From e67660cf374d4301ccb044199e900afaba64baba Mon Sep 17 00:00:00 2001
From: Johann <johannkoenig@google.com>
Date: Fri, 16 Jun 2017 13:14:01 -0700
Subject: fdct32x32 neon implementation

Almost 3x faster in constrained loop testing. Over 10x faster in HBD
builds.

BUG=webm:1424

Change-Id: I2b7f8453e1d4ada63cde729d8115d684c4a71ff9
---
 test/dct_test.cc | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

(limited to 'test/dct_test.cc')

diff --git a/test/dct_test.cc b/test/dct_test.cc
index f78de7a13..848f518f1 100644
--- a/test/dct_test.cc
+++ b/test/dct_test.cc
@@ -529,7 +529,10 @@ INSTANTIATE_TEST_CASE_P(
 #if !CONFIG_EMULATE_HARDWARE
 INSTANTIATE_TEST_CASE_P(
     NEON, TransDCT,
-    ::testing::Values(make_tuple(&vpx_fdct16x16_neon,
+    ::testing::Values(make_tuple(&vpx_fdct32x32_neon,
+                                 &vpx_idct32x32_1024_add_neon, 32, 0,
+                                 VPX_BITS_8),
+                      make_tuple(&vpx_fdct16x16_neon,
                                  &vpx_idct16x16_256_add_neon, 16, 0,
                                  VPX_BITS_8),
                       make_tuple(&vpx_fdct8x8_neon, &vpx_idct8x8_64_add_neon, 8,
-- 
cgit v1.2.3