aarch64: Optimize various intra_predict asm functions
authorJanne Grunau <janne-x264@jannau.net>
Mon, 17 Aug 2015 14:39:20 +0000 (16:39 +0200)
committerHenrik Gramner <henrik@gramner.com>
Sun, 11 Oct 2015 16:44:54 +0000 (18:44 +0200)
commitaec81efd3fe43008551916aa6073eb0732a58210
tree33e35b65b07e859cd715cb6a24d2cd2beb4fd3ba
parentb16268ac0826d78455d0d704ea0fc8b1edc6b6bf
aarch64: Optimize various intra_predict asm functions

Make them at least as fast as the compiled C version (tested on
cortex-a53 vs. gcc 4.9.2).

                        C     NEON (before)   NEON (after)
intra_predict_4x4_dc:   260   335             260
intra_predict_4x4_dct:  210   265             200
intra_predict_8x8c_dc:  497   548             493
intra_predict_8x8c_v:   232   309             179 (arm64)
intra_predict_8x16c_dc: 795   830             790
common/aarch64/predict-a.S
common/aarch64/predict-c.c
common/aarch64/predict.h