quant_h263_ia64.s

wince下的xvidcore开发库,可用于MP4等视频播放开发

 //*******************************************************************************
 //*                                                                             *
 //*     functions quant_inter and dequant_inter have been softwarepipelined     *
 //*     use was made of the pmpyshr2 instruction                                *
 //*                                                                             *
 //*     by Christian Engel and Hans-Joachim Daniels                             *
 //*     [e-mail] [e-mail]              *
 //*                                                                             *
 //*     This was made for the ia64 DivX laboratory (yes, it was really called   *
 //*     this way, originally OpenDivX was intendet, but died shortly before our *
 //*     work started (you will probably already know ...))                      *
 //*     at the Universitat Karlsruhe (TH) held between April and July 2002      *
 //*     http://www.info.uni-karlsruhe.de/~rubino/ia64p/                         *
 //*                                                                             *
 //*******************************************************************************/
         .file   "quant_h263_ia64.s"
         .pred.safe_across_calls p1-p5,p16-p63
                 .section        .rodata
         .align 4
         .type    multipliers#,@object
         .size    multipliers#,128
 multipliers:
         data4   0
         data4   32769
         data4   16385
         data4   10923
         data4   8193
         data4   6554
         data4   5462
         data4   4682
         data4   4097
         data4   3641
         data4   3277
         data4   2979
         data4   2731
         data4   2521
         data4   2341
         data4   2185
         data4   2049
         data4   1928
         data4   1821
         data4   1725
         data4   1639
         data4   1561
         data4   1490
         data4   1425
         data4   1366
         data4   1311
         data4   1261
         data4   1214
         data4   1171
         data4   1130
         data4   1093
         data4   1058
         .global __divdi3#
 .text
         .align 16
         .global quant_h263_intra_ia64#
         .proc quant_h263_intra_ia64#
 quant_h263_intra_ia64:
         .prologue 
         .save ar.pfs, r38
         alloc r38 = ar.pfs, 4, 3, 2, 0
         adds r16 = -8, r12
         .fframe 32
         adds r12 = -32, r12
         mov r17 = ar.lc
         addl r14 = @ltoff(multipliers#), gp
         ld2 r15 = [r33]
         ;;
         .savesp ar.lc, 24
         st8 [r16] = r17, 8
         ld8 r14 = [r14]
         sxt2 r15 = r15
         ;;
         .save.f 0x1
         stf.spill [r16] = f2
         .save rp, r37
         mov r37 = b0
         .body
         dep.z r36 = r34, 1, 15
         dep.z r16 = r34, 2, 32
         cmp4.ge p6, p7 = 0, r15
         ;;
         add r16 = r16, r14
         ;;
         ld4 r16 = [r16]
         ;;
         setf.sig f2 = r16
         (p6) br.cond.dptk .L8
         extr r39 = r35, 1, 31
         sxt4 r40 = r35
         ;;
         add r39 = r39, r15
         br .L21
         ;;
 .L8:
         extr r39 = r35, 1, 31
         sxt4 r40 = r35
         ;;
         sub r39 = r15, r39
         ;;
 .L21:
         sxt4 r39 = r39
         br.call.sptk.many b0 = __divdi3#
         ;;
         addl r14 = 62, r0
         st2 [r32] = r8
         addl r19 = 1, r0
         ;;
         mov ar.lc = r14
         ;;
 .L20:
         dep.z r17 = r19, 1, 32
         ;;
         add r15 = r17, r33
