Cemu/dependencies/ih264d/common/arm/ih264_weighted_pred_a9q.s
2022-08-22 22:21:23 +02:00

480 lines
20 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_pred_a9q.s
@*
@* @brief
@* Contains function definitions for weighted prediction.
@*
@* @author
@* Kaushik Senthoor R
@*
@* @par List of Functions:
@*
@* - ih264_weighted_pred_luma_a9q()
@* - ih264_weighted_pred_chroma_a9q()
@*
@* @remarks
@* None
@*
@*******************************************************************************
@*
@*******************************************************************************
@* @function
@* ih264_weighted_pred_luma_a9q()
@*
@* @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 a ht x wd block, calculates the weighted sample, rounds
@* off, adds offset and stores it in the destination block.
@*
@* @param[in] pu1_src:
@* UWORD8 Pointer to the buffer containing the input block.
@*
@* @param[out] pu1_dst
@* UWORD8 pointer to the destination where the output block is stored.
@*
@* @param[in] src_strd
@* Stride of the input buffer
@*
@* @param[in] dst_strd
@* Stride of the destination buffer
@*
@* @param[in] log_wd
@* number of bits to be rounded off
@*
@* @param[in] wt
@* weight for the weighted prediction
@*
@* @param[in] ofst
@* offset 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_pred_luma_a9q(UWORD8 *pu1_src,
@ UWORD8 *pu1_dst,
@ WORD32 src_strd,
@ WORD32 dst_strd,
@ WORD32 log_wd,
@ WORD32 wt,
@ WORD32 ofst,
@ WORD32 ht,
@ WORD32 wd)
@
@**************Variables Vs Registers*****************************************
@ r0 => pu1_src
@ r1 => pu1_dst
@ r2 => src_strd
@ r3 => dst_strd
@ [sp] => log_wd (r4)
@ [sp+4] => wt (r5)
@ [sp+8] => ofst (r6)
@ [sp+12] => ht (r7)
@ [sp+16] => wd (r8)
@
.text
.p2align 2
.global ih264_weighted_pred_luma_a9q
ih264_weighted_pred_luma_a9q:
stmfd sp!, {r4-r9, r14} @stack stores the values of the arguments
ldr r5, [sp, #32] @Load wt
ldr r4, [sp, #28] @Load log_wd in r4
ldr r6, [sp, #36] @Load ofst
ldr r7, [sp, #40] @Load ht
ldr r8, [sp, #44] @Load wd
vpush {d8-d15}
vdup.16 d2, r5 @D2 = wt (16-bit)
neg r9, r4 @r9 = -log_wd
vdup.8 d3, r6 @D3 = ofst (8-bit)
cmp r8, #16 @check if wd is 16
vdup.16 q0, r9 @Q0 = -log_wd (16-bit)
beq loop_16 @branch if wd is 16
cmp r8, #8 @check if wd is 8
beq loop_8 @branch if wd is 8
loop_4: @each iteration processes four rows
vld1.32 d4[0], [r0], r2 @load row 1 in source
vld1.32 d4[1], [r0], r2 @load row 2 in source
vld1.32 d6[0], [r0], r2 @load row 3 in source
vld1.32 d6[1], [r0], r2 @load row 4 in source
vmovl.u8 q2, d4 @converting rows 1,2 to 16-bit
vmovl.u8 q3, d6 @converting rows 3,4 to 16-bit
vmul.s16 q2, q2, d2[0] @weight mult. for rows 1,2
vmul.s16 q3, q3, d2[0] @weight mult. for rows 3,4
subs r7, r7, #4 @decrement ht by 4
vrshl.s16 q2, q2, q0 @rounds off the weighted samples from rows 1,2
vrshl.s16 q3, q3, q0 @rounds off the weighted samples from rows 3,4
vaddw.s8 q2, q2, d3 @adding offset for rows 1,2
vaddw.s8 q3, q3, d3 @adding offset for rows 3,4
vqmovun.s16 d4, q2 @saturating rows 1,2 to unsigned 8-bit
vqmovun.s16 d6, q3 @saturating rows 3,4 to unsigned 8-bit
vst1.32 d4[0], [r1], r3 @store row 1 in destination
vst1.32 d4[1], [r1], r3 @store row 2 in destination
vst1.32 d6[0], [r1], r3 @store row 3 in destination
vst1.32 d6[1], [r1], r3 @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
vld1.8 d4, [r0], r2 @load row 1 in source
vld1.8 d6, [r0], r2 @load row 2 in source
vld1.8 d8, [r0], r2 @load row 3 in source
vmovl.u8 q2, d4 @converting row 1 to 16-bit
vld1.8 d10, [r0], r2 @load row 4 in source
vmovl.u8 q3, d6 @converting row 2 to 16-bit
vmovl.u8 q4, d8 @converting row 3 to 16-bit
vmul.s16 q2, q2, d2[0] @weight mult. for row 1
vmovl.u8 q5, d10 @converting row 4 to 16-bit
vmul.s16 q3, q3, d2[0] @weight mult. for row 2
vmul.s16 q4, q4, d2[0] @weight mult. for row 3
vmul.s16 q5, q5, d2[0] @weight mult. for row 4
