0N/A/*
2362N/A * Copyright (c) 1998, 2003, Oracle and/or its affiliates. All rights reserved.
0N/A * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
0N/A *
0N/A * This code is free software; you can redistribute it and/or modify it
0N/A * under the terms of the GNU General Public License version 2 only, as
2362N/A * published by the Free Software Foundation. Oracle designates this
0N/A * particular file as subject to the "Classpath" exception as provided
2362N/A * by Oracle in the LICENSE file that accompanied this code.
0N/A *
0N/A * This code is distributed in the hope that it will be useful, but WITHOUT
0N/A * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
0N/A * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
0N/A * version 2 for more details (a copy is included in the LICENSE file that
0N/A * accompanied this code).
0N/A *
0N/A * You should have received a copy of the GNU General Public License version
0N/A * 2 along with this work; if not, write to the Free Software Foundation,
0N/A * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
0N/A *
2362N/A * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
2362N/A * or visit www.oracle.com if you need additional information or have any
2362N/A * questions.
0N/A */
0N/A
0N/A
0N/A
0N/A#include "vis_proto.h"
0N/A#include "mlib_image.h"
0N/A#include "mlib_v_ImageLookUpFunc.h"
0N/A
0N/A/***************************************************************/
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_2_SrcOff0_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_2_SrcOff1_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_2_SrcOff2_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_2_SrcOff3_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_2_DstNonAl_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_2_DstA8D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_2_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_3_SrcOff0_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_3_SrcOff1_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_3_SrcOff2_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_3_SrcOff3_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_3_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_4_SrcOff0_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_f32 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_4_DstNonAl_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_f32 *table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_4_DstOff0_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_4_DstOff1_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_4_DstOff2_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table);
0N/A
0N/Astatic void mlib_v_ImageLookUpSI_U8_U8_4_DstOff3_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table);
0N/A
0N/A/***************************************************************/
0N/A#define VIS_LD_U8_I(X, Y) vis_ld_u8_i((void *)(X), (Y))
0N/A#define VIS_LD_U16_I(X, Y) vis_ld_u16_i((void *)(X), (Y))
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2_SrcOff0_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table)
0N/A{
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0; /* source data */
0N/A mlib_u16 *dl; /* pointer to start of destination */
0N/A mlib_u16 *dend; /* pointer to end of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, acc; /* destination data */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 i, num; /* loop variable */
0N/A
0N/A sa = (mlib_u32*)src;
0N/A dl = (mlib_u16*)dst;
0N/A dp = (mlib_d64 *) dl;
0N/A dend = dl + xsize - 1;
0N/A
0N/A vis_alignaddr((void *) 0, 6);
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s0 = sa[0];
0N/A sa ++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, sa++) {
0N/A t3 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 23) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = sa[0];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t3 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 23) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A sp = (mlib_u8*)sa;
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A
0N/A num = (mlib_u16*) dend - (mlib_u16*) dp;
0N/A sp += num;
0N/A num ++;
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < num; i ++) {
0N/A s0 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U16_I(table, 2*s0);
0N/A acc = vis_faligndata(t0, acc);
0N/A }
0N/A
0N/A emask = vis_edge16(dp, dend);
0N/A vis_pst_16(acc, dp, emask);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2_SrcOff1_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table)
0N/A{
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1; /* source data */
0N/A mlib_u16 *dl; /* pointer to start of destination */
0N/A mlib_u16 *dend; /* pointer to end of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, acc; /* destination data */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 i, num; /* loop variable */
0N/A
0N/A sa = (mlib_u32*)(src-1);
0N/A dl = (mlib_u16*)dst;
0N/A dp = (mlib_d64 *) dl;
0N/A dend = dl + xsize - 1;
0N/A
0N/A vis_alignaddr((void *) 0, 6);
0N/A
0N/A s0 = *sa++;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s1 = sa[0];
0N/A sa ++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, sa++) {
0N/A t3 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s1;
0N/A s1 = sa[0];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t3 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A sp = (mlib_u8*)sa;
0N/A sp -= 3;
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A
0N/A num = (mlib_u16*) dend - (mlib_u16*) dp;
0N/A sp += num;
0N/A num ++;
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < num; i ++) {
0N/A s0 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U16_I(table, 2*s0);
0N/A acc = vis_faligndata(t0, acc);
0N/A }
0N/A
0N/A emask = vis_edge16(dp, dend);
0N/A vis_pst_16(acc, dp, emask);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2_SrcOff2_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table)
0N/A{
0N/A mlib_u32 *sa; /* pointer to source data */
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1; /* source data */
0N/A mlib_u16 *dl; /* pointer to start of destination */
0N/A mlib_u16 *dend; /* pointer to end of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, acc; /* destination data */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 i, num; /* loop variable */
