Cemu/dependencies/ih264d/common/armv8/ih264_weighted_bi_pred_av8.s
2022-08-22 22:21:23 +02:00

574 lines
28 KiB
ArmAsm

//******************************************************************************
//*
//* Copyright (C) 2015 The Android Open Source Project
//*
//* Licensed under the Apache License, Version 2.0 (the "License");
//* you may not use this file except in compliance with the License.
//* You may obtain a copy of the License at:
//*
//* http://www.apache.org/licenses/LICENSE-2.0
//*
//* Unless required by applicable law or agreed to in writing, software
//* distributed under the License is distributed on an "AS IS" BASIS,
//* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
//* See the License for the specific language governing permissions and
//* limitations under the License.
//*
//*****************************************************************************
//* Originally developed and contributed by Ittiam Systems Pvt. Ltd, Bangalore
//*/
///**
//******************************************************************************
//* @file
//* ih264_weighted_bi_pred_av8.s
//*
//* @brief
//* Contains function definitions for weighted biprediction.
//*
//* @author
//* Kaushik Senthoor R
//*
//* @par List of Functions:
//*
//* - ih264_weighted_bi_pred_luma_av8()
//* - ih264_weighted_bi_pred_chroma_av8()
//*
//* @remarks
//* None
//*
//*******************************************************************************
//*/
//*******************************************************************************
//* @function
//* ih264_weighted_bi_pred_luma_av8()
//*
//* @brief
//* This routine performs the default weighted prediction as described in sec
//* 8.4.2.3.2 titled "Weighted sample prediction process" for luma.
//*
//* @par Description:
//* This function gets two ht x wd blocks, calculates the weighted samples,
//* rounds off, adds offset and stores it in the destination block.
//*
//* @param[in] puc_src1
//* UWORD8 Pointer to the buffer containing the input block 1.
//*
//* @param[in] puc_src2
//* UWORD8 Pointer to the buffer containing the input block 2.
//*
//* @param[out] puc_dst
//* UWORD8 pointer to the destination where the output block is stored.
//*
//* @param[in] src_strd1
//* Stride of the input buffer 1
//*
//* @param[in] src_strd2
//* Stride of the input buffer 2
//*
//* @param[in] dst_strd
//* Stride of the destination buffer
//*
//* @param[in] log_WD
//* number of bits to be rounded off
//*
//* @param[in] wt1
//* weight for the weighted prediction
//*
//* @param[in] wt2
//* weight for the weighted prediction
//*
//* @param[in] ofst1
//* offset 1 used after rounding off
//*
//* @param[in] ofst2
//* offset 2 used after rounding off
//*
//* @param[in] ht
//* integer height of the array
//*
//* @param[in] wd
//* integer width of the array
//*
//* @returns
//* None
//*
//* @remarks
//* (ht,wd) can be (4,4), (4,8), (8,4), (8,8), (8,16), (16,8) or (16,16).
//*
//*******************************************************************************
//*/
//void ih264_weighted_bi_pred_luma_av8(UWORD8 *puc_src1,
// UWORD8 *puc_src2,
// UWORD8 *puc_dst,
// WORD32 src_strd1,
// WORD32 src_strd2,
// WORD32 dst_strd,
// WORD32 log_WD,
// WORD32 wt1,
// WORD32 wt2,
// WORD16 ofst1,
// WORD16 ofst2,
// WORD32 ht,
// WORD32 wd)
//
//**************Variables Vs Registers*****************************************
// x0 => puc_src1
// x1 => puc_src2
// x2 => puc_dst
// w3 => src_strd1
// w4 => src_strd2
// w5 => dst_strd
// w6 => log_WD
// w7 => wt1
// [sp] => wt2 (w8)
// [sp+8] => ofst1 (w9)
// [sp+16] => ofst2 (w10)
// [sp+24] => ht (w11)
// [sp+32] => wd (w12)
//
.text
.p2align 2
.include "ih264_neon_macros.s"
.global ih264_weighted_bi_pred_luma_av8
ih264_weighted_bi_pred_luma_av8:
// STMFD sp!, {x4-x12,x14} //stack stores the values of the arguments
push_v_regs
sxtw x3, w3
sxtw x4, w4
sxtw x5, w5
stp x19, x20, [sp, #-16]!