vrshl.s16 q2, q2, q0 @rounds off the weighted samples from row 1
vrshl.s16 q3, q3, q0 @rounds off the weighted samples from row 2
vrshl.s16 q4, q4, q0 @rounds off the weighted samples from row 3
vaddw.s8 q2, q2, d3 @adding offset for row 1
vrshl.s16 q5, q5, q0 @rounds off the weighted samples from row 4
vaddw.s8 q3, q3, d3 @adding offset for row 2
vaddw.s8 q4, q4, d3 @adding offset for row 3
vqmovun.s16 d4, q2 @saturating row 1 to unsigned 8-bit
vaddw.s8 q5, q5, d3 @adding offset for row 4
vqmovun.s16 d6, q3 @saturating row 2 to unsigned 8-bit
vqmovun.s16 d8, q4 @saturating row 3 to unsigned 8-bit
vqmovun.s16 d10, q5 @saturating row 4 to unsigned 8-bit
vst1.8 d4, [r1], r3 @store row 1 in destination
vst1.8 d6, [r1], r3 @store row 2 in destination
subs r7, r7, #4 @decrement ht by 4
vst1.8 d8, [r1], r3 @store row 3 in destination
vst1.8 d10, [r1], r3 @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
vld1.8 {q2}, [r0], r2 @load row 1 in source
vld1.8 {q3}, [r0], r2 @load row 2 in source
vmovl.u8 q6, d4 @converting row 1L to 16-bit
vld1.8 {q4}, [r0], r2 @load row 3 in source
vmovl.u8 q7, d5 @converting row 1H to 16-bit
vld1.8 {q5}, [r0], r2 @load row 4 in source
vmovl.u8 q8, d6 @converting row 2L to 16-bit
vmul.s16 q6, q6, d2[0] @weight mult. for row 1L
vmovl.u8 q9, d7 @converting row 2H to 16-bit
vmul.s16 q7, q7, d2[0] @weight mult. for row 1H
vmovl.u8 q10, d8 @converting row 3L to 16-bit
vmul.s16 q8, q8, d2[0] @weight mult. for row 2L
vmovl.u8 q11, d9 @converting row 3H to 16-bit
vmul.s16 q9, q9, d2[0] @weight mult. for row 2H
vmovl.u8 q12, d10 @converting row 4L to 16-bit
vmul.s16 q10, q10, d2[0] @weight mult. for row 3L
vmovl.u8 q13, d11 @converting row 4H to 16-bit
vmul.s16 q11, q11, d2[0] @weight mult. for row 3H
vmul.s16 q12, q12, d2[0] @weight mult. for row 4L
vrshl.s16 q6, q6, q0 @rounds off the weighted samples from row 1L
vmul.s16 q13, q13, d2[0] @weight mult. for row 4H
vrshl.s16 q7, q7, q0 @rounds off the weighted samples from row 1H
vrshl.s16 q8, q8, q0 @rounds off the weighted samples from row 2L
vaddw.s8 q6, q6, d3 @adding offset for row 1L
vrshl.s16 q9, q9, q0 @rounds off the weighted samples from row 2H
vaddw.s8 q7, q7, d3 @adding offset for row 1H
vqmovun.s16 d4, q6 @saturating row 1L to unsigned 8-bit
vrshl.s16 q10, q10, q0 @rounds off the weighted samples from row 3L
vaddw.s8 q8, q8, d3 @adding offset for row 2L
vqmovun.s16 d5, q7 @saturating row 1H to unsigned 8-bit
vrshl.s16 q11, q11, q0 @rounds off the weighted samples from row 3H
vaddw.s8 q9, q9, d3 @adding offset for row 2H
vqmovun.s16 d6, q8 @saturating row 2L to unsigned 8-bit
vrshl.s16 q12, q12, q0 @rounds off the weighted samples from row 4L
vaddw.s8 q10, q10, d3 @adding offset for row 3L
vqmovun.s16 d7, q9 @saturating row 2H to unsigned 8-bit
vrshl.s16 q13, q13, q0 @rounds off the weighted samples from row 4H
vaddw.s8 q11, q11, d3 @adding offset for row 3H
vqmovun.s16 d8, q10 @saturating row 3L to unsigned 8-bit
vaddw.s8 q12, q12, d3 @adding offset for row 4L
vqmovun.s16 d9, q11 @saturating row 3H to unsigned 8-bit
vaddw.s8 q13, q13, d3 @adding offset for row 4H
vqmovun.s16 d10, q12 @saturating row 4L to unsigned 8-bit
vst1.8 {q2}, [r1], r3 @store row 1 in destination
vqmovun.s16 d11, q13 @saturating row 4H to unsigned 8-bit
vst1.8 {q3}, [r1], r3 @store row 2 in destination
subs r7, r7, #4 @decrement ht by 4
vst1.8 {q4}, [r1], r3 @store row 3 in destination
vst1.8 {q5}, [r1], r3 @store row 4 in destination
bgt loop_16 @if greater than 0 repeat the loop again
end_loops:
vpop {d8-d15}
ldmfd sp!, {r4-r9, r15} @Reload the registers from sp
@*******************************************************************************
@* @function
@* ih264_weighted_pred_chroma_a9q()
@*
@* @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 a ht x wd block, calculates the weighted sample, rounds
@* off, adds offset and stores it in the destination block for U and V.