0N/A
0N/A sa = (mlib_u32*)(src-2);
0N/A dl = (mlib_u16*)dst;
0N/A dp = (mlib_d64 *) dl;
0N/A dend = dl + xsize - 1;
0N/A
0N/A vis_alignaddr((void *) 0, 6);
0N/A
0N/A s0 = *sa++;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s1 = sa[0];
0N/A sa ++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, sa++) {
0N/A t3 = VIS_LD_U16_I(table, (s1 >> 15) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s1;
0N/A s1 = sa[0];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t3 = VIS_LD_U16_I(table, (s1 >> 15) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A sp = (mlib_u8*)sa;
0N/A sp -= 2;
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A
0N/A num = (mlib_u16*) dend - (mlib_u16*) dp;
0N/A sp += num;
0N/A num ++;
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < num; i ++) {
0N/A s0 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U16_I(table, 2*s0);
0N/A acc = vis_faligndata(t0, acc);
0N/A }
0N/A
0N/A emask = vis_edge16(dp, dend);
0N/A vis_pst_16(acc, dp, emask);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2_SrcOff3_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table)
0N/A{
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1; /* source data */
0N/A mlib_u16 *dl; /* pointer to start of destination */
0N/A mlib_u16 *dend; /* pointer to end of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, acc; /* destination data */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 i, num; /* loop variable */
0N/A
0N/A sa = (mlib_u32*)(src-3);
0N/A dl = (mlib_u16*)dst;
0N/A dp = (mlib_d64 *) dl;
0N/A dend = dl + xsize - 1;
0N/A
0N/A vis_alignaddr((void *) 0, 6);
0N/A
0N/A s0 = *sa++;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s1 = sa[0];
0N/A sa ++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, sa++) {
0N/A t3 = VIS_LD_U16_I(table, (s1 >> 7) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s1 >> 15) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s1;
0N/A s1 = sa[0];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t3 = VIS_LD_U16_I(table, (s1 >> 7) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s1 >> 15) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A sp = (mlib_u8*)sa;
0N/A sp -= 1;
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A
0N/A num = (mlib_u16*) dend - (mlib_u16*) dp;
0N/A sp += num;
0N/A num ++;
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < num; i ++) {
0N/A s0 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U16_I(table, 2*s0);
0N/A acc = vis_faligndata(t0, acc);
0N/A }
0N/A
0N/A emask = vis_edge16(dp, dend);
0N/A vis_pst_16(acc, dp, emask);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2_DstNonAl_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u16 *table)
0N/A{
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1, s2, s3; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_u8 *dend; /* pointer to end of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7, acc0; /* destination data */
0N/A mlib_d64 acc1, acc2; /* destination data */
0N/A mlib_d64 acc3, acc4; /* destination data */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 i, num; /* loop variable */
0N/A mlib_s32 off; /* offset */
0N/A
0N/A sa = (mlib_u32*)src;
0N/A dl = dst;
0N/A sp = (void *)src;
0N/A dend = dl + 2*xsize - 1;
0N/A dp = (mlib_d64 *) ((mlib_addr) dl & (~7));
0N/A off = (mlib_addr) dp - (mlib_addr) dl;
0N/A
0N/A emask = vis_edge8(dl, dend);
0N/A num = (xsize < 4) ? xsize : 4;
0N/A
0N/A sp += (num-1);
0N/A
0N/A vis_alignaddr(dp, 6);
0N/A
0N/A for (i = 0; i < num; i ++) {
0N/A s0 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U16_I(table, 2*s0);
0N/A acc0 = vis_faligndata(t0, acc0);
0N/A }
0N/A
0N/A vis_alignaddr(dp, off);
0N/A vis_pst_8(vis_faligndata(acc0, acc0), dp++, emask);
0N/A
0N/A sa++;
0N/A
0N/A xsize -= 4;
0N/A
0N/A i = 0;
0N/A
0N/A if (xsize >= 16) {
0N/A
0N/A s0 = sa[0];
0N/A s1 = sa[1];
0N/A s2 = sa[2];
0N/A s3 = sa[3];
0N/A sa += 4;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 32; i+=16, sa+=4) {
0N/A vis_alignaddr(dp, 6);
0N/A t3 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 23) & 0x1FE);
0N/A acc1 = vis_faligndata(t3, acc1);
0N/A acc1 = vis_faligndata(t2, acc1);
0N/A acc1 = vis_faligndata(t1, acc1);
0N/A acc1 = vis_faligndata(t0, acc1);
0N/A t7 = VIS_LD_U16_I(table, (s1 << 1) & 0x1FE);
0N/A t6 = VIS_LD_U16_I(table, (s1 >> 7) & 0x1FE);
0N/A t5 = VIS_LD_U16_I(table, (s1 >> 15) & 0x1FE);
0N/A t4 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A acc2 = vis_faligndata(t7, acc2);
0N/A acc2 = vis_faligndata(t6, acc2);
0N/A acc2 = vis_faligndata(t5, acc2);
0N/A acc2 = vis_faligndata(t4, acc2);
0N/A t3 = VIS_LD_U16_I(table, (s2 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s2 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s2 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s2 >> 23) & 0x1FE);
0N/A acc3 = vis_faligndata(t3, acc3);
0N/A acc3 = vis_faligndata(t2, acc3);
0N/A acc3 = vis_faligndata(t1, acc3);
0N/A acc3 = vis_faligndata(t0, acc3);
0N/A t7 = VIS_LD_U16_I(table, (s3 << 1) & 0x1FE);
0N/A t6 = VIS_LD_U16_I(table, (s3 >> 7) & 0x1FE);
0N/A t5 = VIS_LD_U16_I(table, (s3 >> 15) & 0x1FE);
0N/A t4 = VIS_LD_U16_I(table, (s3 >> 23) & 0x1FE);
0N/A acc4 = vis_faligndata(t7, acc4);
0N/A acc4 = vis_faligndata(t6, acc4);
0N/A acc4 = vis_faligndata(t5, acc4);
0N/A acc4 = vis_faligndata(t4, acc4);
0N/A vis_alignaddr(dp, off);
0N/A s0 = sa[0];
0N/A s1 = sa[1];
0N/A s2 = sa[2];
0N/A s3 = sa[3];
0N/A *dp++ = vis_faligndata(acc0, acc1);
0N/A *dp++ = vis_faligndata(acc1, acc2);
0N/A *dp++ = vis_faligndata(acc2, acc3);
0N/A *dp++ = vis_faligndata(acc3, acc4);
0N/A acc0 = acc4;
0N/A }
0N/A
0N/A vis_alignaddr(dp, 6);
0N/A t3 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 23) & 0x1FE);
0N/A acc1 = vis_faligndata(t3, acc1);
0N/A acc1 = vis_faligndata(t2, acc1);
0N/A acc1 = vis_faligndata(t1, acc1);
0N/A acc1 = vis_faligndata(t0, acc1);
0N/A t7 = VIS_LD_U16_I(table, (s1 << 1) & 0x1FE);
0N/A t6 = VIS_LD_U16_I(table, (s1 >> 7) & 0x1FE);
0N/A t5 = VIS_LD_U16_I(table, (s1 >> 15) & 0x1FE);
0N/A t4 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A acc2 = vis_faligndata(t7, acc2);
0N/A acc2 = vis_faligndata(t6, acc2);
0N/A acc2 = vis_faligndata(t5, acc2);
0N/A acc2 = vis_faligndata(t4, acc2);