ldr w8, [sp, #80] //Load wt2 in w8
ldr w9, [sp, #88] //Load ofst1 in w9
add w6, w6, #1 //w6 = log_WD + 1
neg w10, w6 //w10 = -(log_WD + 1)
dup v0.8h, w10 //Q0 = -(log_WD + 1) (32-bit)
ldr w10, [sp, #96] //Load ofst2 in w10
ldr w11, [sp, #104] //Load ht in w11
ldr w12, [sp, #112] //Load wd in w12
add w9, w9, #1 //w9 = ofst1 + 1
add w9, w9, w10 //w9 = ofst1 + ofst2 + 1
mov v2.s[0], w7
mov v2.s[1], w8 //D2 = {wt1(32-bit), wt2(32-bit)}
asr w9, w9, #1 //w9 = ofst = (ofst1 + ofst2 + 1) >> 1
dup v3.8b, w9 //D3 = ofst (8-bit)
cmp w12, #16
beq loop_16 //branch if wd is 16
cmp w12, #8 //check if wd is 8
beq loop_8 //branch if wd is 8
loop_4: //each iteration processes four rows
ld1 {v4.s}[0], [x0], x3 //load row 1 in source 1
ld1 {v4.s}[1], [x0], x3 //load row 2 in source 1
ld1 {v6.s}[0], [x1], x4 //load row 1 in source 2
ld1 {v6.s}[1], [x1], x4 //load row 2 in source 2
uxtl v4.8h, v4.8b //converting rows 1,2 in source 1 to 16-bit
ld1 {v8.s}[0], [x0], x3 //load row 3 in source 1
ld1 {v8.s}[1], [x0], x3 //load row 4 in source 1
uxtl v6.8h, v6.8b //converting rows 1,2 in source 2 to 16-bit
ld1 {v10.s}[0], [x1], x4 //load row 3 in source 2
ld1 {v10.s}[1], [x1], x4 //load row 4 in source 2
uxtl v8.8h, v8.8b //converting rows 3,4 in source 1 to 16-bit
uxtl v10.8h, v10.8b //converting rows 3,4 in source 2 to 16-bit
mul v4.8h, v4.8h , v2.h[0] //weight 1 mult. for rows 1,2
mla v4.8h, v6.8h , v2.h[2] //weight 2 mult. for rows 1,2
mul v8.8h, v8.8h , v2.h[0] //weight 1 mult. for rows 3,4
mla v8.8h, v10.8h , v2.h[2] //weight 2 mult. for rows 3,4
subs w11, w11, #4 //decrement ht by 4
srshl v4.8h, v4.8h , v0.8h //rounds off the weighted samples from rows 1,2
srshl v8.8h, v8.8h , v0.8h //rounds off the weighted samples from rows 3,4
saddw v4.8h, v4.8h , v3.8b //adding offset for rows 1,2
saddw v8.8h, v8.8h , v3.8b //adding offset for rows 3,4
sqxtun v4.8b, v4.8h //saturating rows 1,2 to unsigned 8-bit
sqxtun v8.8b, v8.8h //saturating rows 3,4 to unsigned 8-bit
st1 {v4.s}[0], [x2], x5 //store row 1 in destination
st1 {v4.s}[1], [x2], x5 //store row 2 in destination
st1 {v8.s}[0], [x2], x5 //store row 3 in destination
st1 {v8.s}[1], [x2], x5 //store row 4 in destination
bgt loop_4 //if greater than 0 repeat the loop again
b end_loops
loop_8: //each iteration processes four rows
ld1 {v4.8b}, [x0], x3 //load row 1 in source 1
ld1 {v6.8b}, [x1], x4 //load row 1 in source 2
ld1 {v8.8b}, [x0], x3 //load row 2 in source 1
ld1 {v10.8b}, [x1], x4 //load row 2 in source 2
uxtl v4.8h, v4.8b //converting row 1 in source 1 to 16-bit
ld1 {v12.8b}, [x0], x3 //load row 3 in source 1
ld1 {v14.8b}, [x1], x4 //load row 3 in source 2
uxtl v6.8h, v6.8b //converting row 1 in source 2 to 16-bit
ld1 {v16.8b}, [x0], x3 //load row 4 in source 1
ld1 {v18.8b}, [x1], x4 //load row 4 in source 2
uxtl v8.8h, v8.8b //converting row 2 in source 1 to 16-bit
uxtl v10.8h, v10.8b //converting row 2 in source 2 to 16-bit
mul v4.8h, v4.8h , v2.h[0] //weight 1 mult. for row 1
mla v4.8h, v6.8h , v2.h[2] //weight 2 mult. for row 1
uxtl v12.8h, v12.8b //converting row 3 in source 1 to 16-bit