         adds r19 = 1, r19
         ;;
         ld2 r14 = [r15]
         ;;
         sxt2 r14 = r14
         ;;
         mov r16 = r14
         mov r18 = r14
         ;;
         sub r15 = r0, r16
         cmp4.le p8, p9 = r36, r16
         cmp4.le p6, p7 = r0, r16
         ;;
         sxt2 r14 = r15
         (p6) br.cond.dptk .L14
         ;;
         mov r16 = r14
         add r18 = r17, r32
         ;;
         setf.sig f6 = r16
         cmp4.le p6, p7 = r36, r16
         mov r15 = r18
         ;;
         xma.l f6 = f6, f2, f0
         (p7) st2 [r18] = r0
         ;;
         getf.sig r14 = f6
         ;;
         extr r14 = r14, 16, 16
         ;;
         sub r14 = r0, r14
         ;;
         (p6) st2 [r15] = r14
         br .L12
 .L14:
         .pred.rel "mutex", p8, p9
         setf.sig f6 = r18
         add r16 = r17, r32
         ;;
         xma.l f6 = f6, f2, f0
         mov r15 = r16
         (p9) st2 [r16] = r0
         ;;
         getf.sig r14 = f6
         ;;
         extr r14 = r14, 16, 16
         ;;
         (p8) st2 [r15] = r14
 .L12:
         br.cloop.sptk.few .L20
         adds r18 = 24, r12
         ;;
         ld8 r19 = [r18], 8
         mov ar.pfs = r38
         mov b0 = r37
         ;;
         mov ar.lc = r19
         ldf.fill f2 = [r18]
         .restore sp
         adds r12 = 32, r12
         br.ret.sptk.many b0
         .endp quant_h263_intra_ia64#
         .common quant_h263_intra#,8,8
         .common dequant_h263_intra#,8,8
         .align 16
         .global dequant_h263_intra_ia64#
         .proc dequant_h263_intra_ia64#
 dequant_h263_intra_ia64:
         .prologue
         ld2 r14 = [r33]
         andcm r15 = 1, r34
         setf.sig f8 = r35
         ;;
         sxt2 r14 = r14
         sub r15 = r34, r15
         addl r16 = -2048, r0
         ;;
         setf.sig f6 = r14
         setf.sig f7 = r15
         shladd r34 = r34, 1, r0
         ;;
         xma.l f8 = f6, f8, f0
         .save ar.lc, r2
         mov r2 = ar.lc
         ;;
         .body
         getf.sig r14 = f8
         setf.sig f6 = r34
         ;;
         sxt2 r15 = r14
         st2 [r32] = r14
         ;;
         cmp4.le p6, p7 = r16, r15
         ;;
         (p7) st2 [r32] = r16
         (p7) br.cond.dptk .L32
         addl r14 = 2047, r0
         ;;
         cmp4.ge p6, p7 = r14, r15
         ;;
         (p7) st2 [r32] = r14
 .L32:
         addl r14 = 62, r0
         addl r19 = 1, r0
         addl r22 = 2048, r0
         addl r21 = -2048, r0
         addl r20 = 2047, r0
         ;;
         mov ar.lc = r14
         ;;
 .L56:
         dep.z r16 = r19, 1, 32
         ;;
         add r14 = r16, r33
         add r17 = r16, r32
         adds r19 = 1, r19
         ;;
         ld2 r15 = [r14]
         ;;
         sxt2 r15 = r15
         ;;
         cmp4.ne p6, p7 = 0, r15
         cmp4.le p8, p9 = r0, r15
         ;;
         (p7) st2 [r17] = r0
         (p7) br.cond.dpnt .L36
         add r18 = r16, r32
         sub r17 = r0, r15
         ;;
         mov r14 = r18
         (p8) br.cond.dptk .L40
         setf.sig f8 = r17
         ;;
         xma.l f8 = f6, f8, f7
         ;;
         getf.sig r15 = f8
         ;;
         cmp4.lt p6, p7 = r22, r15
         sub r16 = r0, r15
         ;;
         (p7) st2 [r14] = r16
         (p6) st2 [r14] = r21
         br .L36
 .L40:
         setf.sig f8 = r15
         ;;
         xma.l f8 = f6, f8, f7
         ;;
         getf.sig r15 = f8
         ;;
         cmp4.le p6, p7 = r20, r15
         ;;
         (p6) mov r14 = r20
         (p7) mov r14 = r15
         ;;
         st2 [r18] = r14
 .L36:
         br.cloop.sptk.few .L56
         ;;