0N/A t3 = VIS_LD_U16_I(table, (s2 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s2 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s2 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s2 >> 23) & 0x1FE);
0N/A acc3 = vis_faligndata(t3, acc3);
0N/A acc3 = vis_faligndata(t2, acc3);
0N/A acc3 = vis_faligndata(t1, acc3);
0N/A acc3 = vis_faligndata(t0, acc3);
0N/A t7 = VIS_LD_U16_I(table, (s3 << 1) & 0x1FE);
0N/A t6 = VIS_LD_U16_I(table, (s3 >> 7) & 0x1FE);
0N/A t5 = VIS_LD_U16_I(table, (s3 >> 15) & 0x1FE);
0N/A t4 = VIS_LD_U16_I(table, (s3 >> 23) & 0x1FE);
0N/A acc4 = vis_faligndata(t7, acc4);
0N/A acc4 = vis_faligndata(t6, acc4);
0N/A acc4 = vis_faligndata(t5, acc4);
0N/A acc4 = vis_faligndata(t4, acc4);
0N/A vis_alignaddr(dp, off);
0N/A *dp++ = vis_faligndata(acc0, acc1);
0N/A *dp++ = vis_faligndata(acc1, acc2);
0N/A *dp++ = vis_faligndata(acc2, acc3);
0N/A *dp++ = vis_faligndata(acc3, acc4);
0N/A acc0 = acc4; i+=16;
0N/A }
0N/A
0N/A if (i <= xsize - 8) {
0N/A s0 = sa[0];
0N/A s1 = sa[1];
0N/A vis_alignaddr(dp, 6);
0N/A t3 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 23) & 0x1FE);
0N/A acc1 = vis_faligndata(t3, acc1);
0N/A acc1 = vis_faligndata(t2, acc1);
0N/A acc1 = vis_faligndata(t1, acc1);
0N/A acc1 = vis_faligndata(t0, acc1);
0N/A t7 = VIS_LD_U16_I(table, (s1 << 1) & 0x1FE);
0N/A t6 = VIS_LD_U16_I(table, (s1 >> 7) & 0x1FE);
0N/A t5 = VIS_LD_U16_I(table, (s1 >> 15) & 0x1FE);
0N/A t4 = VIS_LD_U16_I(table, (s1 >> 23) & 0x1FE);
0N/A acc2 = vis_faligndata(t7, acc2);
0N/A acc2 = vis_faligndata(t6, acc2);
0N/A acc2 = vis_faligndata(t5, acc2);
0N/A acc2 = vis_faligndata(t4, acc2);
0N/A vis_alignaddr(dp, off);
0N/A *dp++ = vis_faligndata(acc0, acc1);
0N/A *dp++ = vis_faligndata(acc1, acc2);
0N/A acc0 = acc2; i += 8; sa += 2;
0N/A }
0N/A
0N/A if (i <= xsize - 4) {
0N/A s0 = *sa++;
0N/A vis_alignaddr(dp, 6);
0N/A t3 = VIS_LD_U16_I(table, (s0 << 1) & 0x1FE);
0N/A t2 = VIS_LD_U16_I(table, (s0 >> 7) & 0x1FE);
0N/A t1 = VIS_LD_U16_I(table, (s0 >> 15) & 0x1FE);
0N/A t0 = VIS_LD_U16_I(table, (s0 >> 23) & 0x1FE);
0N/A acc1 = vis_faligndata(t3, acc1);
0N/A acc1 = vis_faligndata(t2, acc1);
0N/A acc1 = vis_faligndata(t1, acc1);
0N/A acc1 = vis_faligndata(t0, acc1);
0N/A vis_alignaddr(dp, off);
0N/A *dp++ = vis_faligndata(acc0, acc1);
0N/A acc0 = acc1;
0N/A }
0N/A
0N/A sp = (mlib_u8*)sa;
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A
0N/A num = (((mlib_u8*) dend - (mlib_u8*) dp) + off + 1) >> 1;
0N/A sp += (num - 1);
0N/A vis_alignaddr(dp, 6);
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < num; i ++) {
0N/A s0 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U16_I(table, 2*s0);
0N/A acc1 = vis_faligndata(t0, acc1);
0N/A }
0N/A
0N/A vis_alignaddr(dp, off);
0N/A emask = vis_edge8(dp, dend);
0N/A vis_pst_8(vis_faligndata(acc0, acc1), dp++, emask);
0N/A }
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A emask = vis_edge8(dp, dend);
0N/A vis_pst_8(vis_faligndata(acc1, acc1), dp++, emask);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2_DstA8D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1, s2, s3; /* source data */
0N/A mlib_u16 *dl; /* pointer to start of destination */
0N/A mlib_u16 *dend; /* pointer to end of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7, acc; /* destination data */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 i, num; /* loop variable */
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A
0N/A sp = (void *)src;
0N/A dl = (mlib_u16*)dst;
0N/A dp = (mlib_d64 *) dl;
0N/A dend = dl + xsize - 1;
0N/A
0N/A vis_alignaddr((void *) 0, 7);
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s0 = sp[0];
0N/A s1 = sp[1];
0N/A s2 = sp[2];
0N/A s3 = sp[3];
0N/A sp += 4;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, sp+=4) {
0N/A t7 = VIS_LD_U8_I(tab1, s3);
0N/A t6 = VIS_LD_U8_I(tab0, s3);
0N/A t5 = VIS_LD_U8_I(tab1, s2);
0N/A t4 = VIS_LD_U8_I(tab0, s2);
0N/A t3 = VIS_LD_U8_I(tab1, s1);
0N/A t2 = VIS_LD_U8_I(tab0, s1);
0N/A t1 = VIS_LD_U8_I(tab1, s0);
0N/A t0 = VIS_LD_U8_I(tab0, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = sp[0];
0N/A s1 = sp[1];
0N/A s2 = sp[2];
0N/A s3 = sp[3];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t7 = VIS_LD_U8_I(tab1, s3);
0N/A t6 = VIS_LD_U8_I(tab0, s3);
0N/A t5 = VIS_LD_U8_I(tab1, s2);
0N/A t4 = VIS_LD_U8_I(tab0, s2);
0N/A t3 = VIS_LD_U8_I(tab1, s1);
0N/A t2 = VIS_LD_U8_I(tab0, s1);
0N/A t1 = VIS_LD_U8_I(tab1, s0);
0N/A t0 = VIS_LD_U8_I(tab0, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A
0N/A num = (mlib_u16*) dend - (mlib_u16*) dp;
0N/A sp += num;
0N/A num ++;
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < num; i ++) {
0N/A s0 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U8_I(tab1, s0);
0N/A acc = vis_faligndata(t0, acc);
0N/A
0N/A t0 = VIS_LD_U8_I(tab0, s0);
0N/A acc = vis_faligndata(t0, acc);
0N/A }
0N/A
0N/A emask = vis_edge16(dp, dend);
0N/A vis_pst_16(acc, dp, emask);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1, s2, s3, s4; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_u8 *dend; /* pointer to end of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7, acc; /* destination data */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 i, num; /* loop variable */
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A
0N/A sp = (void *)src;
0N/A dl = dst;
0N/A
0N/A dend = dl + 2 * xsize - 1;
0N/A
0N/A vis_alignaddr((void *) 0, 7);
0N/A
0N/A s0 = *sp++;
0N/A *dl++ = tab0[s0];
0N/A dp = (mlib_d64 *) dl;
0N/A xsize--;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A s3 = sp[2];
0N/A s4 = sp[3];
0N/A sp += 4;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, sp+=4) {
0N/A t7 = VIS_LD_U8_I(tab0, s4);
0N/A t6 = VIS_LD_U8_I(tab1, s3);
0N/A t5 = VIS_LD_U8_I(tab0, s3);
0N/A t4 = VIS_LD_U8_I(tab1, s2);
0N/A t3 = VIS_LD_U8_I(tab0, s2);
0N/A t2 = VIS_LD_U8_I(tab1, s1);
0N/A t1 = VIS_LD_U8_I(tab0, s1);
0N/A t0 = VIS_LD_U8_I(tab1, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s4;
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A s3 = sp[2];
0N/A s4 = sp[3];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t7 = VIS_LD_U8_I(tab0, s4);
0N/A t6 = VIS_LD_U8_I(tab1, s3);
0N/A t5 = VIS_LD_U8_I(tab0, s3);
0N/A t4 = VIS_LD_U8_I(tab1, s2);
0N/A t3 = VIS_LD_U8_I(tab0, s2);
0N/A t2 = VIS_LD_U8_I(tab1, s1);
0N/A t1 = VIS_LD_U8_I(tab0, s1);