uxtl v14.8h, v14.8b //converting row 3 in source 2 to 16-bit
mul v8.8h, v8.8h , v2.h[0] //weight 1 mult. for row 2
mla v8.8h, v10.8h , v2.h[2] //weight 2 mult. for row 2
uxtl v16.8h, v16.8b //converting row 4 in source 1 to 16-bit
uxtl v18.8h, v18.8b //converting row 4 in source 2 to 16-bit
mul v12.8h, v12.8h , v2.h[0] //weight 1 mult. for row 3
mla v12.8h, v14.8h , v2.h[2] //weight 2 mult. for row 3
mul v16.8h, v16.8h , v2.h[0] //weight 1 mult. for row 4
mla v16.8h, v18.8h , v2.h[2] //weight 2 mult. for row 4
srshl v4.8h, v4.8h , v0.8h //rounds off the weighted samples from row 1
srshl v8.8h, v8.8h , v0.8h //rounds off the weighted samples from row 2
srshl v12.8h, v12.8h , v0.8h //rounds off the weighted samples from row 3
saddw v4.8h, v4.8h , v3.8b //adding offset for row 1
srshl v16.8h, v16.8h , v0.8h //rounds off the weighted samples from row 4
saddw v8.8h, v8.8h , v3.8b //adding offset for row 2
saddw v12.8h, v12.8h , v3.8b //adding offset for row 3
sqxtun v4.8b, v4.8h //saturating row 1 to unsigned 8-bit
saddw v16.8h, v16.8h , v3.8b //adding offset for row 4
sqxtun v8.8b, v8.8h //saturating row 2 to unsigned 8-bit
sqxtun v12.8b, v12.8h //saturating row 3 to unsigned 8-bit
sqxtun v16.8b, v16.8h //saturating row 4 to unsigned 8-bit
st1 {v4.8b}, [x2], x5 //store row 1 in destination
st1 {v8.8b}, [x2], x5 //store row 2 in destination
subs w11, w11, #4 //decrement ht by 4
st1 {v12.8b}, [x2], x5 //store row 3 in destination
st1 {v16.8b}, [x2], x5 //store row 4 in destination
bgt loop_8 //if greater than 0 repeat the loop again
b end_loops
loop_16: //each iteration processes two rows
ld1 {v4.8b, v5.8b}, [x0], x3 //load row 1 in source 1
ld1 {v6.8b, v7.8b}, [x1], x4 //load row 1 in source 2
ld1 {v8.8b, v9.8b}, [x0], x3 //load row 2 in source 1
ld1 {v10.8b, v11.8b}, [x1], x4 //load row 2 in source 2
uxtl v20.8h, v4.8b //converting row 1L in source 1 to 16-bit
ld1 {v12.8b, v13.8b}, [x0], x3 //load row 3 in source 1
ld1 {v14.8b, v15.8b}, [x1], x4 //load row 3 in source 2
uxtl v22.8h, v6.8b //converting row 1L in source 2 to 16-bit
ld1 {v16.8b, v17.8b}, [x0], x3 //load row 4 in source 1
ld1 {v18.8b, v19.8b}, [x1], x4 //load row 4 in source 2
uxtl v4.8h, v5.8b //converting row 1H in source 1 to 16-bit
uxtl v6.8h, v7.8b //converting row 1H in source 2 to 16-bit
mul v20.8h, v20.8h , v2.h[0] //weight 1 mult. for row 1L
mla v20.8h, v22.8h , v2.h[2] //weight 2 mult. for row 1L
uxtl v24.8h, v8.8b //converting row 2L in source 1 to 16-bit
uxtl v26.8h, v10.8b //converting row 2L in source 2 to 16-bit
mul v4.8h, v4.8h , v2.h[0] //weight 1 mult. for row 1H
mla v4.8h, v6.8h , v2.h[2] //weight 2 mult. for row 1H
uxtl v8.8h, v9.8b //converting row 2H in source 1 to 16-bit
uxtl v10.8h, v11.8b //converting row 2H in source 2 to 16-bit
mul v24.8h, v24.8h , v2.h[0] //weight 1 mult. for row 2L
mla v24.8h, v26.8h , v2.h[2] //weight 2 mult. for row 2L
uxtl v28.8h, v12.8b //converting row 3L in source 1 to 16-bit
uxtl v30.8h, v14.8b //converting row 3L in source 2 to 16-bit
mul v8.8h, v8.8h , v2.h[0] //weight 1 mult. for row 2H
mla v8.8h, v10.8h , v2.h[2] //weight 2 mult. for row 2H
