From 7971846a5ef42d22385a4dffd1a2aa0b6ee93728 Mon Sep 17 00:00:00 2001 From: hui su Date: Sun, 19 Jul 2015 15:02:56 -0700 Subject: Move intra prediction functions from vp9/common/ to vpx_dsp/ Change-Id: I64edc26cf4aab050c83f2d393df6250628ad43b8 --- vp9/common/mips/dspr2/vp9_intrapred16_dspr2.c | 332 ------------ vp9/common/mips/dspr2/vp9_intrapred4_dspr2.c | 232 -------- vp9/common/mips/dspr2/vp9_intrapred8_dspr2.c | 610 --------------------- vp9/common/mips/msa/vp9_intra_predict_msa.c | 737 -------------------------- 4 files changed, 1911 deletions(-) delete mode 100644 vp9/common/mips/dspr2/vp9_intrapred16_dspr2.c delete mode 100644 vp9/common/mips/dspr2/vp9_intrapred4_dspr2.c delete mode 100644 vp9/common/mips/dspr2/vp9_intrapred8_dspr2.c delete mode 100644 vp9/common/mips/msa/vp9_intra_predict_msa.c (limited to 'vp9/common/mips') diff --git a/vp9/common/mips/dspr2/vp9_intrapred16_dspr2.c b/vp9/common/mips/dspr2/vp9_intrapred16_dspr2.c deleted file mode 100644 index b0dc496ae..000000000 --- a/vp9/common/mips/dspr2/vp9_intrapred16_dspr2.c +++ /dev/null @@ -1,332 +0,0 @@ -/* - * Copyright (c) 2013 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 - -#include "./vp9_rtcd.h" -#include "vp9/common/vp9_common.h" -#include "vp9/common/mips/dspr2/vp9_common_dspr2.h" - -#if HAVE_DSPR2 -void vp9_h_predictor_16x16_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; - int32_t tmp9, tmp10, tmp11, tmp12, tmp13, tmp14, tmp15, tmp16; - - __asm__ __volatile__ ( - "lb %[tmp1], (%[left]) \n\t" - "lb %[tmp2], 1(%[left]) \n\t" - "lb %[tmp3], 2(%[left]) \n\t" - "lb %[tmp4], 3(%[left]) \n\t" - "lb %[tmp5], 4(%[left]) \n\t" - "lb %[tmp6], 5(%[left]) \n\t" - "lb %[tmp7], 6(%[left]) \n\t" - "lb %[tmp8], 7(%[left]) \n\t" - "lb %[tmp9], 8(%[left]) \n\t" - "lb %[tmp10], 9(%[left]) \n\t" - "lb %[tmp11], 10(%[left]) \n\t" - "lb %[tmp12], 11(%[left]) \n\t" - "lb %[tmp13], 12(%[left]) \n\t" - "lb %[tmp14], 13(%[left]) \n\t" - "lb %[tmp15], 14(%[left]) \n\t" - "lb %[tmp16], 15(%[left]) \n\t" - - "replv.qb %[tmp1], %[tmp1] \n\t" - "replv.qb %[tmp2], %[tmp2] \n\t" - "replv.qb %[tmp3], %[tmp3] \n\t" - "replv.qb %[tmp4], %[tmp4] \n\t" - "replv.qb %[tmp5], %[tmp5] \n\t" - "replv.qb %[tmp6], %[tmp6] \n\t" - "replv.qb %[tmp7], %[tmp7] \n\t" - "replv.qb %[tmp8], %[tmp8] \n\t" - "replv.qb %[tmp9], %[tmp9] \n\t" - "replv.qb %[tmp10], %[tmp10] \n\t" - "replv.qb %[tmp11], %[tmp11] \n\t" - "replv.qb %[tmp12], %[tmp12] \n\t" - "replv.qb %[tmp13], %[tmp13] \n\t" - "replv.qb %[tmp14], %[tmp14] \n\t" - "replv.qb %[tmp15], %[tmp15] \n\t" - "replv.qb %[tmp16], %[tmp16] \n\t" - - "sw %[tmp1], (%[dst]) \n\t" - "sw %[tmp1], 4(%[dst]) \n\t" - "sw %[tmp1], 8(%[dst]) \n\t" - "sw %[tmp1], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp2], (%[dst]) \n\t" - "sw %[tmp2], 4(%[dst]) \n\t" - "sw %[tmp2], 8(%[dst]) \n\t" - "sw %[tmp2], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp3], (%[dst]) \n\t" - "sw %[tmp3], 4(%[dst]) \n\t" - "sw %[tmp3], 8(%[dst]) \n\t" - "sw %[tmp3], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp4], (%[dst]) \n\t" - "sw %[tmp4], 4(%[dst]) \n\t" - "sw %[tmp4], 8(%[dst]) \n\t" - "sw %[tmp4], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp5], (%[dst]) \n\t" - "sw %[tmp5], 4(%[dst]) \n\t" - "sw %[tmp5], 8(%[dst]) \n\t" - "sw %[tmp5], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp6], (%[dst]) \n\t" - "sw %[tmp6], 4(%[dst]) \n\t" - "sw %[tmp6], 8(%[dst]) \n\t" - "sw %[tmp6], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp7], (%[dst]) \n\t" - "sw %[tmp7], 4(%[dst]) \n\t" - "sw %[tmp7], 8(%[dst]) \n\t" - "sw %[tmp7], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp8], (%[dst]) \n\t" - "sw %[tmp8], 4(%[dst]) \n\t" - "sw %[tmp8], 8(%[dst]) \n\t" - "sw %[tmp8], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp9], (%[dst]) \n\t" - "sw %[tmp9], 4(%[dst]) \n\t" - "sw %[tmp9], 8(%[dst]) \n\t" - "sw %[tmp9], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp10], (%[dst]) \n\t" - "sw %[tmp10], 4(%[dst]) \n\t" - "sw %[tmp10], 8(%[dst]) \n\t" - "sw %[tmp10], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp11], (%[dst]) \n\t" - "sw %[tmp11], 4(%[dst]) \n\t" - "sw %[tmp11], 8(%[dst]) \n\t" - "sw %[tmp11], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp12], (%[dst]) \n\t" - "sw %[tmp12], 4(%[dst]) \n\t" - "sw %[tmp12], 8(%[dst]) \n\t" - "sw %[tmp12], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp13], (%[dst]) \n\t" - "sw %[tmp13], 4(%[dst]) \n\t" - "sw %[tmp13], 8(%[dst]) \n\t" - "sw %[tmp13], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp14], (%[dst]) \n\t" - "sw %[tmp14], 4(%[dst]) \n\t" - "sw %[tmp14], 8(%[dst]) \n\t" - "sw %[tmp14], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp15], (%[dst]) \n\t" - "sw %[tmp15], 4(%[dst]) \n\t" - "sw %[tmp15], 8(%[dst]) \n\t" - "sw %[tmp15], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp16], (%[dst]) \n\t" - "sw %[tmp16], 4(%[dst]) \n\t" - "sw %[tmp16], 8(%[dst]) \n\t" - "sw %[tmp16], 12(%[dst]) \n\t" - - : [tmp1] "=&r" (tmp1), [tmp2] "=&r" (tmp2), - [tmp3] "=&r" (tmp3), [tmp4] "=&r" (tmp4), - [tmp5] "=&r" (tmp5), [tmp7] "=&r" (tmp7), - [tmp6] "=&r" (tmp6), [tmp8] "=&r" (tmp8), - [tmp9] "=&r" (tmp9), [tmp10] "=&r" (tmp10), - [tmp11] "=&r" (tmp11), [tmp12] "=&r" (tmp12), - [tmp13] "=&r" (tmp13), [tmp14] "=&r" (tmp14), - [tmp15] "=&r" (tmp15), [tmp16] "=&r" (tmp16) - : [left] "r" (left), [dst] "r" (dst), [stride] "r" (stride) - ); -} - -void