0N/A t0 = VIS_LD_U8_I(tab1, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s4;
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A num = ((mlib_u8*) dend - (mlib_u8*) dp) >> 1;
0N/A sp += num;
0N/A num ++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < num; i ++) {
0N/A s1 = (mlib_s32) *sp;
0N/A sp --;
0N/A
0N/A t0 = VIS_LD_U8_I(tab1, s1);
0N/A acc = vis_faligndata(t0, acc);
0N/A
0N/A t0 = VIS_LD_U8_I(tab0, s1);
0N/A acc = vis_faligndata(t0, acc);
0N/A }
0N/A
0N/A t0 = VIS_LD_U8_I(tab1, s0);
0N/A acc = vis_faligndata(t0, acc);
0N/A emask = vis_edge8(dp, dend);
0N/A vis_pst_8(acc, dp, emask);
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_2(const mlib_u8 *src,
0N/A mlib_s32 slb,
0N/A mlib_u8 *dst,
0N/A mlib_s32 dlb,
0N/A mlib_s32 xsize,
0N/A mlib_s32 ysize,
0N/A const mlib_u8 **table)
0N/A{
0N/A if ((xsize * ysize) < 650) {
0N/A mlib_u8 *sl;
0N/A mlib_u8 *dl;
0N/A mlib_s32 i, j;
0N/A
0N/A sl = (void *)src;
0N/A dl = dst;
0N/A
0N/A /* row loop */
0N/A for (j = 0; j < ysize; j ++) {
0N/A mlib_u8 *sp = sl;
0N/A mlib_u8 *dp = dl;
0N/A mlib_s32 off, s0, size = xsize;
0N/A
0N/A off = ((8 - ((mlib_addr)dp & 7)) & 7) >> 1;
0N/A off = (off < size) ? off : size;
0N/A
0N/A for (i = 0; i < off; i++) {
0N/A s0 = *sp++;
0N/A *dp++ = table[0][s0];
0N/A *dp++ = table[1][s0];
0N/A size--;
0N/A }
0N/A
0N/A if (size > 0) {
0N/A
0N/A if (((mlib_addr)dp & 1) == 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_2_DstA8D1_SMALL(sp, dp, size, table);
0N/A } else {
0N/A mlib_v_ImageLookUpSI_U8_U8_2_D1_SMALL(sp, dp, size, table);
0N/A }
0N/A }
0N/A
0N/A sl = (mlib_u8 *) ((mlib_u8 *) sl + slb);
0N/A dl = (mlib_u8 *) ((mlib_u8 *) dl + dlb);
0N/A }
0N/A
0N/A } else {
0N/A mlib_u8 *sl;
0N/A mlib_u8 *dl;
0N/A mlib_u16 tab[256];
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A mlib_s32 i, j, s0, s1, s2;
0N/A
0N/A s0 = tab0[0];
0N/A s1 = tab1[0];
0N/A for (i = 1; i < 256; i++) {
0N/A s2 = (s0 << 8) + s1;
0N/A s0 = tab0[i];
0N/A s1 = tab1[i];
0N/A tab[i-1] = (mlib_u16)s2;
0N/A }
0N/A
0N/A s2 = (s0 << 8) + s1;
0N/A tab[255] = (mlib_u16)s2;
0N/A
0N/A sl = (void *)src;
0N/A dl = dst;
0N/A
0N/A /* row loop */
0N/A for (j = 0; j < ysize; j ++) {
0N/A mlib_u8 *sp = sl;
0N/A mlib_u8 *dp = dl;
0N/A mlib_s32 off, s0, size = xsize;
0N/A
0N/A if (((mlib_addr)dp & 1) == 0) {
0N/A
0N/A off = ((8 - ((mlib_addr)dp & 7)) & 7) >> 1;
0N/A off = (off < size) ? off : size;
0N/A
0N/A for (i = 0; i < off; i++) {
0N/A *(mlib_u16*)dp = tab[(*sp)];
0N/A dp += 2;
0N/A size--; sp++;
0N/A }
0N/A
0N/A if (size > 0) {
0N/A
0N/A off = (mlib_addr)sp & 3;
0N/A
0N/A if (off == 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_2_SrcOff0_D1(sp, dp, size, tab);
0N/A } else if (off == 1) {
0N/A mlib_v_ImageLookUpSI_U8_U8_2_SrcOff1_D1(sp, dp, size, tab);
0N/A } else if (off == 2) {
0N/A mlib_v_ImageLookUpSI_U8_U8_2_SrcOff2_D1(sp, dp, size, tab);
0N/A } else {
0N/A mlib_v_ImageLookUpSI_U8_U8_2_SrcOff3_D1(sp, dp, size, tab);
0N/A }
0N/A }
0N/A
0N/A } else {
0N/A
0N/A off = ((4 - ((mlib_addr)sp & 3)) & 3);
0N/A off = (off < size) ? off : size;
0N/A
0N/A for (i = 0; i < off; i++) {
0N/A s0 = tab[(*sp)];
0N/A *dp++ = (s0 >> 8);
0N/A *dp++ = (s0 & 0xFF);
0N/A size--; sp++;
0N/A }
0N/A
0N/A if (size > 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_2_DstNonAl_D1(sp, dp, size, tab);
0N/A }
0N/A }
0N/A
0N/A sl = (mlib_u8 *) ((mlib_u8 *) sl + slb);
0N/A dl = (mlib_u8 *) ((mlib_u8 *) dl + dlb);
0N/A }
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_3_SrcOff0_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u32 s0; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_f32 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2, t3; /* destination data */
0N/A mlib_d64 acc0, acc1; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A mlib_u8 *ptr;
0N/A
0N/A dl = dst;
0N/A dp = (mlib_f32 *) dl;
0N/A sp = (void *)src;
0N/A sa = (mlib_u32*)sp;
0N/A
0N/A vis_alignaddr((void *) 0, 3);
0N/A
0N/A i = 0;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s0 = *sa++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, dp+=3) {
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 21) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 13) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 5) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A s0 = *sa++;
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A }
0N/A
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 21) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 13) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 5) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A dp += 3;
0N/A i += 4;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (; i < xsize; i++) {
0N/A ptr = (mlib_u8*)(table + src[i]);
0N/A dl[0] = ptr[0];
0N/A dl[1] = ptr[1];
0N/A dl[2] = ptr[2];
0N/A dl += 3;
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_3_SrcOff1_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u32 s0, s1; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_f32 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2, t3; /* destination data */
0N/A mlib_d64 acc0, acc1; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A mlib_u8 *ptr;
0N/A
0N/A dl = dst;
0N/A dp = (mlib_f32 *) dl;
0N/A sp = (void *)src;
0N/A sa = (mlib_u32*)(sp - 1);
0N/A
0N/A vis_alignaddr((void *) 0, 3);
0N/A
0N/A i = 0;
0N/A s0 = *sa++;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s1 = *sa++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, dp+=3) {
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 13) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 5) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 21) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A s0 = s1;
0N/A s1 = *sa++;
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A }
0N/A
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 13) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 5) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 21) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A dp += 3;
0N/A i += 4;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (; i < xsize; i++) {
0N/A ptr = (mlib_u8*)(table + src[i]);
0N/A dl[0] = ptr[0];
0N/A dl[1] = ptr[1];
0N/A dl[2] = ptr[2];
0N/A dl += 3;