uxtl v12.8h, v13.8b //converting row 3H in source 1 to 16-bit
uxtl v14.8h, v15.8b //converting row 3H in source 2 to 16-bit
mul v28.8h, v28.8h , v2.h[0] //weight 1 mult. for row 3L
mla v28.8h, v30.8h , v2.h[2] //weight 2 mult. for row 3L
uxtl v22.8h, v16.8b //converting row 4L in source 1 to 16-bit
uxtl v6.8h, v18.8b //converting row 4L in source 2 to 16-bit
mul v12.8h, v12.8h , v2.h[0] //weight 1 mult. for row 3H
mla v12.8h, v14.8h , v2.h[2] //weight 2 mult. for row 3H
uxtl v16.8h, v17.8b //converting row 4H in source 1 to 16-bit
uxtl v18.8h, v19.8b //converting row 4H in source 2 to 16-bit
mul v22.8h, v22.8h , v2.h[0] //weight 1 mult. for row 4L
mla v22.8h, v6.8h , v2.h[2] //weight 2 mult. for row 4L
srshl v20.8h, v20.8h , v0.8h //rounds off the weighted samples from row 1L
mul v16.8h, v16.8h , v2.h[0] //weight 1 mult. for row 4H
mla v16.8h, v18.8h , v2.h[2] //weight 2 mult. for row 4H
srshl v4.8h, v4.8h , v0.8h //rounds off the weighted samples from row 1H
srshl v24.8h, v24.8h , v0.8h //rounds off the weighted samples from row 2L
saddw v20.8h, v20.8h , v3.8b //adding offset for row 1L
srshl v8.8h, v8.8h , v0.8h //rounds off the weighted samples from row 2H
saddw v4.8h, v4.8h , v3.8b //adding offset for row 1H
srshl v28.8h, v28.8h , v0.8h //rounds off the weighted samples from row 3L
saddw v24.8h, v24.8h , v3.8b //adding offset for row 2L
srshl v12.8h, v12.8h , v0.8h //rounds off the weighted samples from row 3H
saddw v8.8h, v8.8h , v3.8b //adding offset for row 2H
srshl v22.8h, v22.8h , v0.8h //rounds off the weighted samples from row 4L
saddw v28.8h, v28.8h , v3.8b //adding offset for row 3L
srshl v16.8h, v16.8h , v0.8h //rounds off the weighted samples from row 4H
saddw v12.8h, v12.8h , v3.8b //adding offset for row 3H
sqxtun v26.8b, v20.8h //saturating row 1L to unsigned 8-bit
saddw v22.8h, v22.8h , v3.8b //adding offset for row 4L
sqxtun v27.8b, v4.8h //saturating row 1H to unsigned 8-bit
saddw v16.8h, v16.8h , v3.8b //adding offset for row 4H
sqxtun v10.8b, v24.8h //saturating row 2L to unsigned 8-bit
sqxtun v11.8b, v8.8h //saturating row 2H to unsigned 8-bit
sqxtun v30.8b, v28.8h //saturating row 3L to unsigned 8-bit
sqxtun v31.8b, v12.8h //saturating row 3H to unsigned 8-bit
st1 {v26.8b, v27.8b}, [x2], x5 //store row 1 in destination
sqxtun v14.8b, v22.8h //saturating row 4L to unsigned 8-bit
sqxtun v15.8b, v16.8h //saturating row 4H to unsigned 8-bit
st1 {v10.8b, v11.8b}, [x2], x5 //store row 2 in destination
subs w11, w11, #4 //decrement ht by 4
st1 {v30.8b, v31.8b}, [x2], x5 //store row 3 in destination
st1 {v14.8b, v15.8b}, [x2], x5 //store row 4 in destination
bgt loop_16 //if greater than 0 repeat the loop again
end_loops:
// LDMFD sp!,{x4-x12,x15} //Reload the registers from sp
ldp x19, x20, [sp], #16
pop_v_regs
ret
//*******************************************************************************
//* @function
//* ih264_weighted_bi_pred_chroma_av8()
//*
//* @brief
//* This routine performs the default weighted prediction as described in sec
//* 8.4.2.3.2 titled "Weighted sample prediction process" for chroma.