vp9_dc_predictor_16x16_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t expected_dc; - int32_t average; - int32_t tmp, above1, above_l1, above_r1, left1, left_r1, left_l1; - int32_t above2, left2; - - __asm__ __volatile__ ( - "lw %[above1], (%[above]) \n\t" - "lw %[above2], 4(%[above]) \n\t" - "lw %[left1], (%[left]) \n\t" - "lw %[left2], 4(%[left]) \n\t" - - "preceu.ph.qbl %[above_l1], %[above1] \n\t" - "preceu.ph.qbr %[above_r1], %[above1] \n\t" - "preceu.ph.qbl %[left_l1], %[left1] \n\t" - "preceu.ph.qbr %[left_r1], %[left1] \n\t" - - "addu.ph %[average], %[above_r1], %[above_l1] \n\t" - "addu.ph %[average], %[average], %[left_l1] \n\t" - "addu.ph %[average], %[average], %[left_r1] \n\t" - - "preceu.ph.qbl %[above_l1], %[above2] \n\t" - "preceu.ph.qbr %[above_r1], %[above2] \n\t" - "preceu.ph.qbl %[left_l1], %[left2] \n\t" - "preceu.ph.qbr %[left_r1], %[left2] \n\t" - - "addu.ph %[average], %[average], %[above_l1] \n\t" - "addu.ph %[average], %[average], %[above_r1] \n\t" - "addu.ph %[average], %[average], %[left_l1] \n\t" - "addu.ph %[average], %[average], %[left_r1] \n\t" - - "lw %[above1], 8(%[above]) \n\t" - "lw %[above2], 12(%[above]) \n\t" - "lw %[left1], 8(%[left]) \n\t" - "lw %[left2], 12(%[left]) \n\t" - - "preceu.ph.qbl %[above_l1], %[above1] \n\t" - "preceu.ph.qbr %[above_r1], %[above1] \n\t" - "preceu.ph.qbl %[left_l1], %[left1] \n\t" - "preceu.ph.qbr %[left_r1], %[left1] \n\t" - - "addu.ph %[average], %[average], %[above_l1] \n\t" - "addu.ph %[average], %[average], %[above_r1] \n\t" - "addu.ph %[average], %[average], %[left_l1] \n\t" - "addu.ph %[average], %[average], %[left_r1] \n\t" - - "preceu.ph.qbl %[above_l1], %[above2] \n\t" - "preceu.ph.qbr %[above_r1], %[above2] \n\t" - "preceu.ph.qbl %[left_l1], %[left2] \n\t" - "preceu.ph.qbr %[left_r1], %[left2] \n\t" - - "addu.ph %[average], %[average], %[above_l1] \n\t" - "addu.ph %[average], %[average], %[above_r1] \n\t" - "addu.ph %[average], %[average], %[left_l1] \n\t" - "addu.ph %[average], %[average], %[left_r1] \n\t" - - "addiu %[average], %[average], 16 \n\t" - "srl %[tmp], %[average], 16 \n\t" - "addu.ph %[average], %[tmp], %[average] \n\t" - "srl %[expected_dc], %[average], 5 \n\t" - "replv.qb %[expected_dc], %[expected_dc] \n\t" - - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - "sw %[expected_dc], 8(%[dst]) \n\t" - "sw %[expected_dc], 12(%[dst]) \n\t" - - : [left1] "=&r" (left1), [above1] "=&r" (above1), - [left_l1] "=&r" (left_l1), [above_l1] "=&r" (above_l1), - [left_r1] "=&r" (left_r1), [above_r1] "=&r" (above_r1), - [above2] "=&r" (above2), [left2] "=&r" (left2), - [average] "=&r" (average), [tmp] "=&r" (tmp), - [expected_dc] "=&r" (expected_dc) - : [above] "r" (above), [left] "r" (left), - [dst] "r" (dst), [stride] "r" (stride) - ); -} -#endif // #if HAVE_DSPR2 diff --git a/vp9/common/mips/dspr2/vp9_intrapred4_dspr2.c b/vp9/common/mips/dspr2/vp9_intrapred4_dspr2.c deleted file mode 100644 index a53c62381..000000000 --- a/vp9/common/mips/dspr2/vp9_intrapred4_dspr2.c +++ /dev/null @@ -1,232 +0,0 @@ -/* - * Copyright (c) 2013 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 - -#include "./vp9_rtcd.h" -#include "vp9/common/vp9_common.h" -#include "vp9/common/mips/dspr2/vp9_common_dspr2.h" - -#if HAVE_DSPR2 -void vp9_h_predictor_4x4_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t tmp1, tmp2, tmp3, tmp4; - - __asm__ __volatile__ ( - "lb %[tmp1], (%[left]) \n\t" - "lb %[tmp2], 1(%[left]) \n\t" - "lb %[tmp3], 2(%[left]) \n\t" - "lb %[tmp4], 3(%[left]) \n\t" - "replv.qb %[tmp1], %[tmp1] \n\t" - "replv.qb %[tmp2], %[tmp2] \n\t" - "replv.qb %[tmp3], %[tmp3] \n\t" - "replv.qb %[tmp4], %[tmp4] \n\t" - "sw %[tmp1], (%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp2], (%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp3], (%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp4], (%[dst]) \n\t" - - : [tmp1] "=&r" (tmp1), [tmp2] "=&r" (tmp2), - [tmp3] "=&r" (tmp3), [tmp4] "=&r" (tmp4) - : [left] "r" (left), [dst] "r" (dst), [stride] "r" (stride) - ); -} - -void vp9_dc_predictor_4x4_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t expected_dc; - int32_t average; - int32_t tmp, above_c, above_l, above_r, left_c, left_r, left_l; - - __asm__ __volatile__ ( - "lw %[above_c], (%[above]) \n\t" - "lw %[left_c], (%[left]) \n\t" - - "preceu.ph.qbl %[above_l], %[above_c] \n\t" - "preceu.ph.qbr %[above_r], %[above_c] \n\t" - "preceu.ph.qbl %[left_l], %[left_c] \n\t" - "preceu.ph.qbr %[left_r], %[left_c] \n\t" - - "addu.ph %[average], %[above_r], %[above_l] \n\t" - "addu.ph %[average], %[average], %[left_l] \n\t" - "addu.ph %[average], %[average], %[left_r] \n\t" - "addiu %[average], %[average], 4 \n\t" - "srl %[tmp], %[average], 16 \n\t" - "addu.ph %[average], %[tmp], %[average] \n\t" - "srl %[expected_dc], %[average], 3 \n\t" - "replv.qb %[expected_dc], %[expected_dc] \n\t" - - "sw %[expected_dc], (%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - - : [above_c] "=&r" (above_c), [above_l] "=&r" (above_l), - [above_r] "=&r" (above_r), [left_c] "=&r" (left_c), - [left_l] "=&r" (left_l), [left_r] "=&r" (left_r), - [average] "=&r" (average), [tmp] "=&r" (tmp), - [expected_dc] "=&r" (expected_dc) - : [above] "r" (above), [left] "r" (left), - [dst] "r" (dst), [stride] "r" (stride) - ); -} - -void