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_3_SrcOff2_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u32 s0, s1; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_f32 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2, t3; /* destination data */
0N/A mlib_d64 acc0, acc1; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A mlib_u8 *ptr;
0N/A
0N/A dl = dst;
0N/A dp = (mlib_f32 *) dl;
0N/A sp = (void *)src;
0N/A sa = (mlib_u32*)(sp - 2);
0N/A
0N/A vis_alignaddr((void *) 0, 3);
0N/A
0N/A i = 0;
0N/A s0 = *sa++;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s1 = *sa++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, dp+=3) {
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 5) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 21) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 13) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A s0 = s1;
0N/A s1 = *sa++;
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A }
0N/A
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 >> 5) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 21) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 13) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A dp += 3;
0N/A i += 4;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (; i < xsize; i++) {
0N/A ptr = (mlib_u8*)(table + src[i]);
0N/A dl[0] = ptr[0];
0N/A dl[1] = ptr[1];
0N/A dl[2] = ptr[2];
0N/A dl += 3;
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_3_SrcOff3_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_d64 *table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u32 s0, s1; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_f32 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2, t3; /* destination data */
0N/A mlib_d64 acc0, acc1; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A mlib_u8 *ptr;
0N/A
0N/A dl = dst;
0N/A dp = (mlib_f32 *) dl;
0N/A sp = (void *)src;
0N/A sa = (mlib_u32*)(sp - 3);
0N/A
0N/A vis_alignaddr((void *) 0, 3);
0N/A
0N/A i = 0;
0N/A s0 = *sa++;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s1 = *sa++;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, dp+=3) {
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 21) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 13) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 5) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A s0 = s1;
0N/A s1 = *sa++;
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A }
0N/A
0N/A t0 = *(mlib_d64*)((mlib_u8*)table + ((s0 << 3) & 0x7F8 ));
0N/A t1 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 21) & 0x7F8 ));
0N/A t2 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 13) & 0x7F8 ));
0N/A t3 = *(mlib_d64*)((mlib_u8*)table + ((s1 >> 5) & 0x7F8 ));
0N/A acc0 = vis_faligndata(t0, t0);
0N/A acc0 = vis_faligndata(acc0, t1);
0N/A acc1 = vis_faligndata(acc0, acc0);
0N/A acc0 = vis_faligndata(acc0, t2);
0N/A acc1 = vis_faligndata(acc1, acc0);
0N/A acc0 = vis_faligndata(acc0, t3);
0N/A dp[0] = vis_read_lo(acc1);
0N/A dp[1] = vis_read_hi(acc0);
0N/A dp[2] = vis_read_lo(acc0);
0N/A dp += 3;
0N/A i += 4;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (; i < xsize; i++) {
0N/A ptr = (mlib_u8*)(table + src[i]);
0N/A dl[0] = ptr[0];
0N/A dl[1] = ptr[1];
0N/A dl[2] = ptr[2];
0N/A dl += 3;
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_3_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7; /* destination data */
0N/A mlib_d64 acc0, acc1, acc2; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A mlib_u32 s00, s01, s02, s03;
0N/A mlib_u32 s10, s11, s12, s13;
0N/A
0N/A sp = (void *)src;
0N/A dl = dst;
0N/A dp = (mlib_d64 *) dl;
0N/A
0N/A vis_alignaddr((void *) 0, 7);
0N/A
0N/A i = 0;
0N/A
0N/A if (xsize >= 8) {
0N/A
0N/A s00 = sp[0];
0N/A s01 = sp[1];
0N/A s02 = sp[2];
0N/A s03 = sp[3];
0N/A s10 = sp[4];
0N/A s11 = sp[5];
0N/A s12 = sp[6];
0N/A s13 = sp[7];
0N/A sp += 8;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 16; i+=8, sp+=8) {
0N/A t7 = VIS_LD_U8_I(tab1, s02);
0N/A t6 = VIS_LD_U8_I(tab0, s02);
0N/A t5 = VIS_LD_U8_I(tab2, s01);
0N/A t4 = VIS_LD_U8_I(tab1, s01);
0N/A t3 = VIS_LD_U8_I(tab0, s01);
0N/A t2 = VIS_LD_U8_I(tab2, s00);
0N/A t1 = VIS_LD_U8_I(tab1, s00);
0N/A t0 = VIS_LD_U8_I(tab0, s00);
0N/A acc0 = vis_faligndata(t7, acc0);
0N/A acc0 = vis_faligndata(t6, acc0);
0N/A acc0 = vis_faligndata(t5, acc0);
0N/A acc0 = vis_faligndata(t4, acc0);
0N/A acc0 = vis_faligndata(t3, acc0);
0N/A acc0 = vis_faligndata(t2, acc0);
0N/A acc0 = vis_faligndata(t1, acc0);
0N/A acc0 = vis_faligndata(t0, acc0);
0N/A t7 = VIS_LD_U8_I(tab0, s11);
0N/A t6 = VIS_LD_U8_I(tab2, s10);
0N/A t5 = VIS_LD_U8_I(tab1, s10);
0N/A t4 = VIS_LD_U8_I(tab0, s10);
0N/A t3 = VIS_LD_U8_I(tab2, s03);
0N/A t2 = VIS_LD_U8_I(tab1, s03);
0N/A t1 = VIS_LD_U8_I(tab0, s03);
0N/A t0 = VIS_LD_U8_I(tab2, s02);
0N/A acc1 = vis_faligndata(t7, acc1);
0N/A acc1 = vis_faligndata(t6, acc1);
0N/A acc1 = vis_faligndata(t5, acc1);
0N/A acc1 = vis_faligndata(t4, acc1);
0N/A acc1 = vis_faligndata(t3, acc1);
0N/A acc1 = vis_faligndata(t2, acc1);
0N/A acc1 = vis_faligndata(t1, acc1);
0N/A acc1 = vis_faligndata(t0, acc1);
0N/A t7 = VIS_LD_U8_I(tab2, s13);
0N/A t6 = VIS_LD_U8_I(tab1, s13);
0N/A t5 = VIS_LD_U8_I(tab0, s13);
0N/A t4 = VIS_LD_U8_I(tab2, s12);
0N/A t3 = VIS_LD_U8_I(tab1, s12);
0N/A t2 = VIS_LD_U8_I(tab0, s12);
0N/A t1 = VIS_LD_U8_I(tab2, s11);
0N/A t0 = VIS_LD_U8_I(tab1, s11);
0N/A acc2 = vis_faligndata(t7, acc2);
0N/A acc2 = vis_faligndata(t6, acc2);
0N/A acc2 = vis_faligndata(t5, acc2);
0N/A acc2 = vis_faligndata(t4, acc2);
0N/A acc2 = vis_faligndata(t3, acc2);
0N/A acc2 = vis_faligndata(t2, acc2);
0N/A acc2 = vis_faligndata(t1, acc2);
0N/A acc2 = vis_faligndata(t0, acc2);
0N/A s00 = sp[0];
0N/A s01 = sp[1];
0N/A s02 = sp[2];
0N/A s03 = sp[3];
0N/A s10 = sp[4];
0N/A s11 = sp[5];
0N/A s12 = sp[6];
0N/A s13 = sp[7];
0N/A *dp++ = acc0;
0N/A *dp++ = acc1;
0N/A *dp++ = acc2;
0N/A }
0N/A
0N/A t7 = VIS_LD_U8_I(tab1, s02);
0N/A t6 = VIS_LD_U8_I(tab0, s02);
0N/A t5 = VIS_LD_U8_I(tab2, s01);
0N/A t4 = VIS_LD_U8_I(tab1, s01);
0N/A t3 = VIS_LD_U8_I(tab0, s01);
0N/A t2 = VIS_LD_U8_I(tab2, s00);
0N/A t1 = VIS_LD_U8_I(tab1, s00);
0N/A t0 = VIS_LD_U8_I(tab0, s00);
0N/A acc0 = vis_faligndata(t7, acc0);
0N/A acc0 = vis_faligndata(t6, acc0);
0N/A acc0 = vis_faligndata(t5, acc0);
0N/A acc0 = vis_faligndata(t4, acc0);
0N/A acc0 = vis_faligndata(t3, acc0);
0N/A acc0 = vis_faligndata(t2, acc0);