//*
//* @par Description:
//* This function gets two ht x wd blocks, calculates the weighted samples,
//* rounds off, adds offset and stores it in the destination block for U and V.
//*
//* @param[in] puc_src1
//* UWORD8 Pointer to the buffer containing the input block 1.
//*
//* @param[in] puc_src2
//* UWORD8 Pointer to the buffer containing the input block 2.
//*
//* @param[out] puc_dst
//* UWORD8 pointer to the destination where the output block is stored.
//*
//* @param[in] src_strd1
//* Stride of the input buffer 1
//*
//* @param[in] src_strd2
//* Stride of the input buffer 2
//*
//* @param[in] dst_strd
//* Stride of the destination buffer
//*
//* @param[in] log_WD
//* number of bits to be rounded off
//*
//* @param[in] wt1
//* weights for the weighted prediction in U and V
//*
//* @param[in] wt2
//* weights for the weighted prediction in U and V
//*
//* @param[in] ofst1
//* offset 1 used after rounding off for U an dV
//*
//* @param[in] ofst2
//* offset 2 used after rounding off for U and V
//*
//* @param[in] ht
//* integer height of the array
//*
//* @param[in] wd
//* integer width of the array
//*
//* @returns
//* None
//*
//* @remarks
//* (ht,wd) can be (2,2), (2,4), (4,2), (4,4), (4,8), (8,4) or (8,8).
//*
//*******************************************************************************
//*/
//void ih264_weighted_bi_pred_chroma_av8(UWORD8 *puc_src1,
// UWORD8 *puc_src2,
// UWORD8 *puc_dst,
// WORD32 src_strd1,
// WORD32 src_strd2,
// WORD32 dst_strd,
// WORD32 log_WD,
// WORD32 wt1,
// WORD32 wt2,
// WORD32 ofst1,
// WORD32 ofst2,
// WORD32 ht,
// WORD32 wd)
//
//**************Variables Vs Registers*****************************************
// x0 => puc_src1
// x1 => puc_src2
// x2 => puc_dst
// w3 => src_strd1
// w4 => src_strd2
// w5 => dst_strd
// w6 => log_WD
// w7 => wt1
// [sp] => wt2 (w8)
// [sp+8] => ofst1 (w9)
// [sp+16] => ofst2 (w10)
// [sp+24] => ht (w11)
// [sp+32] => wd (w12)
//
.global ih264_weighted_bi_pred_chroma_av8
ih264_weighted_bi_pred_chroma_av8:
// STMFD sp!, {x4-x12,x14} //stack stores the values of the arguments
push_v_regs
sxtw x3, w3
sxtw x4, w4
sxtw x5, w5
stp x19, x20, [sp, #-16]!
ldr w8, [sp, #80] //Load wt2 in w8
dup v4.4s, w8 //Q2 = (wt2_u, wt2_v) (32-bit)
dup v2.4s, w7 //Q1 = (wt1_u, wt1_v) (32-bit)
add w6, w6, #1 //w6 = log_WD + 1
ldr w9, [sp, #88] //Load ofst1 in w9
ldr w10, [sp, #96] //Load ofst2 in w10
neg w20, w6 //w20 = -(log_WD + 1)
dup v0.8h, w20 //Q0 = -(log_WD + 1) (16-bit)
ldr w11, [sp, #104] //Load ht in x11
ldr w12, [sp, #112] //Load wd in x12
dup v20.8h, w9 //0ffset1
dup v21.8h, w10 //0ffset2
srhadd v6.8b, v20.8b, v21.8b
sxtl v6.8h, v6.8b
cmp w12, #8 //check if wd is 8
beq loop_8_uv //branch if wd is 8
cmp w12, #4 //check if wd is 4
beq loop_4_uv //branch if wd is 4
loop_2_uv: //each iteration processes two rows
ld1 {v8.s}[0], [x0], x3 //load row 1 in source 1
ld1 {v8.s}[1], [x0], x3 //load row 2 in source 1
ld1 {v10.s}[0], [x1], x4 //load row 1 in source 2
ld1 {v10.s}[1], [x1], x4 //load row 2 in source 2
uxtl v8.8h, v8.8b //converting rows 1,2 in source 1 to 16-bit