vp9_tm_predictor_4x4_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t abovel, abover; - int32_t left0, left1, left2, left3; - int32_t res0, res1; - int32_t resl; - int32_t resr; - int32_t top_left; - uint8_t *cm = vp9_ff_cropTbl; - - __asm__ __volatile__ ( - "ulw %[resl], (%[above]) \n\t" - - "lbu %[left0], (%[left]) \n\t" - "lbu %[left1], 1(%[left]) \n\t" - "lbu %[left2], 2(%[left]) \n\t" - "lbu %[left3], 3(%[left]) \n\t" - - "lbu %[top_left], -1(%[above]) \n\t" - - "preceu.ph.qbl %[abovel], %[resl] \n\t" - "preceu.ph.qbr %[abover], %[resl] \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "replv.ph %[left1], %[left1] \n\t" - "replv.ph %[left2], %[left2] \n\t" - "replv.ph %[left3], %[left3] \n\t" - - "replv.ph %[top_left], %[top_left] \n\t" - - "addu.ph %[resl], %[abovel], %[left0] \n\t" - "subu.ph %[resl], %[resl], %[top_left] \n\t" - - "addu.ph %[resr], %[abover], %[left0] \n\t" - "subu.ph %[resr], %[resr], %[top_left] \n\t" - - "sll %[res0], %[resr], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - - "sra %[res1], %[resr], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "sb %[res0], (%[dst]) \n\t" - - "sll %[res0], %[resl], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - - "sra %[res1], %[resl], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - - "addu.ph %[resl], %[abovel], %[left1] \n\t" - "subu.ph %[resl], %[resl], %[top_left] \n\t" - - "addu.ph %[resr], %[abover], %[left1] \n\t" - "subu.ph %[resr], %[resr], %[top_left] \n\t" - - "sb %[res0], 2(%[dst]) \n\t" - "sb %[res1], 3(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - - "sll %[res0], %[resr], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - - "sra %[res1], %[resr], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "sb %[res0], (%[dst]) \n\t" - - "sll %[res0], %[resl], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - - "sb %[res1], 1(%[dst]) \n\t" - "sra %[res1], %[resl], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - - "addu.ph %[resl], %[abovel], %[left2] \n\t" - "subu.ph %[resl], %[resl], %[top_left] \n\t" - - "addu.ph %[resr], %[abover], %[left2] \n\t" - "subu.ph %[resr], %[resr], %[top_left] \n\t" - - "sb %[res0], 2(%[dst]) \n\t" - "sb %[res1], 3(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - - "sll %[res0], %[resr], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - - - "sra %[res1], %[resr], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "sb %[res0], (%[dst]) \n\t" - - "sll %[res0], %[resl], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - - - "sb %[res1], 1(%[dst]) \n\t" - "sra %[res1], %[resl], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - - "addu.ph %[resl], %[abovel], %[left3] \n\t" - "subu.ph %[resl], %[resl], %[top_left] \n\t" - - "addu.ph %[resr], %[abover], %[left3] \n\t" - "subu.ph %[resr], %[resr], %[top_left] \n\t" - - "sb %[res0], 2(%[dst]) \n\t" - "sb %[res1], 3(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - - "sll %[res0], %[resr], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - - "sra %[res1], %[resr], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "sb %[res0], (%[dst]) \n\t" - - "sll %[res0], %[resl], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "lbux %[res0], %[res0](%[cm]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - - "sra %[res1], %[resl], 16 \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - - "sb %[res0], 2(%[dst]) \n\t" - "sb %[res1], 3(%[dst]) \n\t" - - : [abovel] "=&r" (abovel), [abover] "=&r" (abover), - [left0] "=&r" (left0), [left1] "=&r" (left1), [left2] "=&r" (left2), - [res0] "=&r" (res0), [res1] "=&r" (res1), [left3] "=&r" (left3), - [resl] "=&r" (resl), [resr] "=&r" (resr), [top_left] "=&r" (top_left) - : [above] "r" (above), [left] "r" (left), - [dst] "r" (dst), [stride] "r" (stride), [cm] "r" (cm) - ); -} -#endif // #if HAVE_DSPR2 diff --git a/vp9/common/mips/dspr2/vp9_intrapred8_dspr2.c b/vp9/common/mips/dspr2/vp9_intrapred8_dspr2.c deleted file mode 100644 index 40d93ae35..000000000 --- a/vp9/common/mips/dspr2/vp9_intrapred8_dspr2.c +++ /dev/null @@ -1,610 +0,0 @@ -/* - * Copyright (c) 2013 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 - -#include "./vp9_rtcd.h" -#include "vp9/common/vp9_common.h" -#include "vp9/common/mips/dspr2/vp9_common_dspr2.h" - -#if HAVE_DSPR2 -void vp9_h_predictor_8x8_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; - - __asm__ __volatile__ ( - "lb %[tmp1], (%[left]) \n\t" - "lb %[tmp2], 1(%[left]) \n\t" - "lb %[tmp3], 2(%[left]) \n\t" - "lb %[tmp4], 3(%[left]) \n\t" - "lb %[tmp5], 4(%[left]) \n\t" - "lb %[tmp6], 5(%[left]) \n\t" - "lb %[tmp7], 6(%[left]) \n\t" - "lb %[tmp8], 7(%[left]) \n\t" - - "replv.qb %[tmp1], %[tmp1] \n\t" - "replv.qb %[tmp2], %[tmp2] \n\t" - "replv.qb %[tmp3], %[tmp3] \n\t" - "replv.qb %[tmp4], %[tmp4] \n\t" - "replv.qb %[tmp5], %[tmp5] \n\t" - "replv.qb %[tmp6], %[tmp6] \n\t" - "replv.qb %[tmp7], %[tmp7] \n\t" - "replv.qb %[tmp8], %[tmp8] \n\t" - - "sw %[tmp1], (%[dst]) \n\t" - "sw %[tmp1], 4(%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp2], (%[dst]) \n\t" - "sw %[tmp2], 4(%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp3], (%[dst]) \n\t" - "sw %[tmp3], 4(%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp4], (%[dst]) \n\t" - "sw %[tmp4], 4(%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp5], (%[dst]) \n\t" - "sw %[tmp5], 4(%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp6], (%[dst]) \n\t" - "sw %[tmp6], 4(%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp7], (%[dst]) \n\t" - "sw %[tmp7], 4(%[dst]) \n\t" - "add %[dst], %[dst], %[stride] \n\t" - "sw %[tmp8], (%[dst]) \n\t" - "sw %[tmp8], 4(%[dst]) \n\t" - - : [tmp1] "=&r" (tmp1), [tmp2] "=&r" (tmp2), - [tmp3] "=&r" (tmp3), [tmp4] "=&r" (tmp4), - [tmp5] "=&r" (tmp5), [tmp7] "=&r" (tmp7), - [tmp6] "=&r" (tmp6), [tmp8] "=&r" (tmp8) - : [left] "r" (left), [dst] "r" (dst), - [stride] "r" (stride) - ); -} - -void vp9_dc_predictor_8x8_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t expected_dc; - int32_t average; - int32_t tmp, above1, above_l1, above_r1, left1, left_r1, left_l1; - int32_t above2, above_l2, above_r2, left2, left_r2, left_l2; - - __asm__ __volatile__ ( - "lw %[above1], (%[above]) \n\t" - "lw %[above2], 4(%[above]) \n\t" - "lw %[left1], (%[left]) \n\t" - "lw %[left2], 4(%[left]) \n\t" - - "preceu.ph.qbl %[above_l1], %[above1] \n\t" - "preceu.ph.qbr %[above_r1], %[above1] \n\t" - "preceu.ph.qbl %[left_l1], %[left1] \n\t" - "preceu.ph.qbr %[left_r1], %[left1] \n\t" - - "preceu.ph.qbl %[above_l2], %[above2] \n\t" - "preceu.ph.qbr %[above_r2], %[above2] \n\t" - "preceu.ph.qbl %[left_l2], %[left2] \n\t" - "preceu.ph.qbr %[left_r2], %[left2] \n\t" - - "addu.ph %[average], %[above_r1], %[above_l1] \n\t" - "addu.ph %[average], %[average], %[left_l1] \n\t" - "addu.ph %[average], %[average], %[left_r1] \n\t" - - "addu.ph %[average], %[average], %[above_l2] \n\t" - "addu.ph %[average], %[average], %[above_r2] \n\t" - "addu.ph %[average], %[average], %[left_l2] \n\t" - "addu.ph %[average], %[average], %[left_r2] \n\t" - - "addiu %[average], %[average], 8 \n\t" - - "srl %[tmp], %[average], 16 \n\t" - "addu.ph %[average], %[tmp], %[average] \n\t" - "srl %[expected_dc], %[average], 4 \n\t" - "replv.qb %[expected_dc], %[expected_dc] \n\t" - - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - "add %[dst], %[dst], %[stride] \n\t" - "sw %[expected_dc], (%[dst]) \n\t" - "sw %[expected_dc], 4(%[dst]) \n\t" - - : [above1] "=&r" (above1), [above_l1] "=&r" (above_l1), - [above_r1] "=&r" (above_r1), [left1] "=&r" (left1), - [left_l1] "=&r" (left_l1), [left_r1] "=&r" (left_r1), - [above2] "=&r" (above2), [above_l2] "=&r" (above_l2), - [above_r2] "=&r" (above_r2), [left2] "=&r" (left2), - [left_l2] "=&r" (left_l2), [left_r2] "=&r" (left_r2), - [average] "=&r" (average), [tmp] "=&r" (tmp), - [expected_dc] "=&r" (expected_dc) - : [above] "r" (above), [left] "r" (left), [dst] "r" (dst), - [stride] "r" (stride) - ); -} - -void vp9_tm_predictor_8x8_dspr2(uint8_t *dst, ptrdiff_t stride, - const uint8_t *above, const uint8_t *left) { - int32_t abovel, abover; - int32_t abovel_1, abover_1; - int32_t left0; - int32_t res0, res1, res2, res3; - int32_t reshw; - int32_t top_left; - uint8_t *cm = vp9_ff_cropTbl; - - __asm__ __volatile__ ( - "ulw %[reshw], (%[above]) \n\t" - "ulw %[top_left], 4(%[above]) \n\t" - - "lbu %[left0], (%[left]) \n\t" - - "preceu.ph.qbl %[abovel], %[reshw] \n\t" - "preceu.ph.qbr %[abover], %[reshw] \n\t" - "preceu.ph.qbl %[abovel_1], %[top_left] \n\t" - "preceu.ph.qbr %[abover_1], %[top_left] \n\t" - - "lbu %[top_left], -1(%[above]) \n\t" - "replv.ph %[left0], %[left0] \n\t" - - "replv.ph %[top_left], %[top_left] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbu %[left0], 1(%[left]) \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "add %[dst], %[dst], %[stride] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbu %[left0], 2(%[left]) \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "add %[dst], %[dst], %[stride] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbu %[left0], 3(%[left]) \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "add %[dst], %[dst], %[stride] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbu %[left0], 4(%[left]) \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "add %[dst], %[dst], %[stride] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbu %[left0], 5(%[left]) \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "add %[dst], %[dst], %[stride] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbu %[left0], 6(%[left]) \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "add %[dst], %[dst], %[stride] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbu %[left0], 7(%[left]) \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - "replv.ph %[left0], %[left0] \n\t" - "add %[dst], %[dst], %[stride] \n\t" - - "addu.ph %[reshw], %[abovel], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], (%[dst]) \n\t" - "sb %[res1], 1(%[dst]) \n\t" - "sb %[res2], 2(%[dst]) \n\t" - "sb %[res3], 3(%[dst]) \n\t" - - "addu.ph %[reshw], %[abovel_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res2], %[reshw], 16 \n\t" - "sra %[res2], %[res2], 16 \n\t" - "sra %[res3], %[reshw], 16 \n\t" - - "addu.ph %[reshw], %[abover_1], %[left0] \n\t" - "subu.ph %[reshw], %[reshw], %[top_left] \n\t" - - "sll %[res0], %[reshw], 16 \n\t" - "sra %[res0], %[res0], 16 \n\t" - "sra %[res1], %[reshw], 16 \n\t" - - "lbux %[res0], %[res0](%[cm]) \n\t" - "lbux %[res1], %[res1](%[cm]) \n\t" - "lbux %[res2], %[res2](%[cm]) \n\t" - "lbux %[res3], %[res3](%[cm]) \n\t" - - "sb %[res0], 4(%[dst]) \n\t" - "sb %[res1], 5(%[dst]) \n\t" - "sb %[res2], 6(%[dst]) \n\t" - "sb %[res3], 7(%[dst]) \n\t" - - : [abovel] "=&r" (abovel), [abover] "=&r" (abover), - [abovel_1] "=&r" (abovel_1), [abover_1] "=&r" (abover_1), - [left0] "=&r" (left0), [res2] "=&r" (res2), [res3] "=&r" (res3), - [res0] "=&r" (res0), [res1] "=&r" (res1), - [reshw] "=&r" (reshw), [top_left] "=&r" (top_left) - : [above] "r" (above), [left] "r" (left), - [dst] "r" (dst), [stride] "r" (stride), [cm] "r" (cm) - ); -} -#endif // #if HAVE_DSPR2 diff --git a/vp9/common/mips/msa/vp9_intra_predict_msa.c b/vp9/common/mips/msa/vp9_intra_predict_msa.c deleted file mode 100644 index abf2704ca..000000000 --- a/vp9/common/mips/msa/vp9_intra_predict_msa.c +++ /dev/null @@ -1,737 +0,0 @@ -/* - * Copyright (c) 2015 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 "./vp9_rtcd.h" -#include "vpx_dsp/mips/macros_msa.h" - -#define IPRED_SUBS_UH2_UH(in0, in1, out0, out1) { \ - out0 = __msa_subs_u_h(out0, in0); \ - out1 = __msa_subs_u_h(out1, in1); \ -} - -static void intra_predict_vert_4x4_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t src_data; - - src_data = LW(src); - - SW4(src_data, src_data, src_data, src_data, dst, dst_stride); -} - -static void intra_predict_vert_8x8_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t row; - uint32_t src_data1, src_data2; - - src_data1 = LW(src); - src_data2 = LW(src + 4); - - for (row = 8; row--;) { - SW(src_data1, dst); - SW(src_data2, (dst + 4)); - dst += dst_stride; - } -} - -static void intra_predict_vert_16x16_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t row; - v16u8 src0; - - src0 = LD_UB(src); - - for (row = 16; row--;) { - ST_UB(src0, dst); - dst += dst_stride; - } -} - -static void intra_predict_vert_32x32_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t row; - v16u8 src1, src2; - - src1 = LD_UB(src); - src2 = LD_UB(src + 16); - - for (row = 32; row--;) { - ST_UB2(src1, src2, dst, 16); - dst += dst_stride; - } -} - -static void intra_predict_horiz_4x4_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t out0, out1, out2, out3; - - out0 = src[0] * 0x01010101; - out1 = src[1] * 0x01010101; - out2 = src[2] * 0x01010101; - out3 = src[3] * 0x01010101; - - SW4(out0, out1, out2, out3, dst, dst_stride); -} - -static void intra_predict_horiz_8x8_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint64_t out0, out1, out2, out3, out4, out5, out6, out7; - - out0 = src[0] * 0x0101010101010101ull; - out1 = src[1] * 0x0101010101010101ull; - out2 = src[2] * 0x0101010101010101ull; - out3 = src[3] * 0x0101010101010101ull; - out4 = src[4] * 0x0101010101010101ull; - out5 = src[5] * 0x0101010101010101ull; - out6 = src[6] * 0x0101010101010101ull; - out7 = src[7] * 0x0101010101010101ull; - - SD4(out0, out1, out2, out3, dst, dst_stride); - dst += (4 * dst_stride); - SD4(out4, out5, out6, out7, dst, dst_stride); -} - -static void intra_predict_horiz_16x16_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t row; - uint8_t inp0, inp1, inp2, inp3; - v16u8 src0, src1, src2, src3; - - for (row = 4; row--;) { - inp0 = src[0]; - inp1 = src[1]; - inp2 = src[2]; - inp3 = src[3]; - src += 4; - - src0 = (v16u8)__msa_fill_b(inp0); - src1 = (v16u8)__msa_fill_b(inp1); - src2 = (v16u8)__msa_fill_b(inp2); - src3 = (v16u8)__msa_fill_b(inp3); - - ST_UB4(src0, src1, src2, src3, dst, dst_stride); - dst += (4 * dst_stride); - } -} - -static void intra_predict_horiz_32x32_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t row; - uint8_t inp0, inp1, inp2, inp3; - v16u8 src0, src1, src2, src3; - - for (row = 8; row--;) { - inp0 = src[0]; - inp1 = src[1]; - inp2 = src[2]; - inp3 = src[3]; - src += 4; - - src0 = (v16u8)__msa_fill_b(inp0); - src1 = (v16u8)__msa_fill_b(inp1); - src2 = (v16u8)__msa_fill_b(inp2); - src3 = (v16u8)__msa_fill_b(inp3); - - ST_UB2(src0, src0, dst, 16); - dst += dst_stride; - ST_UB2(src1, src1, dst, 16); - dst += dst_stride; - ST_UB2(src2, src2, dst, 16); - dst += dst_stride; - ST_UB2(src3, src3, dst, 16); - dst += dst_stride; - } -} - -static void intra_predict_dc_4x4_msa(const uint8_t *src_top, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - uint32_t val0, val1; - v16i8 store, src = { 0 }; - v8u16 sum_h; - v4u32 sum_w; - v2u64 sum_d; - - val0 = LW(src_top); - val1 = LW(src_left); - INSERT_W2_SB(val0, val1, src); - sum_h = __msa_hadd_u_h((v16u8)src, (v16u8)src); - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 3); - store = __msa_splati_b((v16i8)sum_w, 0); - val0 = __msa_copy_u_w((v4i32)store, 0); - - SW4(val0, val0, val0, val0, dst, dst_stride); -} - -static void intra_predict_dc_tl_4x4_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t val0; - v16i8 store, data = { 0 }; - v8u16 sum_h; - v4u32 sum_w; - - val0 = LW(src); - data = (v16i8)__msa_insert_w((v4i32)data, 0, val0); - sum_h = __msa_hadd_u_h((v16u8)data, (v16u8)data); - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_w, 2); - store = __msa_splati_b((v16i8)sum_w, 