/**********************************************

   該函數的功能主要是進行整幀的圖象內插
   在該函數內一次完成，水平，垂直和交叉
   內插。
完成r8*8的基礎塊內插耗時為：374cycles
此函數比原函數增加四句，在后面有標注
**********************************************/
.section L1_code;
.global __fullinterpolate;
.align 8;
__fullinterpolate:
       [--SP] = (R7:4,P5:3);
       P0 = R0;                //輸入數據的地址
       P1 = R1;                //水平內插的輸出地址
       P2 = R2;                //垂直內插的輸出地址
       P3 = [SP+40];           //交叉內插的輸出地址
       P4 = [SP+44];           //Whide
       R3 = [SP+48];           //Hight
       R4 = [SP+52];           //stride
       M0 = R3;     //高
       M1 = R4;     //跨距
       R5 = P4;     //寬
       I2 = P0;     //輸入數據
       L2 = 0;
       I3 = M0;     //高
       P5 = I3;     //高
       M3 = 1;
       R3 = R4-R5(S)||I2 += M1;             //stride-Whide；I2指向下一行
       M2 = R3;                         //stride-Whide
       I3 = P0;                       //輸入數據
       I1 = M2;          //比原函數添加句1
       I1+=4;           //添加句2
       P0 = I1;         //添加句3
       I0 = 0;
       LSETUP(INSERTO_ST,INSERTO_END) LC1 = P5; //核心算法開始
INSERTO_ST:
       P5 = P4;
       P5 = P5>>2; 
       R6=B[P5--](x);     //添加句4                 //調整循環的次數,相當于P5=P5-1,將內循環的最后一次循環放在后面執行;
       LSETUP(INSERTI_ST,INSERTI_END) LC0 = P5;
INSERTI_ST:
       R0 = [I3++]||R2 = [I2++];                //P0指向輸入數據的第n行
                                                //I2指向輸入數據的第n+1行
       I1 = 0;
       R4 = BYTEOP1P(R1:0,R3:2)||R1 = [I3--]||R3 = [I2];                //垂直內插的計算
       R4 = BYTEOP2P(R1:0,R3:2)(RNDL)||[P2++] = R4;          //交叉內插的計算
       I0 = 1;
       I1 = 1;
       R5 = BYTEOP2P(R1:0,R3:2)(RNDH);          //4四個8位數取平均結果放在高位字節
       R4 = R4+|+R5||I0 -= M3||R2 =[I3++];    //I0=0
       R3 = R1;
       R4 = BYTEOP1P(R1:0,R3:2)||[P3++] = R4;                //水平內插的計算
INSERTI_END: [P1++] =R4;
       R0 = [I3++]||R2 = [I2++];                //P0指向輸入數據的第n行
                                                //I2指向輸入數據的第n+1行
       I1 = 0;
       R4 = BYTEOP1P(R1:0,R3:2)||R1 = [I3--]||R3 = [I2];                //垂直內插的計算
       R6 = BYTEOP2P(R1:0,R3:2)(RNDL);          //交叉內插的計算
       [P2++P0] = R4;
       I0 = 1;
       I1 = 1;
       R7 = BYTEOP2P(R1:0,R3:2)(RNDH);          //4四個8位數取平均結果放在高位字節
       R6 = R6+|+R7||I0 -= M3||R2 =[I3++];    //I0=0
       R3 = R1;
       R4 = BYTEOP1P(R1:0,R3:2);                //水平內插的計算
       [P3++P0] = R6;
       [P1++P0] =R4;                         //輸出數據換行，原函數編程只是輸入數據換行，沒有實現輸出數據換行
                           
                                 
INSERTO_END:  R7 = [I2++M2]||R6 = [I3++M2]; //輸入的數據換行
      (R7:4,P5:3) = [SP++];
       RTS;                                    //函數返回
       NOP; 
__fullinterpolate.end: