tree 0ad8305d59dfddd9064cbe2ff3ccf6313db7431d
parent eb6289891935407ffee4481cd45d99fa10a23749
author Vitalii Dziumenko <vdziumenko@luxoft.corp-partner.google.com> 1585587414 +0300
committer Yunqing Wang <yunqingwang@google.com> 1586974796 +0000

Add NEON version of few quantize functions

Speedup:
av1_quantize_fp_32x32_c vs av1_quantize_fp_32x32_neon avg gain 4.4
av1_quantize_fp_64x64_c vs &av1_quantize_fp_64x64_neon avg gain 3.7
aom_quantize_b_c vs aom_quantize_b_neon avg gain 5.2
aom_quantize_b_32x32_c vs &aom_quantize_b_32x32_neon(aom_quantize_b_helper inside) avg gain 5.0
aom_quantize_b_64x64_c vs &aom_quantize_b_64x64_neon(aom_quantize_b_helper inside) avg gain 4.2

Tested via NEON/QuantizeTest.DISABLED_Speed

Change-Id: I829578b6a11c0bac69d4514d54c1652eb50144f3