uxtl v10.8h, v10.8b //converting rows 1,2 in source 2 to 16-bit
mul v8.8h, v8.8h , v2.8h //weight 1 mult. for rows 1,2
mla v8.8h, v10.8h , v4.8h //weight 2 mult. for rows 1,2
srshl v8.8h, v8.8h , v0.8h //rounds off the weighted samples from rows 1,2
add v8.8h, v8.8h , v6.8h //adding offset for rows 1,2
sqxtun v8.8b, v8.8h //saturating rows 1,2 to unsigned 8-bit/
st1 {v8.s}[0], [x2], x5 //store row 1 in destination
st1 {v8.s}[1], [x2], x5 //store row 2 in destination
subs w11, w11, #2 //decrement ht by 2
bgt loop_2_uv //if greater than 0 repeat the loop again
b end_loops_uv
loop_4_uv: //each iteration processes two rows
ld1 {v8.8b}, [x0], x3 //load row 1 in source 1
ld1 {v10.8b}, [x1], x4 //load row 1 in source 2
uxtl v8.8h, v8.8b //converting row 1 in source 1 to 16-bit
ld1 {v12.8b}, [x0], x3 //load row 2 in source 1
uxtl v10.8h, v10.8b //converting row 1 in source 2 to 16-bit
ld1 {v14.8b}, [x1], x4 //load row 2 in source 2
uxtl v12.8h, v12.8b //converting row 2 in source 1 to 16-bit
mul v8.8h, v8.8h , v2.8h //weight 1 mult. for row 1
mla v8.8h, v10.8h , v4.8h //weight 2 mult. for row 1
uxtl v14.8h, v14.8b //converting row 2 in source 2 to 16-bit
mul v12.8h, v12.8h , v2.8h //weight 1 mult. for row 2
mla v12.8h, v14.8h , v4.8h //weight 2 mult. for row 2
subs w11, w11, #2 //decrement ht by 2
srshl v8.8h, v8.8h , v0.8h //rounds off the weighted samples from row 1
srshl v12.8h, v12.8h , v0.8h //rounds off the weighted samples from row 2
add v8.8h, v8.8h , v6.8h //adding offset for row 1
add v12.8h, v12.8h , v6.8h //adding offset for row 2
sqxtun v8.8b, v8.8h //saturating row 1 to unsigned 8-bit
sqxtun v12.8b, v12.8h //saturating row 2 to unsigned 8-bit
st1 {v8.8b}, [x2], x5 //store row 1 in destination
st1 {v12.8b}, [x2], x5 //store row 2 in destination
bgt loop_4_uv //if greater than 0 repeat the loop again
b end_loops_uv
loop_8_uv: //each iteration processes two rows
ld1 {v8.8b, v9.8b}, [x0], x3 //load row 1 in source 1
ld1 {v10.8b, v11.8b}, [x1], x4 //load row 1 in source 2
ld1 {v12.8b, v13.8b}, [x0], x3 //load row 2 in source 1
ld1 {v14.8b, v15.8b}, [x1], x4 //load row 2 in source 2
uxtl v24.8h, v8.8b //converting row 1L in source 1 to 16-bit
ld1 {v16.8b, v17.8b}, [x0], x3 //load row 3 in source 1
ld1 {v18.8b, v19.8b}, [x1], x4 //load row 3 in source 2
uxtl v26.8h, v10.8b //converting row 1L in source 2 to 16-bit
ld1 {v20.8b, v21.8b}, [x0], x3 //load row 4 in source 1
ld1 {v22.8b, v23.8b}, [x1], x4 //load row 4 in source 2
uxtl v8.8h, v9.8b //converting row 1H in source 1 to 16-bit
uxtl v10.8h, v11.8b //converting row 1H in source 2 to 16-bit
mul v24.8h, v24.8h , v2.8h //weight 1 mult. for row 1L
mla v24.8h, v26.8h , v4.8h //weight 2 mult. for row 1L
uxtl v28.8h, v12.8b //converting row 2L in source 1 to 16-bit
uxtl v30.8h, v14.8b //converting row 2L in source 2 to 16-bit
mul v8.8h, v8.8h , v2.8h //weight 1 mult. for row 1H
mla v8.8h, v10.8h , v4.8h //weight 2 mult. for row 1H
uxtl v12.8h, v13.8b //converting row 2H in source 1 to 16-bit
uxtl v14.8h, v15.8b //converting row 2H in source 2 to 16-bit