0N/A acc0 = vis_faligndata(t1, acc0);
0N/A acc0 = vis_faligndata(t0, acc0);
0N/A t7 = VIS_LD_U8_I(tab0, s11);
0N/A t6 = VIS_LD_U8_I(tab2, s10);
0N/A t5 = VIS_LD_U8_I(tab1, s10);
0N/A t4 = VIS_LD_U8_I(tab0, s10);
0N/A t3 = VIS_LD_U8_I(tab2, s03);
0N/A t2 = VIS_LD_U8_I(tab1, s03);
0N/A t1 = VIS_LD_U8_I(tab0, s03);
0N/A t0 = VIS_LD_U8_I(tab2, s02);
0N/A acc1 = vis_faligndata(t7, acc1);
0N/A acc1 = vis_faligndata(t6, acc1);
0N/A acc1 = vis_faligndata(t5, acc1);
0N/A acc1 = vis_faligndata(t4, acc1);
0N/A acc1 = vis_faligndata(t3, acc1);
0N/A acc1 = vis_faligndata(t2, acc1);
0N/A acc1 = vis_faligndata(t1, acc1);
0N/A acc1 = vis_faligndata(t0, acc1);
0N/A t7 = VIS_LD_U8_I(tab2, s13);
0N/A t6 = VIS_LD_U8_I(tab1, s13);
0N/A t5 = VIS_LD_U8_I(tab0, s13);
0N/A t4 = VIS_LD_U8_I(tab2, s12);
0N/A t3 = VIS_LD_U8_I(tab1, s12);
0N/A t2 = VIS_LD_U8_I(tab0, s12);
0N/A t1 = VIS_LD_U8_I(tab2, s11);
0N/A t0 = VIS_LD_U8_I(tab1, s11);
0N/A acc2 = vis_faligndata(t7, acc2);
0N/A acc2 = vis_faligndata(t6, acc2);
0N/A acc2 = vis_faligndata(t5, acc2);
0N/A acc2 = vis_faligndata(t4, acc2);
0N/A acc2 = vis_faligndata(t3, acc2);
0N/A acc2 = vis_faligndata(t2, acc2);
0N/A acc2 = vis_faligndata(t1, acc2);
0N/A acc2 = vis_faligndata(t0, acc2);
0N/A *dp++ = acc0;
0N/A *dp++ = acc1;
0N/A *dp++ = acc2;
0N/A i += 8;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (; i < xsize; i++) {
0N/A s00 = sp[0];
0N/A dl[0] = tab0[s00];
0N/A dl[1] = tab1[s00];
0N/A dl[2] = tab2[s00];
0N/A dl += 3; sp ++;
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_3(const mlib_u8 *src,
0N/A mlib_s32 slb,
0N/A mlib_u8 *dst,
0N/A mlib_s32 dlb,
0N/A mlib_s32 xsize,
0N/A mlib_s32 ysize,
0N/A const mlib_u8 **table)
0N/A{
0N/A if ((xsize * ysize) < 650) {
0N/A mlib_u8 *sl;
0N/A mlib_u8 *dl;
0N/A mlib_s32 i, j;
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A
0N/A sl = (void *)src;
0N/A dl = dst;
0N/A
0N/A /* row loop */
0N/A for (j = 0; j < ysize; j ++) {
0N/A mlib_u8 *sp = sl;
0N/A mlib_u8 *dp = dl;
0N/A mlib_s32 off, s0, size = xsize;
0N/A
0N/A off = (mlib_addr)dp & 7;
0N/A off = (off * 5) & 7;
0N/A off = (off < size) ? off : size;
0N/A
0N/A for (i = 0; i < off; i++) {
0N/A s0 = *sp++;
0N/A *dp++ = tab0[s0];
0N/A *dp++ = tab1[s0];
0N/A *dp++ = tab2[s0];
0N/A size--;
0N/A }
0N/A
0N/A if (size > 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_3_D1_SMALL(sp, dp, size, table);
0N/A }
0N/A
0N/A sl = (mlib_u8 *) ((mlib_u8 *) sl + slb);
0N/A dl = (mlib_u8 *) ((mlib_u8 *) dl + dlb);
0N/A }
0N/A
0N/A } else {
0N/A mlib_u8 *sl;
0N/A mlib_u8 *dl;
0N/A mlib_u32 tab[512];
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A mlib_s32 i, j;
0N/A mlib_u32 s0, s1, s2, s3;
0N/A
0N/A s0 = tab0[0];
0N/A s1 = tab1[0];
0N/A s2 = tab2[0];
0N/A for (i = 1; i < 256; i++) {
0N/A s3 = (s0 << 24) + (s1 << 16) + (s2 << 8);
0N/A s0 = tab0[i];
0N/A s1 = tab1[i];
0N/A s2 = tab2[i];
0N/A tab[2*i-2] = s3;
0N/A }
0N/A
0N/A s3 = (s0 << 24) + (s1 << 16) + (s2 << 8);
0N/A tab[510] = s3;
0N/A
0N/A sl = (void *)src;
0N/A dl = dst;
0N/A
0N/A /* row loop */
0N/A for (j = 0; j < ysize; j ++) {
0N/A mlib_u8 *sp = sl;
0N/A mlib_u8 *dp = dl;
0N/A mlib_s32 off, size = xsize;
0N/A mlib_u8 *ptr;
0N/A
0N/A off = ((mlib_addr)dp & 3);
0N/A off = (off < size) ? off : size;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < off; i++) {
0N/A ptr = (mlib_u8*)(tab + 2*sp[i]);
0N/A dp[0] = ptr[0];
0N/A dp[1] = ptr[1];
0N/A dp[2] = ptr[2];
0N/A dp += 3;
0N/A }
0N/A
0N/A size -= off;
0N/A sp += off;
0N/A
0N/A if (size > 0) {
0N/A off = (mlib_addr)sp & 3;
0N/A
0N/A if (off == 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_3_SrcOff0_D1(sp, dp, size, (mlib_d64*)tab);
0N/A } else if (off == 1) {
0N/A mlib_v_ImageLookUpSI_U8_U8_3_SrcOff1_D1(sp, dp, size, (mlib_d64*)tab);
0N/A } else if (off == 2) {
0N/A mlib_v_ImageLookUpSI_U8_U8_3_SrcOff2_D1(sp, dp, size, (mlib_d64*)tab);
0N/A } else if (off == 3) {
0N/A mlib_v_ImageLookUpSI_U8_U8_3_SrcOff3_D1(sp, dp, size, (mlib_d64*)tab);
0N/A }
0N/A }
0N/A
0N/A sl = (mlib_u8 *) ((mlib_u8 *) sl + slb);
0N/A dl = (mlib_u8 *) ((mlib_u8 *) dl + dlb);
0N/A }
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_4_SrcOff0_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_f32 *table)
0N/A{
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0; /* source data */
0N/A mlib_f32 *dp; /* aligned pointer to destination */
0N/A mlib_f32 acc0, acc1; /* destination data */
0N/A mlib_f32 acc2, acc3; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A mlib_u32 s00, s01, s02, s03;
0N/A
0N/A sa = (mlib_u32*)src;
0N/A dp = (mlib_f32 *) dst;
0N/A
0N/A i = 0;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s0 = *sa++;
0N/A s00 = (s0 >> 22) & 0x3FC;
0N/A s01 = (s0 >> 14) & 0x3FC;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, dp += 4) {
0N/A s02 = (s0 >> 6) & 0x3FC;
0N/A s03 = (s0 << 2) & 0x3FC;
0N/A acc0 = *(mlib_f32*)((mlib_u8*)table + s00);
0N/A acc1 = *(mlib_f32*)((mlib_u8*)table + s01);
0N/A acc2 = *(mlib_f32*)((mlib_u8*)table + s02);
0N/A acc3 = *(mlib_f32*)((mlib_u8*)table + s03);
0N/A s0 = *sa++;
0N/A s00 = (s0 >> 22) & 0x3FC;
0N/A s01 = (s0 >> 14) & 0x3FC;
0N/A dp[0] = acc0;
0N/A dp[1] = acc1;
0N/A dp[2] = acc2;
0N/A dp[3] = acc3;
0N/A }
0N/A
0N/A s02 = (s0 >> 6) & 0x3FC;
0N/A s03 = (s0 << 2) & 0x3FC;
0N/A acc0 = *(mlib_f32*)((mlib_u8*)table + s00);
0N/A acc1 = *(mlib_f32*)((mlib_u8*)table + s01);
0N/A acc2 = *(mlib_f32*)((mlib_u8*)table + s02);
0N/A acc3 = *(mlib_f32*)((mlib_u8*)table + s03);
0N/A dp[0] = acc0;
0N/A dp[1] = acc1;
0N/A dp[2] = acc2;
0N/A dp[3] = acc3;
0N/A dp += 4;
0N/A i += 4;
0N/A }
0N/A
0N/A sp = (mlib_u8*)sa;
0N/A
0N/A if ( i <= xsize - 2) {
0N/A *dp++ = table[sp[0]];
0N/A *dp++ = table[sp[1]];
0N/A i+=2; sp += 2;
0N/A }
0N/A
0N/A if ( i < xsize) *dp = table[sp[0]];
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_4_DstNonAl_D1(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_f32 *table)
0N/A{
0N/A mlib_u32 *sa; /* aligned pointer to source data */
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 acc0, acc1, acc2; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A mlib_u8 *dend; /* pointer to end of destination */
0N/A mlib_s32 emask; /* edge mask */
0N/A mlib_s32 off;
0N/A mlib_u32 s00, s01, s02, s03;
0N/A
0N/A sa = (mlib_u32*)src;
0N/A sp = (void *)src;
0N/A dl = dst;
0N/A dend = dl + (xsize << 2) - 1;