0); - val0 = __msa_copy_u_w((v4i32)store, 0); - - SW4(val0, val0, val0, val0, dst, dst_stride); -} - -static void intra_predict_128dc_4x4_msa(uint8_t *dst, int32_t dst_stride) { - uint32_t out; - const v16i8 store = __msa_ldi_b(128); - - out = __msa_copy_u_w((v4i32)store, 0); - - SW4(out, out, out, out, dst, dst_stride); -} - -static void intra_predict_dc_8x8_msa(const uint8_t *src_top, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - uint64_t val0, val1; - v16i8 store; - v16u8 src = { 0 }; - v8u16 sum_h; - v4u32 sum_w; - v2u64 sum_d; - - val0 = LD(src_top); - val1 = LD(src_left); - INSERT_D2_UB(val0, val1, src); - sum_h = __msa_hadd_u_h(src, src); - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_pckev_w((v4i32)sum_d, (v4i32)sum_d); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 4); - store = __msa_splati_b((v16i8)sum_w, 0); - val0 = __msa_copy_u_d((v2i64)store, 0); - - SD4(val0, val0, val0, val0, dst, dst_stride); - dst += (4 * dst_stride); - SD4(val0, val0, val0, val0, dst, dst_stride); -} - -static void intra_predict_dc_tl_8x8_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint64_t val0; - v16i8 store; - v16u8 data = { 0 }; - v8u16 sum_h; - v4u32 sum_w; - v2u64 sum_d; - - val0 = LD(src); - data = (v16u8)__msa_insert_d((v2i64)data, 0, val0); - sum_h = __msa_hadd_u_h(data, data); - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 3); - store = __msa_splati_b((v16i8)sum_w, 0); - val0 = __msa_copy_u_d((v2i64)store, 0); - - SD4(val0, val0, val0, val0, dst, dst_stride); - dst += (4 * dst_stride); - SD4(val0, val0, val0, val0, dst, dst_stride); -} - -static void intra_predict_128dc_8x8_msa(uint8_t *dst, int32_t dst_stride) { - uint64_t out; - const v16i8 store = __msa_ldi_b(128); - - out = __msa_copy_u_d((v2i64)store, 0); - - SD4(out, out, out, out, dst, dst_stride); - dst += (4 * dst_stride); - SD4(out, out, out, out, dst, dst_stride); -} - -static void intra_predict_dc_16x16_msa(const uint8_t *src_top, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - v16u8 top, left, out; - v8u16 sum_h, sum_top, sum_left; - v4u32 sum_w; - v2u64 sum_d; - - top = LD_UB(src_top); - left = LD_UB(src_left); - HADD_UB2_UH(top, left, sum_top, sum_left); - sum_h = sum_top + sum_left; - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_pckev_w((v4i32)sum_d, (v4i32)sum_d); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 5); - out = (v16u8)__msa_splati_b((v16i8)sum_w, 0); - - ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride); - dst += (8 * dst_stride); - ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride); -} - -static void intra_predict_dc_tl_16x16_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - v16u8 data, out; - v8u16 sum_h; - v4u32 sum_w; - v2u64 sum_d; - - data = LD_UB(src); - sum_h = __msa_hadd_u_h(data, data); - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_pckev_w((v4i32)sum_d, (v4i32)sum_d); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 4); - out = (v16u8)__msa_splati_b((v16i8)sum_w, 0); - - ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride); - dst += (8 * dst_stride); - ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride); -} - -static void intra_predict_128dc_16x16_msa(uint8_t *dst, int32_t dst_stride) { - const v16u8 out = (v16u8)__msa_ldi_b(128); - - ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride); - dst += (8 * dst_stride); - ST_UB8(out, out, out, out, out, out, out, out, dst, dst_stride); -} - -static void intra_predict_dc_32x32_msa(const uint8_t *src_top, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - uint32_t row; - v16u8 top0, top1, left0, left1, out; - v8u16 sum_h, sum_top0, sum_top1, sum_left0, sum_left1; - v4u32 sum_w; - v2u64 sum_d; - - LD_UB2(src_top, 16, top0, top1); - LD_UB2(src_left, 16, left0, left1); - HADD_UB2_UH(top0, top1, sum_top0, sum_top1); - HADD_UB2_UH(left0, left1, sum_left0, sum_left1); - sum_h = sum_top0 + sum_top1; - sum_h += sum_left0 + sum_left1; - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_pckev_w((v4i32)sum_d, (v4i32)sum_d); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 6); - out = (v16u8)__msa_splati_b((v16i8)sum_w, 0); - - for (row = 16; row--;) { - ST_UB2(out, out, dst, 16); - dst += dst_stride; - ST_UB2(out, out, dst, 16); - dst += dst_stride; - } -} - -static void intra_predict_dc_tl_32x32_msa(const uint8_t *src, uint8_t *dst, - int32_t dst_stride) { - uint32_t row; - v16u8 data0, data1, out; - v8u16 sum_h, sum_data0, sum_data1; - v4u32 sum_w; - v2u64 sum_d; - - LD_UB2(src, 16, data0, data1); - HADD_UB2_UH(data0, data1, sum_data0, sum_data1); - sum_h = sum_data0 + sum_data1; - sum_w = __msa_hadd_u_w(sum_h, sum_h); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_pckev_w((v4i32)sum_d, (v4i32)sum_d); - sum_d = __msa_hadd_u_d(sum_w, sum_w); - sum_w = (v4u32)__msa_srari_w((v4i32)sum_d, 5); - out = (v16u8)__msa_splati_b((v16i8)sum_w, 0); - - for (row = 16; row--;) { - ST_UB2(out, out, dst, 16); - dst += dst_stride; - ST_UB2(out, out, dst, 16); - dst += dst_stride; - } -} - -static void intra_predict_128dc_32x32_msa(uint8_t *dst, int32_t dst_stride) { - uint32_t row; - const v16u8 out = (v16u8)__msa_ldi_b(128); - - for (row = 16; row--;) { - ST_UB2(out, out, dst, 16); - dst += dst_stride; - ST_UB2(out, out, dst, 