scale_h2_h.asm

基于DM642平台的视频缩小放大功能 程序源代码

* ========================================================================= *
*   NAME                                                                    *
*       scale_h2 -- Horizontal scaling by 2                                 *
*                                                                           *
*   USAGE                                                                   *
*       This routine has the following C prototype:                         *
*                                                                           *
*       void scale_h2_cn(                                                   *
*                            const unsigned char *restrict inp,             *
*                            int   cols,                                    *
*                            unsigned char *restrict outp                   *
*                        )                                                  *
*                                                                           *
*                                                                           *
*       The scale_h2 function accepts an input array "inp" that contains    *
*       "cols" contiguous pixels and produces an output array "outp" of     *
*       "cols/2" pixels. Simple averaging and rounding are performed.       *
*       This step may also be viewed as simple bi-linear interpolation.     *
*                                                                           *
*   DESCRIPTION                                                             *
*       The following C code model shows the behaviour of the code.         *
*                                                                           *
*       void scale_h2_cn(                                                   *
*                          const unsigned char *restrict inp,               *
*                          int   cols,                                      *
*                          unsigned char *restrict outp                     *
*                        )                                                  *
*        {                                                                  *
*            int i;                                                         *
*                                                                           *
*            for( i = 0; i < (cols >> 1); i++)                              *
*            {                                                              *
*                outp[i] = (inp[2*i] + inp[(2*i)+1] + 1) >> 1;              *
*            }                                                              *
*        }                                                                  *
*                                                                           *
*      Every two consecutive pixels of the input image are averaged to      *
*      form an output pixel.                                                *
*                                                                           *
*   TECHNIQUES                                                              *
*      This code can be used to process either one line, or an arbitrary    *
*      number of lines as long as they are contiguous in memory.            *
*                                                                           *
*   ASSUMPTIONS                                                             *
*      The input array is aligned on a double-word boundary. The output     *
*      array is aligned on a double word boundary. The number of input      *
*      pixels is a multiple of 32.                                          *
*                                                                           *
*   NOTES                                                                   *
*     None                                                                  *
*                                                                           *
*   SOURCE                                                                  *
*     None                                                                  *
*                                                                           *
*   MEMORY NOTE                                                             *
*      No bank conflcits should occurr irrespective of the alignemt of      *
*      the arrays in memory.                                                *
*                                                                           *
*   INTERRUPT NOTE                                                          *
*      This code is interrupt tolerant though not interruptible.            *
*                                                                           *
*   CYCLES                                                                  *
*      (5*cols)/32 - 5                                                      *
*                                                                           *
*   cols = 640,  cycles = 95.                                               *
*                                                                           *
* ------------------------------------------------------------------------- *
*             Copyright (c) 2002 Texas Instruments, Incorporated.           *
*                            All Rights Reserved.                           *
* ========================================================================= *

               .sect ".text:hand"
               .include "scale_h2_h.h64"
  
_scale_h2_asm:

*S-------------------------------------------------------------------------S*
*S                TI Proprietary Information: Internal Data                S*
*S-------------------------------------------------------------------------S*
*S Swipe6NG datecode: 020206 (12:26:58 AM) [REL]
*S Swipe6NG commandline:
*S swipe6ng -Ma -s -v -l -N -mv6400 prol.sa
*S Target architecture: "C64xx ISA"
*S-------------------------------------------------------------------------S*
*S Total register usage: 6 (a:3 b:3)
* ===================== SYMBOLIC REGISTER ASSIGNMENTS ===================== *
        .asg            A4,         A_inp                       ; Preferred
        .asg            B4,         B_cols                      ; Preferred
        .asg            A6,         A_outp                      ; Preferred
        .asg            B6,         B_inp                       ; Preferred
        .asg            B8,         B_outp                      ; Preferred
        .asg            A0,         A_pd                        ; Preferred
        .asg            B3,         B_return                    ; Return
        .asg            B30,        B_csr                       ; CSR
        .asg            B31,        B_no_gie                    ; NO GIE
* ========================================================================= *

        ADD     .L2X    A_inp,      8,          B_inp           ;[ 4,0] 
||      MVC     .S2     CSR,        B_csr                       ; CSR

        MVK     .S1     3,          A_pd                        ;[ 4,0]
||      ADD     .S2X    A_outp,     8,          B_outp          ;[ 3,0]
||      AND     .L2     B_csr,      -2,         B_no_gie        ;

        ;==== Branch occurs  

*S-------------------------------------------------------------------------S*
*S                TI Proprietary Information: Internal Data                S*
*S-------------------------------------------------------------------------S*
*S Swipe6NG datecode: 020206 (12:26:58 AM) [REL]
*S Swipe6NG commandline:
*S swipe6ng -Ma -s -v -2 -mv6400 --Bl scale_h2.cmd
*S Target architecture: "C64xx ISA"
*S-------------------------------------------------------------------------S*
*S
*S Resource MII: 4
*S
*S    L1         L1-unit hardware                             * Bound: 4
*S    S1         S1-unit hardware                             * Bound: 4
*S    D1         D1-unit hardware                               Bound: 3
*S    M1         M1-unit hardware                             * Bound: 4
*S    T1         T1-unit LD/ST transport (DAB1)                 Bound: 3
*S    L2         L2-unit hardware                             * Bound: 4
*S    S2         S2-unit hardware                             * Bound: 4
*S    D2         D2-unit hardware                               Bound: 3
*S    M2         M2-unit hardware                             * Bound: 4
*S    T2         T2-unit LD/ST transport (DAB2)                 Bound: 3
*S    M1h        M1-unit write port (hi half)                 * Bound: 4
*S    M2h        M2-unit write port (hi half)                 * Bound: 4
*S    xoper1     Unshared cross path on 1X                      Bound: 1
*S    xoper2     Unshared cross path on 2X                      Bound: 1
*S    slots      Instruction issue slots: 30                  * Bound: 4
*S
*S Recurrence MII: 3
*S
*S------------------------- Running Adjust Phase --------------------------S*
*S-------------------------------------------------------------------------S*
*S File   : scale_h2.cmd
*S ResMII : 4
*S RecMII : 3
*S MII    : 4
*S II     : 4
*S No ILP : 0
*S List   : 18
*S Length : 16
*S Start  : 0
*S Stop   : 15
*S Operations         : 30
*S Avg. para @ II     : 7.500000
*S Calc. Operations   : 30