Lines Matching refs:dst1
67 #define dst1 r25
99 mov dst1=dst // copy because of rotation
116 EX(.failure_out,(EPI) st1 [dst1]=val1[PIPE_DEPTH-1],1)
131 and dst2=0x7,dst1 // dst offset
134 // that we can reach 8-byte alignment for both src1 and dst1.
141 // to the dst1. Then copy zeros for the rest of dst1.
153 // Optimization. If dst1 is 8-byte aligned (quite common), we don't need
154 // to copy the head to dst1, to start 8-byte copy software pipeline.
198 EX(.failure_out,(EPI) st1 [dst1]=val1[PIPE_DEPTH-1],1)
215 // Now both src1 and dst1 point to an 8-byte aligned address. And
227 // 3 (EPI): Store tmp to dst1
241 EX(.failure_out,(EPI) st8 [dst1]=tmp,8); \
249 (EPI) st8 [dst1]=tmp,8; \
290 (p14) adds dst1=-8,dst1
291 (p15) sub dst1=dst1,t1
309 EX(.failure_out,(EPI) st1 [dst1]=val1[PIPE_DEPTH-1],1)
348 EX(.failure_out,(p6) st1 [dst1]=val1[0],1)
356 EX(.failure_out,(p7) st2 [dst1]=val1[1],2)
359 EX(.failure_out, (p8) st4 [dst1]=val2[0],4)
363 EX(.failure_out, (p9) st8 [dst1]=val2[1],8)
370 adds dst2=8,dst1
380 EX(.failure_out, (EPI) st8 [dst1]=val1[PIPE_DEPTH-1],16)
400 EX(.failure_out, (p6) st8 [dst1]=val1[0],8)
405 EX(.failure_out,(p7) st4 [dst1]=val1[1],4)
408 EX(.failure_out, (p8) st2 [dst1]=val2[0],2)
411 EX(.failure_out, (p9) st1 [dst1]=val2[1])
422 // In the same loop iteration, the dst1 pointer does not directly
443 (EPI) st1 [dst1]=val1[PIPE_DEPTH-1],1
460 (EPI) st1 [dst1]=val1[PIPE_DEPTH-1],1
463 sub len=enddst,dst1,1 // precompute len
554 st1 [dst1]=r0,1
569 // we MUST use src1/endsrc here and not dst1/enddst because
578 (EPI) st8 [dst1]=val1[PIPE_DEPTH-1],16
582 cmp.ne p6,p0=dst1,enddst // Do we need to finish the tail ?
583 sub len=enddst,dst1,1 // precompute len
593 cmp.ne p6,p0=dst1,enddst // Do we need to finish the tail ?
594 sub len=enddst,dst1,1 // precompute len
606 sub ret0=enddst,dst1