@*
@* @param[in] pu1_src:
@* UWORD8 Pointer to the buffer containing the input block.
@*
@* @param[out] pu1_dst
@* UWORD8 pointer to the destination where the output block is stored.
@*
@* @param[in] src_strd
@* Stride of the input buffer
@*
@* @param[in] dst_strd
@* Stride of the destination buffer
@*
@* @param[in] log_wd
@* number of bits to be rounded off
@*
@* @param[in] wt
@* weights for the weighted prediction for U and V
@*
@* @param[in] ofst
@* offsets 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_pred_chroma_a9q(UWORD8 *pu1_src,
@ UWORD8 *pu1_dst,
@ WORD32 src_strd,
@ WORD32 dst_strd,
@ WORD32 log_wd,
@ WORD32 wt,
@ WORD32 ofst,
@ WORD32 ht,
@ WORD32 wd)
@
@**************Variables Vs Registers*****************************************
@ r0 => pu1_src
@ r1 => pu1_dst
@ r2 => src_strd
@ r3 => dst_strd
@ [sp] => log_wd (r4)
@ [sp+4] => wt (r5)
@ [sp+8] => ofst (r6)
@ [sp+12] => ht (r7)
@ [sp+16] => wd (r8)
@
.global ih264_weighted_pred_chroma_a9q
ih264_weighted_pred_chroma_a9q:
stmfd sp!, {r4-r9, r14} @stack stores the values of the arguments
ldr r4, [sp, #28] @Load log_wd in r4
ldr r5, [sp, #32] @Load wt = {wt_u (16-bit), wt_v (16-bit)}
ldr r6, [sp, #36] @Load ofst = {ofst_u (8-bit), ofst_v (8-bit)}
ldr r8, [sp, #44] @Load wd
neg r9, r4 @r9 = -log_wd
vdup.32 q1, r5 @Q1 = {wt_u (16-bit), wt_v (16-bit)}
ldr r7, [sp, #40] @Load ht
vpush {d8-d15}
vdup.16 d4, r6 @D4 = {ofst_u (8-bit), ofst_v (8-bit)}
cmp r8, #8 @check if wd is 8
vdup.16 q0, r9 @Q0 = -log_wd (16-bit)
beq loop_8_uv @branch if wd is 8
cmp r8, #4 @check if ws is 4
beq loop_4_uv @branch if wd is 4
loop_2_uv: @each iteration processes two rows
vld1.32 d6[0], [r0], r2 @load row 1 in source
vld1.32 d6[1], [r0], r2 @load row 2 in source
vmovl.u8 q3, d6 @converting rows 1,2 to 16-bit
vmul.s16 q3, q3, q1 @weight mult. for rows 1,2
vrshl.s16 q3, q3, q0 @rounds off the weighted samples from rows 1,2
vaddw.s8 q3, q3, d4 @adding offset for rows 1,2
vqmovun.s16 d6, q3 @saturating rows 1,2 to unsigned 8-bit
subs r7, r7, #2 @decrement ht by 2
vst1.32 d6[0], [r1], r3 @store row 1 in destination
vst1.32 d6[1], [r1], r3 @store row 2 in destination
bgt loop_2_uv @if greater than 0 repeat the loop again
b end_loops_uv
loop_4_uv: @each iteration processes two rows
vld1.8 d6, [r0], r2 @load row 1 in source
vld1.8 d8, [r0], r2 @load row 2 in source
vmovl.u8 q3, d6 @converting row 1 to 16-bit
vmovl.u8 q4, d8 @converting row 2 to 16-bit
vmul.s16 q3, q3, q1 @weight mult. for row 1
vmul.s16 q4, q4, q1 @weight mult. for row 2
subs r7, r7, #2 @decrement ht by 2
vrshl.s16 q3, q3, q0 @rounds off the weighted samples from row 1
vrshl.s16 q4, q4, q0 @rounds off the weighted samples from row 2
vaddw.s8 q3, q3, d4 @adding offset for row 1
vaddw.s8 q4, q4, d4 @adding offset for row 2
vqmovun.s16 d6, q3 @saturating row 1 to unsigned 8-bit
