Lines Matching refs:Sad
54 @* @brief computes distortion (SAD) between 2 16x16 blocks (fast mode)
57 @* This functions computes SAD between 2 16x16 blocks. There is a provision
58 @* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
77 @* integer evaluated sad
140 @* @brief computes distortion (SAD) between 2 16x8 blocks
144 @* This functions computes SAD between 2 16x8 blocks. There is a provision
145 @* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
164 @* integer evaluated sad
222 @* @brief computes distortion (SAD) between 2 16x16 blocks with early exit
225 @* This functions computes SAD between 2 16x16 blocks. There is a provision
226 @* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
245 @* integer evaluated sad
350 @// Detail Description : This function find the sad values of 4 Progressive MBs
415 @ Compute SAD
437 @// Detail Description : This function find the sad values of 4 Progressive MBs
515 @ Compute SAD
542 @* @brief computes distortion (SAD) for sub-pel motion estimation
545 @* This functions computes SAD for all the 8 half pel points
548 @* integer evaluated sad
682 @* @brief computes distortion (SAD) between 2 16x16 blocks
685 @* This functions computes SAD between 2 16x16 blocks. There is a provision
686 @* for early exit if the up-to computed SAD exceeds maximum allowed SAD. To
705 @* integer evaluated sad
770 @// Detail Description : This function find the sad values of 4 Progressive MBs
951 @* Description : This fucntion computes SAD for a 16x16 block.
982 @R5 :Distortion,ie SAD
990 mov r10, #0 @Sad
1085 vpadd.s16 d10, d16, d17 @I Get sad by adding s1 s2 s3 s4
1098 vdup.s16 q6, d10[0] @I Get the sad blk 1
1102 vmov.s16 r9, d10[0] @I Get the sad for block 1
1105 vmov.s16 r5, d10[1] @I Get the sad for block 2
1110 vdup.s16 q15, d10[1] @I Get the sad blk 1
1118 ADD R10, R10, R9 @I Add to the global sad blk 1
1122 ADD R10, R10, R5 @I Add to the global sad blk 2
1142 bne compute_sad_only @I if a non zero block is der,From now on compute sad only
1175 vmov.s16 r9, d10[0] @Get the sad for block 1
1178 vmov.s16 r5, d10[1] @Get the sad for block 2
1198 vpadd.s16 d10, d16, d17 @I Get sad by adding s1 s2 s3 s4
1200 vmov.u16 r9, d10[0] @Get the sad for block 1
1201 vmov.u16 r5, d10[1] @Get the sad for block 2
1207 ADD R10, R10, R9 @Add to the global sad blk 1
1213 ADD R10, R10, R5 @Add to the global sad blk 2
1215 vdup.u16 q6, d10[0] @Get the sad blk 1
1226 vdup.u16 q14, d10[1] @Get the sad blk 1
1242 compute_sad_only: @This block computes SAD only, so will be lighter
1244 @It will compute sad for odd blok,
1259 beq skip_odd_blk @If the blk is odd we have to compute sad
1262 vadd.u16 q4, q0, q1 @Add SAD of row1 and row2
1263 vadd.u16 q5, q2, q3 @Add SAD of row3 and row4
1264 vadd.u16 q6, q4, q5 @Add SAD of row 1-4
1269 vmov.u16 r9, d18[0] @Move sad to arm
1270 ADD R10, R10, R9 @Add to the global sad
1328 bne sad_loop @Go back to SAD computation
1344 vadd.u16 d6, d4, d5 @Add two blk sad
1345 vpadd.u16 d8, d6, d7 @Add col 1-2 and 3-4 sad
1346 vpadd.u16 d10, d8, d9 @Add col 12-34 sad
1348 vmov.u16 r9, d10[0] @move SAD to ARM
1349 ADD R10, R10, R9 @Add to the global sad
1358 str r10, [r5] @Store sad