16); - dst += dst_stride; - } -} - -static void intra_predict_tm_4x4_msa(const uint8_t *src_top_ptr, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - uint32_t val; - uint8_t top_left = src_top_ptr[-1]; - v16i8 src_left0, src_left1, src_left2, src_left3, tmp0, tmp1, src_top = { 0 }; - v16u8 src0, src1, src2, src3; - v8u16 src_top_left, vec0, vec1, vec2, vec3; - - src_top_left = (v8u16)__msa_fill_h(top_left); - val = LW(src_top_ptr); - src_top = (v16i8)__msa_insert_w((v4i32)src_top, 0, val); - - src_left0 = __msa_fill_b(src_left[0]); - src_left1 = __msa_fill_b(src_left[1]); - src_left2 = __msa_fill_b(src_left[2]); - src_left3 = __msa_fill_b(src_left[3]); - - ILVR_B4_UB(src_left0, src_top, src_left1, src_top, src_left2, src_top, - src_left3, src_top, src0, src1, src2, src3); - HADD_UB4_UH(src0, src1, src2, src3, vec0, vec1, vec2, vec3); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, vec0, vec1); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, vec2, vec3); - SAT_UH4_UH(vec0, vec1, vec2, vec3, 7); - PCKEV_B2_SB(vec1, vec0, vec3, vec2, tmp0, tmp1); - ST4x4_UB(tmp0, tmp1, 0, 2, 0, 2, dst, dst_stride); -} - -static void intra_predict_tm_8x8_msa(const uint8_t *src_top_ptr, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - uint64_t val; - uint8_t top_left = src_top_ptr[-1]; - uint32_t loop_cnt; - v16i8 src_left0, src_left1, src_left2, src_left3, tmp0, tmp1, src_top = { 0 }; - v8u16 src_top_left, vec0, vec1, vec2, vec3; - v16u8 src0, src1, src2, src3; - - val = LD(src_top_ptr); - src_top = (v16i8)__msa_insert_d((v2i64)src_top, 0, val); - src_top_left = (v8u16)__msa_fill_h(top_left); - - for (loop_cnt = 2; loop_cnt--;) { - src_left0 = __msa_fill_b(src_left[0]); - src_left1 = __msa_fill_b(src_left[1]); - src_left2 = __msa_fill_b(src_left[2]); - src_left3 = __msa_fill_b(src_left[3]); - src_left += 4; - - ILVR_B4_UB(src_left0, src_top, src_left1, src_top, src_left2, src_top, - src_left3, src_top, src0, src1, src2, src3); - HADD_UB4_UH(src0, src1, src2, src3, vec0, vec1, vec2, vec3); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, vec0, vec1); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, vec2, vec3); - SAT_UH4_UH(vec0, vec1, vec2, vec3, 7); - PCKEV_B2_SB(vec1, vec0, vec3, vec2, tmp0, tmp1); - ST8x4_UB(tmp0, tmp1, dst, dst_stride); - dst += (4 * dst_stride); - } -} - -static void intra_predict_tm_16x16_msa(const uint8_t *src_top_ptr, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - uint8_t top_left = src_top_ptr[-1]; - uint32_t loop_cnt; - v16i8 src_top, src_left0, src_left1, src_left2, src_left3; - v8u16 src_top_left, res_r, res_l; - - src_top = LD_SB(src_top_ptr); - src_top_left = (v8u16)__msa_fill_h(top_left); - - for (loop_cnt = 4; loop_cnt--;) { - src_left0 = __msa_fill_b(src_left[0]); - src_left1 = __msa_fill_b(src_left[1]); - src_left2 = __msa_fill_b(src_left[2]); - src_left3 = __msa_fill_b(src_left[3]); - src_left += 4; - - ILVRL_B2_UH(src_left0, src_top, res_r, res_l); - HADD_UB2_UH(res_r, res_l, res_r, res_l); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r, res_l); - - SAT_UH2_UH(res_r, res_l, 7); - PCKEV_ST_SB(res_r, res_l, dst); - dst += dst_stride; - - ILVRL_B2_UH(src_left1, src_top, res_r, res_l); - HADD_UB2_UH(res_r, res_l, res_r, res_l); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r, res_l); - SAT_UH2_UH(res_r, res_l, 7); - PCKEV_ST_SB(res_r, res_l, dst); - dst += dst_stride; - - ILVRL_B2_UH(src_left2, src_top, res_r, res_l); - HADD_UB2_UH(res_r, res_l, res_r, res_l); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r, res_l); - SAT_UH2_UH(res_r, res_l, 7); - PCKEV_ST_SB(res_r, res_l, dst); - dst += dst_stride; - - ILVRL_B2_UH(src_left3, src_top, res_r, res_l); - HADD_UB2_UH(res_r, res_l, res_r, res_l); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r, res_l); - SAT_UH2_UH(res_r, res_l, 7); - PCKEV_ST_SB(res_r, res_l, dst); - dst += dst_stride; - } -} - -static void intra_predict_tm_32x32_msa(const uint8_t *src_top, - const uint8_t *src_left, - uint8_t *dst, int32_t dst_stride) { - uint8_t top_left = src_top[-1]; - uint32_t loop_cnt; - v16i8 src_top0, src_top1, src_left0, src_left1, src_left2, src_left3; - v8u16 src_top_left, res_r0, res_r1, res_l0, res_l1; - - LD_SB2(src_top, 16, src_top0, src_top1); - src_top_left = (v8u16)__msa_fill_h(top_left); - - for (loop_cnt = 8; loop_cnt--;) { - src_left0 = __msa_fill_b(src_left[0]); - src_left1 = __msa_fill_b(src_left[1]); - src_left2 = __msa_fill_b(src_left[2]); - src_left3 = __msa_fill_b(src_left[3]); - src_left += 4; - - ILVR_B2_UH(src_left0, src_top0, src_left0, src_top1, res_r0, res_r1); - ILVL_B2_UH(src_left0, src_top0, src_left0, src_top1, res_l0, res_l1); - HADD_UB4_UH(res_r0, res_l0, res_r1, res_l1, res_r0, res_l0, res_r1, res_l1); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r0, res_l0); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r1, res_l1); - SAT_UH4_UH(res_r0, res_l0, res_r1, res_l1, 7); - PCKEV_ST_SB(res_r0, res_l0, dst); - PCKEV_ST_SB(res_r1, res_l1, dst + 16); - dst += dst_stride; - - ILVR_B2_UH(src_left1, src_top0, src_left1, src_top1, res_r0, res_r1); - ILVL_B2_UH(src_left1, src_top0, src_left1, src_top1, res_l0, res_l1); - HADD_UB4_UH(res_r0, res_l0, res_r1, res_l1, res_r0, res_l0, res_r1, res_l1); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r0, res_l0); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r1, res_l1); - SAT_UH4_UH(res_r0, res_l0, res_r1, res_l1, 