL, NULL, NULL, NULL, NULL,
+                NULL, NULL)
+
+INTRA_PRED_TEST(VSX, TestIntraPred32, NULL, NULL, NULL, NULL,
+                vpx_v_predictor_32x32_vsx, NULL, NULL, NULL, NULL, NULL, NULL,
+                NULL, NULL)
+#endif  // HAVE_VSX
+
 // -----------------------------------------------------------------------------
 
 #if CONFIG_VP9_HIGHBITDEPTH
diff --git a/vpx_dsp/ppc/intrapred_vsx.c b/vpx_dsp/ppc/intrapred_vsx.c
new file mode 100644
index 000000000..16e08eec4
--- /dev/null
+++ b/vpx_dsp/ppc/intrapred_vsx.c
@@ -0,0 +1,36 @@
+/*
+ *  Copyright (c) 2017 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.
+ */
+
+#include "./vpx_dsp_rtcd.h"
+#include "vpx_dsp/ppc/types_vsx.h"
+
+void vpx_v_predictor_16x16_vsx(uint8_t *dst, ptrdiff_t stride,
+                               const uint8_t *above, const uint8_t *left) {
+  const uint8x16_t d = vec_vsx_ld(0, above);
+  int i;
+  (void)left;
+
+  for (i = 0; i < 16; i++, dst += stride) {
+    vec_vsx_st(d, 0, dst);
+  }
+}
+
+void vpx_v_predictor_32x32_vsx(uint8_t *dst, ptrdiff_t stride,
+                               const uint8_t *above, const uint8_t *left) {
+  const uint8x16_t d0 = vec_vsx_ld(0, above);
+  const uint8x16_t d1 = vec_vsx_ld(16, above);
+  int i;
+  (void)left;
+
+  for (i = 0; i < 32; i++, dst += stride) {
+    vec_vsx_st(d0, 0, dst);
+    vec_vsx_st(d1, 16, dst);
+  }
+}
diff --git a/vpx_dsp/ppc/types_vsx.h b/vpx_dsp/ppc/types_vsx.h
index 2f3aa2049..66a9a03fe 100644
--- a/vpx_dsp/ppc/types_vsx.h
+++ b/vpx_dsp/ppc/types_vsx.h
@@ -13,8 +13,11 @@
 
 #include <altivec.h>
 
+typedef vector signed char int8x16_t;
+typedef vector unsigned char uint8x16_t;
 typedef vector signed short int16x8_t;
 typedef vector unsigned short uint16x8_t;
 typedef vector signed int int32x4_t;
+typedef vector unsigned int uint32x4_t;
 
 #endif  // VPX_DSP_PPC_TYPES_VSX_H_
diff --git a/vpx_dsp/vpx_dsp.mk b/vpx_dsp/vpx_dsp.mk
index b959cdcda..64df59cfe 100644
--- a/vpx_dsp/vpx_dsp.mk
+++ b/vpx_dsp/vpx_dsp.mk
@@ -51,6 +51,7 @@ DSP_SRCS-$(HAVE_SSE) += x86/intrapred_sse2.asm
 DSP_SRCS-$(HAVE_SSE2) += x86/intrapred_sse2.asm
 DSP_SRCS-$(HAVE_SSSE3) += x86/intrapred_ssse3.asm
 DSP_SRCS-$(HAVE_SSSE3) += x86/vpx_subpixel_8t_ssse3.asm
+DSP_SRCS-$(HAVE_VSX) += ppc/intrapred_vsx.c
 
 ifeq ($(CONFIG_VP9_HIGHBITDEPTH),yes)
 DSP_SRCS-$(HAVE_SSE)  += x86/highbd_intrapred_sse2.asm
diff --git a/vpx_dsp/vpx_dsp_rtcd_defs.pl b/vpx_dsp/vpx_dsp_rtcd_defs.pl
index e1181223d..f0609f591 100644
--- a/vpx_dsp/vpx_dsp_rtcd_defs.pl
+++ b/vpx_dsp/vpx_dsp_rtcd_defs.pl
@@ -130,7 +130,7 @@ add_proto qw/void vpx_d153_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride,
 specialize qw/vpx_d153_predictor_16x16 ssse3/;
 
 add_proto qw/void vpx_v_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_v_predictor_16x16 neon msa sse2/;
+specialize qw/vpx_v_predictor_16x16 neon msa sse2 vsx/;
 
 add_proto qw/void vpx_tm_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_tm_predictor_16x16 neon msa sse2/;
@@ -168,7 +168,7 @@ add_proto qw/void vpx_d153_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride,
 specialize qw/vpx_d153_predictor_32x32 ssse3/;
 
 add_proto qw/void vpx_v_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_v_predictor_32x32 neon msa sse2/;
+specialize qw/vpx_v_predictor_32x32 neon msa sse2 vsx/;
 
 add_proto qw/void vpx_tm_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_tm_predictor_32x32 neon msa sse2/;