0N/A dp = (mlib_d64 *) ((mlib_addr) dl & (~7));
0N/A off = (mlib_addr) dp - (mlib_addr) dl;
0N/A vis_alignaddr(dp, off);
0N/A
0N/A emask = vis_edge8(dl, dend);
0N/A acc0 = vis_freg_pair(table[sp[0]], table[sp[1]]);
0N/A vis_pst_8(vis_faligndata(acc0, acc0), dp++, emask);
0N/A sp += 2;
0N/A
0N/A xsize -= 2;
0N/A
0N/A if (xsize >= 2) {
0N/A acc1 = vis_freg_pair(table[sp[0]], table[sp[1]]);
0N/A *dp++ = vis_faligndata(acc0, acc1);
0N/A acc0 = acc1;
0N/A sp += 2; xsize -= 2;
0N/A }
0N/A
0N/A sa++;
0N/A
0N/A i = 0;
0N/A
0N/A if (xsize >= 4) {
0N/A
0N/A s0 = *sa++;
0N/A s00 = (s0 >> 22) & 0x3FC;
0N/A s01 = (s0 >> 14) & 0x3FC;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 8; i+=4, dp += 2) {
0N/A s02 = (s0 >> 6) & 0x3FC;
0N/A s03 = (s0 << 2) & 0x3FC;
0N/A acc1 = vis_freg_pair(*(mlib_f32*)((mlib_u8*)table + s00),
0N/A *(mlib_f32*)((mlib_u8*)table + s01));
0N/A acc2 = vis_freg_pair(*(mlib_f32*)((mlib_u8*)table + s02),
0N/A *(mlib_f32*)((mlib_u8*)table + s03));
0N/A s0 = *sa++;
0N/A s00 = (s0 >> 22) & 0x3FC;
0N/A s01 = (s0 >> 14) & 0x3FC;
0N/A dp[0] = vis_faligndata(acc0, acc1);
0N/A dp[1] = vis_faligndata(acc1, acc2);
0N/A acc0 = acc2;
0N/A }
0N/A
0N/A s02 = (s0 >> 6) & 0x3FC;
0N/A s03 = (s0 << 2) & 0x3FC;
0N/A acc1 = vis_freg_pair(*(mlib_f32*)((mlib_u8*)table + s00),
0N/A *(mlib_f32*)((mlib_u8*)table + s01));
0N/A acc2 = vis_freg_pair(*(mlib_f32*)((mlib_u8*)table + s02),
0N/A *(mlib_f32*)((mlib_u8*)table + s03));
0N/A dp[0] = vis_faligndata(acc0, acc1);
0N/A dp[1] = vis_faligndata(acc1, acc2);
0N/A acc0 = acc2;
0N/A sp = (mlib_u8*)sa;
0N/A dp += 2;
0N/A i += 4;
0N/A }
0N/A
0N/A if ( i <= xsize - 2) {
0N/A acc1 = vis_freg_pair(table[sp[0]], table[sp[1]]);
0N/A *dp++ = vis_faligndata(acc0, acc1);
0N/A acc0 = acc1;
0N/A i+=2; sp += 2;
0N/A }
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A emask = vis_edge8(dp, dend);
0N/A acc1 = vis_freg_pair(table[sp[0]], table[sp[1]]);
0N/A vis_pst_8(vis_faligndata(acc0, acc1), dp++, emask);
0N/A }
0N/A
0N/A if ((mlib_addr) dp <= (mlib_addr) dend) {
0N/A emask = vis_edge8(dp, dend);
0N/A vis_pst_8(vis_faligndata(acc1, acc1), dp++, emask);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_4_DstOff0_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7, acc; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A const mlib_u8 *tab3 = table[3];
0N/A
0N/A sp = (void *)src;
0N/A dl = dst;
0N/A dp = (mlib_d64 *) dl;
0N/A
0N/A vis_alignaddr((void *) 0, 7);
0N/A
0N/A if (xsize >= 2) {
0N/A
0N/A s0 = sp[0];
0N/A s1 = sp[1];
0N/A sp += 2;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 4; i+=2, sp+=2) {
0N/A t7 = VIS_LD_U8_I(tab3, s1);
0N/A t6 = VIS_LD_U8_I(tab2, s1);
0N/A t5 = VIS_LD_U8_I(tab1, s1);
0N/A t4 = VIS_LD_U8_I(tab0, s1);
0N/A t3 = VIS_LD_U8_I(tab3, s0);
0N/A t2 = VIS_LD_U8_I(tab2, s0);
0N/A t1 = VIS_LD_U8_I(tab1, s0);
0N/A t0 = VIS_LD_U8_I(tab0, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = sp[0];
0N/A s1 = sp[1];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t7 = VIS_LD_U8_I(tab3, s1);
0N/A t6 = VIS_LD_U8_I(tab2, s1);
0N/A t5 = VIS_LD_U8_I(tab1, s1);
0N/A t4 = VIS_LD_U8_I(tab0, s1);
0N/A t3 = VIS_LD_U8_I(tab3, s0);
0N/A t2 = VIS_LD_U8_I(tab2, s0);
0N/A t1 = VIS_LD_U8_I(tab1, s0);
0N/A t0 = VIS_LD_U8_I(tab0, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A if ((xsize & 1) != 0) {
0N/A s0 = sp[0];
0N/A t7 = VIS_LD_U8_I(tab3, s0);
0N/A t6 = VIS_LD_U8_I(tab2, s0);
0N/A t5 = VIS_LD_U8_I(tab1, s0);
0N/A t4 = VIS_LD_U8_I(tab0, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A *(mlib_f32*)dp = vis_read_hi(acc);
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_4_DstOff1_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1, s2; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7, acc; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A const mlib_u8 *tab3 = table[3];
0N/A
0N/A sp = (void *)src;
0N/A dl = dst;
0N/A dp = (mlib_d64 *) dl;
0N/A
0N/A vis_alignaddr((void *) 0, 7);
0N/A
0N/A s0 = *sp++;
0N/A
0N/A if (xsize >= 2) {
0N/A
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A sp += 2;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 4; i+=2, sp+=2) {
0N/A t7 = VIS_LD_U8_I(tab0, s2);
0N/A t6 = VIS_LD_U8_I(tab3, s1);
0N/A t5 = VIS_LD_U8_I(tab2, s1);
0N/A t4 = VIS_LD_U8_I(tab1, s1);
0N/A t3 = VIS_LD_U8_I(tab0, s1);
0N/A t2 = VIS_LD_U8_I(tab3, s0);
0N/A t1 = VIS_LD_U8_I(tab2, s0);
0N/A t0 = VIS_LD_U8_I(tab1, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s2;
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t7 = VIS_LD_U8_I(tab0, s2);
0N/A t6 = VIS_LD_U8_I(tab3, s1);
0N/A t5 = VIS_LD_U8_I(tab2, s1);
0N/A t4 = VIS_LD_U8_I(tab1, s1);
0N/A t3 = VIS_LD_U8_I(tab0, s1);
0N/A t2 = VIS_LD_U8_I(tab3, s0);
0N/A t1 = VIS_LD_U8_I(tab2, s0);
0N/A t0 = VIS_LD_U8_I(tab1, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s2;
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A if ((xsize & 1) != 0) {
0N/A s1 = sp[0];
0N/A t7 = VIS_LD_U8_I(tab0, s1);
0N/A t6 = VIS_LD_U8_I(tab3, s0);
0N/A t5 = VIS_LD_U8_I(tab2, s0);
0N/A t4 = VIS_LD_U8_I(tab1, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A *(mlib_f32*)dl = vis_read_hi(acc);
0N/A dl += 4;
0N/A s0 = s1;
0N/A }
0N/A
0N/A dl[0] = tab1[s0];
0N/A dl[1] = tab2[s0];
0N/A dl[2] = tab3[s0];
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_4_DstOff2_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1, s2; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7, acc; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A const mlib_u8 *tab3 = table[3];
0N/A
0N/A sp = (void *)src;
0N/A dl = dst;
0N/A dp = (mlib_d64 *) dl;
0N/A
0N/A vis_alignaddr((void *) 0, 7);
0N/A
0N/A s0 = *sp++;
0N/A
0N/A if (xsize >= 2) {
0N/A
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A sp += 2;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 4; i+=2, sp+=2) {
0N/A t7 = VIS_LD_U8_I(tab1, s2);
0N/A t6 = VIS_LD_U8_I(tab0, s2);
0N/A t5 = VIS_LD_U8_I(tab3, s1);
0N/A t4 = VIS_LD_U8_I(tab2, s1);
0N/A t3 = VIS_LD_U8_I(tab1, s1);
0N/A t2 = VIS_LD_U8_I(tab0, s1);
0N/A t1 = VIS_LD_U8_I(tab3, s0);
0N/A t0 = VIS_LD_U8_I(tab2, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s2;
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t7 = VIS_LD_U8_I(tab1, s2);
0N/A t6 = VIS_LD_U8_I(tab0, s2);
0N/A t5 = VIS_LD_U8_I(tab3, s1);
0N/A t4 = VIS_LD_U8_I(tab2, s1);
0N/A t3 = VIS_LD_U8_I(tab1, s1);