vqmovun.s16 d8, q4 @saturating row 2 to unsigned 8-bit
vst1.8 d6, [r1], r3 @store row 1 in destination
vst1.8 d8, [r1], r3 @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
vld1.8 {q3}, [r0], r2 @load row 1 in source
vld1.8 {q4}, [r0], r2 @load row 2 in source
vmovl.u8 q7, d6 @converting row 1L to 16-bit
vld1.8 {q5}, [r0], r2 @load row 3 in source
vmovl.u8 q8, d7 @converting row 1H to 16-bit
vld1.8 {q6}, [r0], r2 @load row 4 in source
vmul.s16 q7, q7, q1 @weight mult. for row 1L
vmovl.u8 q9, d8 @converting row 2L to 16-bit
vmul.s16 q8, q8, q1 @weight mult. for row 1H
vmovl.u8 q10, d9 @converting row 2H to 16-bit
vmul.s16 q9, q9, q1 @weight mult. for row 2L
vmovl.u8 q11, d10 @converting row 3L to 16-bit
vmul.s16 q10, q10, q1 @weight mult. for row 2H
vmovl.u8 q12, d11 @converting row 3H to 16-bit
vmul.s16 q11, q11, q1 @weight mult. for row 3L
vmovl.u8 q13, d12 @converting row 4L to 16-bit
vmul.s16 q12, q12, q1 @weight mult. for row 3H
vmovl.u8 q14, d13 @converting row 4H to 16-bit
vmul.s16 q13, q13, q1 @weight mult. for row 4L
vrshl.s16 q7, q7, q0 @rounds off the weighted samples from row 1L
vmul.s16 q14, q14, q1 @weight mult. for row 4H
vrshl.s16 q8, q8, q0 @rounds off the weighted samples from row 1H
vrshl.s16 q9, q9, q0 @rounds off the weighted samples from row 2L
vaddw.s8 q7, q7, d4 @adding offset for row 1L
vrshl.s16 q10, q10, q0 @rounds off the weighted samples from row 2H
vaddw.s8 q8, q8, d4 @adding offset for row 1H
vqmovun.s16 d6, q7 @saturating row 1L to unsigned 8-bit
vrshl.s16 q11, q11, q0 @rounds off the weighted samples from row 3L
vaddw.s8 q9, q9, d4 @adding offset for row 2L
vqmovun.s16 d7, q8 @saturating row 1H to unsigned 8-bit
vrshl.s16 q12, q12, q0 @rounds off the weighted samples from row 3H
vaddw.s8 q10, q10, d4 @adding offset for row 2H
vqmovun.s16 d8, q9 @saturating row 2L to unsigned 8-bit
vrshl.s16 q13, q13, q0 @rounds off the weighted samples from row 4L
vaddw.s8 q11, q11, d4 @adding offset for row 3L
vqmovun.s16 d9, q10 @saturating row 2H to unsigned 8-bit
vrshl.s16 q14, q14, q0 @rounds off the weighted samples from row 4H
vaddw.s8 q12, q12, d4 @adding offset for row 3H
vqmovun.s16 d10, q11 @saturating row 3L to unsigned 8-bit
vaddw.s8 q13, q13, d4 @adding offset for row 4L
vqmovun.s16 d11, q12 @saturating row 3H to unsigned 8-bit
vaddw.s8 q14, q14, d4 @adding offset for row 4H
vqmovun.s16 d12, q13 @saturating row 4L to unsigned 8-bit
vst1.8 {q3}, [r1], r3 @store row 1 in destination
vqmovun.s16 d13, q14 @saturating row 4H to unsigned 8-bit
vst1.8 {q4}, [r1], r3 @store row 2 in destination
subs r7, r7, #4 @decrement ht by 4
vst1.8 {q5}, [r1], r3 @store row 3 in destination
vst1.8 {q6}, [r1], r3 @store row 4 in destination
bgt loop_8_uv @if greater than 0 repeat the loop again
end_loops_uv:
vpop {d8-d15}
ldmfd sp!, {r4-r9, r15} @Reload the registers from sp