7); - PCKEV_ST_SB(res_r0, res_l0, dst); - PCKEV_ST_SB(res_r1, res_l1, dst + 16); - dst += dst_stride; - - ILVR_B2_UH(src_left2, src_top0, src_left2, src_top1, res_r0, res_r1); - ILVL_B2_UH(src_left2, src_top0, src_left2, src_top1, res_l0, res_l1); - HADD_UB4_UH(res_r0, res_l0, res_r1, res_l1, res_r0, res_l0, res_r1, res_l1); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r0, res_l0); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r1, res_l1); - SAT_UH4_UH(res_r0, res_l0, res_r1, res_l1, 7); - PCKEV_ST_SB(res_r0, res_l0, dst); - PCKEV_ST_SB(res_r1, res_l1, dst + 16); - dst += dst_stride; - - ILVR_B2_UH(src_left3, src_top0, src_left3, src_top1, res_r0, res_r1); - ILVL_B2_UH(src_left3, src_top0, src_left3, src_top1, res_l0, res_l1); - HADD_UB4_UH(res_r0, res_l0, res_r1, res_l1, res_r0, res_l0, res_r1, res_l1); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r0, res_l0); - IPRED_SUBS_UH2_UH(src_top_left, src_top_left, res_r1, res_l1); - SAT_UH4_UH(res_r0, res_l0, res_r1, res_l1, 7); - PCKEV_ST_SB(res_r0, res_l0, dst); - PCKEV_ST_SB(res_r1, res_l1, dst + 16); - dst += dst_stride; - } -} - -void vp9_v_predictor_4x4_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_vert_4x4_msa(above, dst, y_stride); -} - -void vp9_v_predictor_8x8_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_vert_8x8_msa(above, dst, y_stride); -} - -void vp9_v_predictor_16x16_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_vert_16x16_msa(above, dst, y_stride); -} - -void vp9_v_predictor_32x32_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_vert_32x32_msa(above, dst, y_stride); -} - -void vp9_h_predictor_4x4_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - - intra_predict_horiz_4x4_msa(left, dst, y_stride); -} - -void vp9_h_predictor_8x8_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - - intra_predict_horiz_8x8_msa(left, dst, y_stride); -} - -void vp9_h_predictor_16x16_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - - intra_predict_horiz_16x16_msa(left, dst, y_stride); -} - -void vp9_h_predictor_32x32_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - - intra_predict_horiz_32x32_msa(left, dst, y_stride); -} - -void vp9_dc_predictor_4x4_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_dc_4x4_msa(above, left, dst, y_stride); -} - -void vp9_dc_predictor_8x8_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_dc_8x8_msa(above, left, dst, y_stride); -} - -void vp9_dc_predictor_16x16_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_dc_16x16_msa(above, left, dst, y_stride); -} - -void vp9_dc_predictor_32x32_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_dc_32x32_msa(above, left, dst, y_stride); -} - -void vp9_dc_top_predictor_4x4_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_dc_tl_4x4_msa(above, dst, y_stride); -} - -void vp9_dc_top_predictor_8x8_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_dc_tl_8x8_msa(above, dst, y_stride); -} - -void vp9_dc_top_predictor_16x16_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_dc_tl_16x16_msa(above, dst, y_stride); -} - -void vp9_dc_top_predictor_32x32_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)left; - - intra_predict_dc_tl_32x32_msa(above, dst, y_stride); -} - -void vp9_dc_left_predictor_4x4_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - - intra_predict_dc_tl_4x4_msa(left, dst, y_stride); -} - -void vp9_dc_left_predictor_8x8_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - - intra_predict_dc_tl_8x8_msa(left, dst, y_stride); -} - -void vp9_dc_left_predictor_16x16_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, - const uint8_t *left) { - (void)above; - - intra_predict_dc_tl_16x16_msa(left, dst, y_stride); -} - -void vp9_dc_left_predictor_32x32_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, - const uint8_t *left) { - (void)above; - - intra_predict_dc_tl_32x32_msa(left, dst, y_stride); -} - -void vp9_dc_128_predictor_4x4_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - (void)left; - - intra_predict_128dc_4x4_msa(dst, y_stride); -} - -void vp9_dc_128_predictor_8x8_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - (void)left; - - intra_predict_128dc_8x8_msa(dst, y_stride); -} - -void vp9_dc_128_predictor_16x16_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - (void)left; - - intra_predict_128dc_16x16_msa(dst, y_stride); -} - -void vp9_dc_128_predictor_32x32_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - (void)above; - (void)left; - - intra_predict_128dc_32x32_msa(dst, y_stride); -} - -void vp9_tm_predictor_4x4_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_tm_4x4_msa(above, left, dst, y_stride); -} - -void vp9_tm_predictor_8x8_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_tm_8x8_msa(above, left, dst, y_stride); -} - -void vp9_tm_predictor_16x16_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_tm_16x16_msa(above, left, dst, y_stride); -} - -void vp9_tm_predictor_32x32_msa(uint8_t *dst, ptrdiff_t y_stride, - const uint8_t *above, const uint8_t *left) { - intra_predict_tm_32x32_msa(above, left, dst, y_stride); -} -- cgit v1.2.3