@/****************************************************************************** @ * @ * 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_inter_pred_luma_vert_qpel_a9q.s @* @* @brief @* Contains function definitions for inter prediction vertical quarter pel interpolation. @* @* @author @* Mohit @* @* @par List of Functions: @* @* - ih264_inter_pred_luma_vert_qpel_a9q() @* @* @remarks @* None @* @******************************************************************************* @* @* All the functions here are replicated from ih264_inter_pred_filters.c @ @******************************************************************************* @* @* @brief @* Quarter pel interprediction luma filter for vertical input @* @* @par Description: @* Applies a 6 tap horizontal filter .The output is clipped to 8 bits @* sec 8.4.2.2.1 titled "Luma sample interpolation process" @* @* @param[in] pu1_src @* UWORD8 pointer to the source @* @* @param[out] pu1_dst @* UWORD8 pointer to the destination @* @* @param[in] src_strd @* integer source stride @* @* @param[in] dst_strd @* integer destination stride @* @* @param[in] ht @* integer height of the array @* @* @param[in] wd @* integer width of the array @* @* @param[in] pu1_tmp: temporary buffer: UNUSED in this function @* @* @param[in] dydx: x and y reference offset for qpel calculations. @* @returns @* @ @remarks @* None @* @******************************************************************************* @* @void ih264_inter_pred_luma_vert ( @ UWORD8 *pu1_src, @ UWORD8 *pu1_dst, @ WORD32 src_strd, @ WORD32 dst_strd, @ WORD32 ht, @ WORD32 wd, @ UWORD8* pu1_tmp, @ UWORD32 dydx) @**************Variables Vs Registers***************************************** @ r0 => *pu1_src @ r1 => *pu1_dst @ r2 => src_strd @ r3 => dst_strd @ r5 => ht @ r6 => wd @ r7 => dydx .text .p2align 2 .global ih264_inter_pred_luma_vert_qpel_a9q ih264_inter_pred_luma_vert_qpel_a9q: stmfd sp!, {r4-r12, r14} @store register values to stack vstmdb sp!, {d8-d15} @push neon registers to stack ldr r5, [sp, #104] @Loads ht ldr r6, [sp, #108] @Loads wd ldr r7, [sp, #116] @Loads dydx and r7, r7, #12 @Finds y-offset lsr r7, r7, #3 @dydx>>3 mul r7, r2, r7 add r7, r0, r7 @pu1_src + (y_offset>>1)*src_strd vmov.u16 q11, #20 @ Filter coeff 0x14 into Q11 sub r0, r0, r2, lsl #1 @pu1_src-2*src_strd subs r12, r6, #8 @if wd=8 branch to loop_8 vmov.u16 q12, #5 @ Filter coeff 0x5 into Q12 beq loop_8 subs r12, r6, #4 @if wd=4 branch to loop_4 beq loop_4 loop_16: @when wd=16 vld1.u32 {q0}, [r0], r2 @ Vector load from src[0_0] vld1.u32 {q1}, [r0], r2 @ Vector load from src[1_0] vld1.u32 {q2}, [r0], r2 @ Vector load from src[2_0] vld1.u32 {q3}, [r0], r2 @ Vector load from src[3_0] vld1.u32 {q4}, [r0], r2 @ Vector load from src[4_0] vaddl.u8 q6, d4, d6 @ temp1 = src[2_0] + src[3_0] vld1.u32 {q5}, [r0], r2 @ Vector load from src[5_0] vaddl.u8 q7, d0, d10 @ temp = src[0_0] + src[5_0] vaddl.u8 q8, d2, d8 @ temp2 = src[1_0] + src[4_0] vmla.u16 q7, q6, q11 @ temp += temp1 * 20 vaddl.u8 q10, d1, d11 @ temp4 = src[0_8] + src[5_8] vaddl.u8 q9, d5, d7 @ temp3 = src[2_8] + src[3_8] vmla.u16 q10, q9, q11 @ temp4 += temp3 * 20 vld1.u32 {q0}, [r0], r2 vaddl.u8 q13, d3, d9 @ temp5 = src[1_8] + src[4_8] vaddl.u8 q6, d6, d8 vmls.u16 q7, q8, q12 @ temp -= temp2 * 5 vaddl.u8 q8, d2, d0 vaddl.u8 q9, d4, d10 vmla.u16 q8, q6, q11 vmls.u16 q10, q13, q12 @ temp4 -= temp5 * 5 vaddl.u8 q13, d5, d11 vaddl.u8 q6, d7, d9 vqrshrun.s16 d30, q7, #5 @ dst[0_0] = CLIP_U8((temp +16) >> 5) vaddl.u8 q7, d3, d1 vld1.u32 {q1}, [r0], r2 vmla.u16 q7, q6, q11 vmls.u16 q8, q9, q12 vqrshrun.s16 d31, q10, #5 @ dst[0_8] = CLIP_U8((temp4 +16) >> 5) vld1.u32 {q10}, [r7], r2 @ Load for interpolation row 0 vrhadd.u8 q15, q10, q15 @ Interpolation to obtain qpel value vaddl.u8 q9, d4, d2 vaddl.u8 q6, d8, d10 vst1.u32 {q15}, [r1], r3 @ Vector store to dst[0_0] vmla.u16 q9, q6, q11 vaddl.u8 q10, d6, d0 vmls.u16 q7, q13, q12 vqrshrun.s16 d30, q8, #5 vaddl.u8 q6, d9, d11 vaddl.u8 q8, d5, d3 vaddl.u8 q13, d7, d1 vmla.u16 q8, q6, q11 vmls.u16 q9, q10, q12 vld1.u32 {q2}, [r0], r2 vqrshrun.s16 d31, q7, #5 vld1.u32 {q7}, [r7], r2 @ Load for interpolation row 1 vaddl.u8 q6, d10, d0 vrhadd.u8 q15, q7, q15 @ Interpolation to obtain qpel value vaddl.u8 q7, d6, d4 vaddl.u8 q10, d8, d2 vmla.u16 q7, q6, q11 vmls.u16 q8, q13, q12 vst1.u32 {q15}, [r1], r3 @store row 1 vqrshrun.s16 d30, q9, #5 vaddl.u8 q9, d7, d5 vaddl.u8 q6, d11, d1 vmla.u16 q9, q6, q11 vaddl.u8 q13, d9, d3 vmls.u16 q7, q10, q12 vqrshrun.s16 d31, q8, #5 vld1.u32 {q8}, [r7], r2 @ Load for interpolation row 2 vmls.u16 q9, q13, q12 vrhadd.u8 q15, q8, q15 @ Interpolation to obtain qpel value vaddl.u8 q6, d0, d2 @ temp1 = src[2_0] + src[3_0] vst1.u32 {q15}, [r1], r3 @store row 2 vaddl.u8 q8, d10, d4 @ temp2 = src[1_0] + src[4_0] vaddl.u8 q10, d9, d7 @ temp4 = src[0_8] + src[5_8] vqrshrun.s16 d30, q7, #5 vaddl.u8 q13, d5, d11 @ temp5 = src[1_8] + src[4_8] vaddl.u8 q7, d8, d6 @ temp = src[0_0] + src[5_0] vqrshrun.s16 d31, q9, #5 vld1.u32 {q9}, [r7], r2 @ Load for interpolation row 3 vmla.u16 q7, q6, q11 @ temp += temp1 * 20 vrhadd.u8 q15, q9, q15 @ Interpolation to obtain qpel value vaddl.u8 q9, d1, d3 @ temp3 = src[2_8] + src[3_8] vst1.u32 {q15}, [r1], r3 @store row 3 subs r5, r5, #4 @ 4 rows processed, decrement by 4 subne r0, r0 , r2, lsl #2 subne r0, r0, r2 beq end_func @ Branch if height==4 b loop_16 @ looping if height = 8 or 16 loop_8: @ Processing row0 and row1 vld1.u32 d0, [r0], r2 @ Vector load from src[0_0] vld1.u32 d1, [r0], r2 @ Vector load from src[1_0] vld1.u32 d2, [r0], r2 @ Vector load from src[2_0] vld1.u32 d3, [r0], r2 @ Vector load from src[3_0] vld1.u32 d4, [r0], r2 @ Vector load from src[4_0] vld1.u32 d5, [r0], r2 @ Vector load from src[5_0] vaddl.u8 q3, d2, d3 @ temp1 = src[2_0] + src[3_0] vaddl.u8 q4, d0, d5 @ temp = src[0_0] + src[5_0] vaddl.u8 q5, d1, d4 @ temp2 = src[1_0] + src[4_0] vmla.u16 q4, q3, q11 @ temp += temp1 * 20 vld1.u32 d6, [r0], r2 vaddl.u8 q7, d3, d4 vaddl.u8 q8, d1, d6 vaddl.u8 q9, d2, d5 vmls.u16 q4, q5, q12 @ temp -= temp2 * 5 vmla.u16 q8, q7, q11 vld1.u32 d7, [r0], r2 vaddl.u8 q10, d4, d5 vaddl.u8 q6, d2, d7 vaddl.u8 q5, d3, d6 vmls.u16 q8, q9, q12 vqrshrun.s16 d26, q4, #5 @ dst[0_0] = CLIP_U8( (temp + 16) >> 5) vmla.u16 q6, q10, q11 vld1.32 d8, [r7], r2 @Load value for interpolation (row0) vld1.32 d9, [r7], r2 @Load value for interpolation (row1) vld1.u32 d0, [r0], r2 vaddl.u8 q7, d5, d6 vqrshrun.s16 d27, q8, #5 vrhadd.u8 q13, q4, q13 @ Interpolation step for qpel calculation vaddl.u8 q10, d3, d0 vmls.u16 q6, q5, q12 vst1.u32 d26, [r1], r3 @ Vector store to dst[0_0] vaddl.u8 q9, d4, d7 vmla.u16 q10, q7, q11 vst1.u32 d27, [r1], r3 @ Vector store to dst[1_0] vqrshrun.s16 d28, q6, #5 vmls.u16 q10, q9, q12 vld1.32 d12, [r7], r2 @Load value for interpolation (row2) vld1.32 d13, [r7], r2 @Load value for interpolation (row3) vqrshrun.s16 d29, q10, #5 subs r9, r5, #4 vrhadd.u8 q14, q6, q14 vst1.u32 d28, [r1], r3 @store row 2 vst1.u32 d29, [r1], r3 @store row 3 subs r5, r5, #4 @ 4 rows processed, decrement by 4 subne r0, r0 , r2, lsl #2 subne r0, r0, r2 beq end_func @ Branch if height==4 b loop_8 @looping if height == 8 or 16 loop_4: @ Processing row0 and row1 vld1.u32 d0[0], [r0], r2 @ Vector load from src[0_0] vld1.u32 d1[0], [r0], r2 @ Vector load from src[1_0] vld1.u32 d2[0], [r0], r2 @ Vector load from src[2_0] vld1.u32 d3[0], [r0], r2 @ Vector load from src[3_0] vld1.u32 d4[0], [r0], r2 @ Vector load from src[4_0] vld1.u32 d5[0], [r0], r2 @ Vector load from src[5_0] vaddl.u8 q3, d2, d3 @ temp1 = src[2_0] + src[3_0] vaddl.u8 q4, d0, d5 @ temp = src[0_0] + src[5_0] vaddl.u8 q5, d1, d4 @ temp2 = src[1_0] + src[4_0] vmla.u16 q4, q3, q11 @ temp += temp1 * 20 vld1.u32 d6, [r0], r2 vaddl.u8 q7, d3, d4 vaddl.u8 q8, d1, d6 vaddl.u8 q9, d2, d5 vmls.u16 q4, q5, q12 @ temp -= temp2 * 5 vld1.u32 d7[0], [r0], r2 vmla.u16 q8, q7, q11 vaddl.u8 q10, d4, d5 vaddl.u8 q6, d2, d7 vaddl.u8 q5, d3, d6 vmls.u16 q8, q9, q12 vqrshrun.s16 d26, q4, #5 @ dst[0_0] = CLIP_U8( (temp + 16) >> 5) vld1.u32 d8[0], [r7], r2 @Load value for interpolation - row 0 vld1.u32 d9[0], [r7], r2 @Load value for interpolation - row 1 vmla.u16 q6, q10, q11 vld1.u32 d0[0], [r0], r2 vaddl.u8 q7, d5, d6 vqrshrun.s16 d27, q8, #5 vaddl.u8 q10, d3, d0 vrhadd.u8 q13, q13, q4 @Interpolation step for qpel calculation vmls.u16 q6, q5, q12 vst1.u32 d26[0], [r1], r3 @ Vector store to dst[0_0] vaddl.u8 q9, d4, d7 vmla.u16 q10, q7, q11 vst1.u32 d27[0], [r1], r3 @ store row 1 vqrshrun.s16 d28, q6, #5 vld1.u32 d12[0], [r7], r2 @Load value for interpolation - row 2 vld1.u32 d13[0], [r7], r2 @Load value for interpolation - row 3 vmls.u16 q10, q9, q12 vqrshrun.s16 d29, q10, #5 vrhadd.u8 q14, q6, q14 @Interpolation step for qpel calculation vst1.u32 d28[0], [r1], r3 @store row 2 vst1.u32 d29[0], [r1], r3 @store row 3 subs r5, r5, #8 subeq r0, r0, r2, lsl #2 subeq r0, r0, r2 beq loop_4 @ Loop if height==8 end_func: vldmia sp!, {d8-d15} @ Restore neon registers that were saved ldmfd sp!, {r4-r12, pc} @Restoring registers from stack