mul v28.8h, v28.8h , v2.8h //weight 1 mult. for row 2L
mla v28.8h, v30.8h , v4.8h //weight 2 mult. for row 2L
uxtl v26.8h, v16.8b //converting row 3L in source 1 to 16-bit
uxtl v10.8h, v18.8b //converting row 3L in source 2 to 16-bit
mul v12.8h, v12.8h , v2.8h //weight 1 mult. for row 2H
mla v12.8h, v14.8h , v4.8h //weight 2 mult. for row 2H
uxtl v16.8h, v17.8b //converting row 3H in source 1 to 16-bit
uxtl v18.8h, v19.8b //converting row 3H in source 2 to 16-bit
mul v26.8h, v26.8h , v2.8h //weight 1 mult. for row 3L
mla v26.8h, v10.8h , v4.8h //weight 2 mult. for row 3L
uxtl v30.8h, v20.8b //converting row 4L in source 1 to 16-bit
uxtl v14.8h, v22.8b //converting row 4L in source 2 to 16-bit
mul v16.8h, v16.8h , v2.8h //weight 1 mult. for row 3H
mla v16.8h, v18.8h , v4.8h //weight 2 mult. for row 3H
uxtl v20.8h, v21.8b //converting row 4H in source 1 to 16-bit
uxtl v22.8h, v23.8b //converting row 4H in source 2 to 16-bit
mul v30.8h, v30.8h , v2.8h //weight 1 mult. for row 4L
mla v30.8h, v14.8h , v4.8h //weight 2 mult. for row 4L
srshl v24.8h, v24.8h , v0.8h //rounds off the weighted samples from row 1L
mul v20.8h, v20.8h , v2.8h //weight 1 mult. for row 4H
mla v20.8h, v22.8h , v4.8h //weight 2 mult. for row 4H
srshl v8.8h, v8.8h , v0.8h //rounds off the weighted samples from row 1H
srshl v28.8h, v28.8h , v0.8h //rounds off the weighted samples from row 2L
add v24.8h, v24.8h , v6.8h //adding offset for row 1L
srshl v12.8h, v12.8h , v0.8h //rounds off the weighted samples from row 2H
add v8.8h, v8.8h , v6.8h //adding offset for row 1H
srshl v26.8h, v26.8h , v0.8h //rounds off the weighted samples from row 3L
add v28.8h, v28.8h , v6.8h //adding offset for row 2L
srshl v16.8h, v16.8h , v0.8h //rounds off the weighted samples from row 3H
add v12.8h, v12.8h , v6.8h //adding offset for row 2H
srshl v30.8h, v30.8h , v0.8h //rounds off the weighted samples from row 4L
add v26.8h, v26.8h , v6.8h //adding offset for row 3L
srshl v20.8h, v20.8h , v0.8h //rounds off the weighted samples from row 4H
add v16.8h, v16.8h , v6.8h //adding offset for row 3H
sqxtun v10.8b, v24.8h //saturating row 1L to unsigned 8-bit
add v30.8h, v30.8h , v6.8h //adding offset for row 4L
sqxtun v11.8b, v8.8h //saturating row 1H to unsigned 8-bit
add v20.8h, v20.8h , v6.8h //adding offset for row 4H
sqxtun v18.8b, v28.8h //saturating row 2L to unsigned 8-bit
sqxtun v19.8b, v12.8h //saturating row 2H to unsigned 8-bit
sqxtun v14.8b, v26.8h //saturating row 3L to unsigned 8-bit
sqxtun v15.8b, v16.8h //saturating row 3H to unsigned 8-bit
st1 {v10.8b, v11.8b}, [x2], x5 //store row 1 in destination
sqxtun v22.8b, v30.8h //saturating row 4L to unsigned 8-bit
sqxtun v23.8b, v20.8h //saturating row 4H to unsigned 8-bit
st1 {v18.8b, v19.8b}, [x2], x5 //store row 2 in destination
subs w11, w11, #4 //decrement ht by 4
st1 {v14.8b, v15.8b}, [x2], x5 //store row 3 in destination
st1 {v22.8b, v23.8b}, [x2], x5 //store row 4 in destination
bgt loop_8_uv //if greater than 0 repeat the loop again
end_loops_uv:
// LDMFD sp!,{x4-x12,x15} //Reload the registers from sp
ldp x19, x20, [sp], #16
pop_v_regs
ret