0N/A t2 = VIS_LD_U8_I(tab0, s1);
0N/A t1 = VIS_LD_U8_I(tab3, s0);
0N/A t0 = VIS_LD_U8_I(tab2, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s2;
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A if ((xsize & 1) != 0) {
0N/A s1 = sp[0];
0N/A t7 = VIS_LD_U8_I(tab1, s1);
0N/A t6 = VIS_LD_U8_I(tab0, s1);
0N/A t5 = VIS_LD_U8_I(tab3, s0);
0N/A t4 = VIS_LD_U8_I(tab2, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A *(mlib_f32*)dl = vis_read_hi(acc);
0N/A dl += 4;
0N/A s0 = s1;
0N/A }
0N/A
0N/A dl[0] = tab2[s0];
0N/A dl[1] = tab3[s0];
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_4_DstOff3_D1_SMALL(const mlib_u8 *src,
0N/A mlib_u8 *dst,
0N/A mlib_s32 xsize,
0N/A const mlib_u8 **table)
0N/A{
0N/A mlib_u8 *sp; /* pointer to source data */
0N/A mlib_u32 s0, s1, s2; /* source data */
0N/A mlib_u8 *dl; /* pointer to start of destination */
0N/A mlib_d64 *dp; /* aligned pointer to destination */
0N/A mlib_d64 t0, t1, t2; /* destination data */
0N/A mlib_d64 t3, t4, t5; /* destination data */
0N/A mlib_d64 t6, t7, acc; /* destination data */
0N/A mlib_s32 i; /* loop variable */
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A const mlib_u8 *tab3 = table[3];
0N/A
0N/A sp = (void *)src;
0N/A dl = dst;
0N/A dp = (mlib_d64 *) dl;
0N/A
0N/A vis_alignaddr((void *) 0, 7);
0N/A
0N/A s0 = *sp++;
0N/A
0N/A if (xsize >= 2) {
0N/A
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A sp += 2;
0N/A
0N/A#pragma pipeloop(0)
0N/A for(i = 0; i <= xsize - 4; i+=2, sp+=2) {
0N/A t7 = VIS_LD_U8_I(tab2, s2);
0N/A t6 = VIS_LD_U8_I(tab1, s2);
0N/A t5 = VIS_LD_U8_I(tab0, s2);
0N/A t4 = VIS_LD_U8_I(tab3, s1);
0N/A t3 = VIS_LD_U8_I(tab2, s1);
0N/A t2 = VIS_LD_U8_I(tab1, s1);
0N/A t1 = VIS_LD_U8_I(tab0, s1);
0N/A t0 = VIS_LD_U8_I(tab3, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s2;
0N/A s1 = sp[0];
0N/A s2 = sp[1];
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A t7 = VIS_LD_U8_I(tab2, s2);
0N/A t6 = VIS_LD_U8_I(tab1, s2);
0N/A t5 = VIS_LD_U8_I(tab0, s2);
0N/A t4 = VIS_LD_U8_I(tab3, s1);
0N/A t3 = VIS_LD_U8_I(tab2, s1);
0N/A t2 = VIS_LD_U8_I(tab1, s1);
0N/A t1 = VIS_LD_U8_I(tab0, s1);
0N/A t0 = VIS_LD_U8_I(tab3, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A acc = vis_faligndata(t3, acc);
0N/A acc = vis_faligndata(t2, acc);
0N/A acc = vis_faligndata(t1, acc);
0N/A acc = vis_faligndata(t0, acc);
0N/A s0 = s2;
0N/A *dp++ = acc;
0N/A }
0N/A
0N/A dl = (mlib_u8*)dp;
0N/A
0N/A if ((xsize & 1) != 0) {
0N/A s1 = sp[0];
0N/A t7 = VIS_LD_U8_I(tab2, s1);
0N/A t6 = VIS_LD_U8_I(tab1, s1);
0N/A t5 = VIS_LD_U8_I(tab0, s1);
0N/A t4 = VIS_LD_U8_I(tab3, s0);
0N/A acc = vis_faligndata(t7, acc);
0N/A acc = vis_faligndata(t6, acc);
0N/A acc = vis_faligndata(t5, acc);
0N/A acc = vis_faligndata(t4, acc);
0N/A *(mlib_f32*)dl = vis_read_hi(acc);
0N/A dl += 4;
0N/A s0 = s1;
0N/A }
0N/A
0N/A dl[0] = tab3[s0];
0N/A}
0N/A
0N/A/***************************************************************/
0N/Avoid mlib_v_ImageLookUpSI_U8_U8_4(const mlib_u8 *src,
0N/A mlib_s32 slb,
0N/A mlib_u8 *dst,
0N/A mlib_s32 dlb,
0N/A mlib_s32 xsize,
0N/A mlib_s32 ysize,
0N/A const mlib_u8 **table)
0N/A{
0N/A if ((xsize * ysize) < 500) {
0N/A mlib_u8 *sl;
0N/A mlib_u8 *dl;
0N/A mlib_s32 j;
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A const mlib_u8 *tab3 = table[3];
0N/A
0N/A sl = (void *)src;
0N/A dl = dst;
0N/A
0N/A /* row loop */
0N/A for (j = 0; j < ysize; j ++) {
0N/A mlib_u8 *sp = sl;
0N/A mlib_u8 *dp = dl;
0N/A mlib_s32 off, s0, size = xsize;
0N/A
0N/A off = (8 - ((mlib_addr)dp & 7)) & 7;
0N/A
0N/A if ((off >= 4) && (size > 0)) {
0N/A s0 = *sp++;
0N/A *dp++ = tab0[s0];
0N/A *dp++ = tab1[s0];
0N/A *dp++ = tab2[s0];
0N/A *dp++ = tab3[s0];
0N/A size--;
0N/A }
0N/A
0N/A if (size > 0) {
0N/A off = (4 - ((mlib_addr)dp & 3)) & 3;
0N/A
0N/A if (off == 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_4_DstOff0_D1_SMALL(sp, dp, size, table);
0N/A } else if (off == 1) {
0N/A s0 = *sp;
0N/A *dp++ = tab0[s0];
0N/A size--;
0N/A mlib_v_ImageLookUpSI_U8_U8_4_DstOff1_D1_SMALL(sp, dp, size, table);
0N/A } else if (off == 2) {
0N/A s0 = *sp;
0N/A *dp++ = tab0[s0];
0N/A *dp++ = tab1[s0];
0N/A size--;
0N/A mlib_v_ImageLookUpSI_U8_U8_4_DstOff2_D1_SMALL(sp, dp, size, table);
0N/A } else if (off == 3) {
0N/A s0 = *sp;
0N/A *dp++ = tab0[s0];
0N/A *dp++ = tab1[s0];
0N/A *dp++ = tab2[s0];
0N/A size--;
0N/A mlib_v_ImageLookUpSI_U8_U8_4_DstOff3_D1_SMALL(sp, dp, size, table);
0N/A }
0N/A }
0N/A
0N/A sl = (mlib_u8 *) ((mlib_u8 *) sl + slb);
0N/A dl = (mlib_u8 *) ((mlib_u8 *) dl + dlb);
0N/A }
0N/A
0N/A } else {
0N/A mlib_u8 *sl;
0N/A mlib_u8 *dl;
0N/A mlib_u32 tab[256];
0N/A const mlib_u8 *tab0 = table[0];
0N/A const mlib_u8 *tab1 = table[1];
0N/A const mlib_u8 *tab2 = table[2];
0N/A const mlib_u8 *tab3 = table[3];
0N/A mlib_s32 i, j;
0N/A mlib_u32 s0, s1, s2, s3, s4;
0N/A
0N/A s0 = tab0[0];
0N/A s1 = tab1[0];
0N/A s2 = tab2[0];
0N/A s3 = tab3[0];
0N/A for (i = 1; i < 256; i++) {
0N/A s4 = (s0 << 24) + (s1 << 16) + (s2 << 8) + s3;
0N/A s0 = tab0[i];
0N/A s1 = tab1[i];
0N/A s2 = tab2[i];
0N/A s3 = tab3[i];
0N/A tab[i-1] = s4;
0N/A }
0N/A
0N/A s4 = (s0 << 24) + (s1 << 16) + (s2 << 8) + s3;
0N/A tab[255] = s4;
0N/A
0N/A sl = (void *)src;
0N/A dl = dst;
0N/A
0N/A /* row loop */
0N/A for (j = 0; j < ysize; j ++) {
0N/A mlib_u8 *sp = sl;
0N/A mlib_u8 *dp = dl;
0N/A mlib_s32 off, size = xsize;
0N/A
0N/A if (((mlib_addr)dp & 3) == 0) {
0N/A off = (4 - (mlib_addr)sp & 3) & 3;
0N/A
0N/A off = (off < size) ? off : size;
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < off; i++) {
0N/A *(mlib_u32*)dp = tab[(*sp)];
0N/A dp += 4; sp++;
0N/A }
0N/A
0N/A size -= off;
0N/A
0N/A if (size > 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_4_SrcOff0_D1(sp, dp, size, (mlib_f32*)tab);
0N/A }
0N/A
0N/A } else {
0N/A
0N/A off = ((4 - ((mlib_addr)sp & 3)) & 3);
0N/A off = (off < size) ? off : size;
0N/A
0N/A for (i = 0; i < off; i++) {
0N/A s0 = tab[(*sp)];
0N/A *dp++ = (s0 >> 24);
0N/A *dp++ = (s0 >> 16);
0N/A *dp++ = (s0 >> 8);
0N/A *dp++ = s0;
0N/A size--; sp++;
0N/A }
0N/A
0N/A if (size > 0) {
0N/A mlib_v_ImageLookUpSI_U8_U8_4_DstNonAl_D1(sp, dp, size, (mlib_f32*)tab);
0N/A }
0N/A }
0N/A
0N/A sl = (mlib_u8 *) ((mlib_u8 *) sl + slb);
0N/A dl = (mlib_u8 *) ((mlib_u8 *) dl + dlb);
0N/A }
0N/A }
0N/A}
0N/A
0N/A/***************************************************************/