vc1dsp.c

君正早期ucos系统(只有早期的才不没有打包成库),MPLAYER,文件系统,图片解码,浏览,电子书,录音,想学ucos,识货的人就下吧 russblock fmradio explore set

{
 vc1_mspel_mc(dst,src,stride,1,0,rnd);
} 
*/
//PUT_VC1_MSPEL(2, 0)
/*static void put_vc1_mspel_mc20_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{

 vc1_mspel_mc(dst,src,stride,2,0,rnd);

} */
#ifdef JZ4740_MXU_OPT
static void put_vc1_mspel_mc20_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{
  uint32_t  i,j;
  uint32_t src_aln0,src_aln1,src_rs0,src_rs1,src_rs2,src_rs3;
  int mul_ct0,mul_ct1;
  int ct0,rnd1;
  mul_ct0=0x09090909;   //9 9 9 9
  mul_ct1=0x01010101;   //1 1 1 1
  ct0 = 0x00080008;
  rnd1   = rnd * 0x00010001;
  src_aln0 = (((uint32_t)src-1) & 0xFFFFFFFC);
  src_aln1 = (((uint32_t)src+1) & 0xFFFFFFFC);
  src_rs0 = 4-(((uint32_t)src - 1) & 3);
  src_rs1 = src_rs0-1;
  src_rs2 = 4-(((int)src + 1) & 3);
  src_rs3 = src_rs2-1; 
  S32I2M(xr15,mul_ct0); //xr15:9 9 9 9
  S32I2M(xr14,mul_ct1); //xr14:1 1 1 1
  S32I2M(xr13,ct0);     //xr13:  8   8
  S32I2M(xr12,rnd1);     //xr12:rnd rnd
  for(j=0;j<8;j++){
   S32LDD(xr1,src_aln0,0);
   S32LDD(xr2,src_aln0,4);
   
   S32LDD(xr5,src_aln1,0);       
   S32LDD(xr6,src_aln1,4);      
   
   S32ALN(xr3,xr2,xr1,src_rs0);  //xr3:src[2] src[1] src[0] src[-1]
   S32ALN(xr4,xr2,xr1,src_rs1);  //xr4:src[3] src[2] src[1] src[0]
   S32ALN(xr7,xr6,xr5,src_rs2);  //xr7:src[4] src[3] src[2] src[1]
   S32ALN(xr8,xr6,xr5,src_rs3);  //xr8:src[5] src[4] src[3] src[2]
   Q8MUL(xr5,xr4,xr15,xr6);   //xr5:9*src[3] 9*src[2] xr6:9*src[1] 9*src[0]
   Q8MAC_SS(xr5,xr3,xr14,xr6);//xr5:9*src[3]-src[2] 9*src[2]-src[1] xr6:9*src[1]-src[0] 9*src[0]-src[-1]
   Q8MUL(xr9,xr7,xr15,xr10);  //xr9:9*src[4] 9*src[3] xr10:9*src[2] 9*src[1]
   Q8MAC_SS(xr9,xr8,xr14,xr10); //xr9:9*src[4]-src[5] 9*src[3]-src[4] xr10:9*src[2]-src[3] 9*src[1]-src[2]
   Q16ADD_AA_WW(xr0,xr5,xr9,xr5); //xr5:dst[3] 9*src[3]-src[2]+9*src[4]-src[5] dst[2] 9*src[2]-src[1]+9*src[3]-src[4]
   Q16ADD_AA_WW(xr0,xr6,xr10,xr6); //xr6:dst[1] 9*src[1]-src[0]+9*src[2]-src[3] dst[0] 9*src[0]-src[-1]+9*src[1]-src[2]
   Q16ACC_SS(xr5,xr13,xr12,xr6); //two instructions with one
   Q16SAR(xr5,xr5,xr6,xr6,4);
   Q16SAT(xr5,xr5,xr6);      //xr5:dst[3]dst[2]dst[1]dst[0]
   S32LDI(xr1,src_aln0,4);
   S32LDD(xr2,src_aln0,4);
   S32ALN(xr3,xr2,xr1,src_rs0);  //xr3:src[6] src[5] src[4] src[3]
   S32ALN(xr4,xr2,xr1,src_rs1);  //xr4:src[7] src[6] src[5] src[4]

   S32LDI(xr1,src_aln1,4);
   S32LDD(xr2,src_aln1,4);
   S32ALN(xr7,xr2,xr1,src_rs2);  //xr7:src[8] src[7] src[6] src[5]
   S32ALN(xr8,xr2,xr1,src_rs3);  //xr8:src[9] src[8] src[7] src[6]
   Q8MUL(xr1,xr4,xr15,xr2);   //xr5:9*src[3] 9*src[2] xr6:9*src[1] 9*src[0]
   Q8MAC_SS(xr1,xr3,xr14,xr2);//xr5:9*src[3]-src[2] 9*src[2]-src[1] xr6:9*src[1]-src[0] 9*src[0]-src[-1]

   Q8MUL(xr9,xr7,xr15,xr10);  //xr9:9*src[4] 9*src[3] xr10:9*src[2] 9*src[1]
   Q8MAC_SS(xr9,xr8,xr14,xr10); //xr9:9*src[4]-src[5] 9*src[3]-src[4] xr10:9*src[2]-src[3] 9*src[1]-src[2]
   Q16ADD_AA_WW(xr0,xr1,xr9,xr1); //xr5:dst[3] 9*src[3]-src[2]+9*src[4]-src[5] dst[2] 9*src[2]-src[1]+9*src[3]-src[4]

   Q16ADD_AA_WW(xr0,xr2,xr10,xr2); //xr6:dst[1] 9*src[1]-src[0]+9*src[2]-src[3] dst[0] 9*src[0]-src[-1]+9*src[1]-src[2]
   Q16ACC_SS(xr1,xr13,xr12,xr2);
   Q16SAR(xr1,xr1,xr2,xr2,4);
   Q16SAT(xr1,xr1,xr2);      //xr5:dst[3]dst[2]dst[1]dst[0]
   S32STD(xr5,dst,0);
   S32STD(xr1,dst,4);
   dst+=stride;
   src_aln0+=(stride-4);
   src_aln1+=(stride-4);
}

}
#else
static void put_vc1_mspel_mc20_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{

int i,j;
for(j = 0; j < 8; j++) 
   {
       dst[0]=av_clip_uint8((-src[-1] + 9*src[0] + 9*src[1] - src[2] + 8 - rnd) >> 4);
       dst[1]=av_clip_uint8((-src[0]  + 9*src[1] + 9*src[2] - src[3] + 8 - rnd) >> 4);
       dst[2]=av_clip_uint8((-src[1]  + 9*src[2] + 9*src[3] - src[4] + 8 - rnd) >> 4);
       dst[3]=av_clip_uint8((-src[2]  + 9*src[3] + 9*src[4] - src[5] + 8 - rnd) >> 4);
       dst[4]=av_clip_uint8((-src[3]  + 9*src[4] + 9*src[5] - src[6] + 8 - rnd) >> 4);
       dst[5]=av_clip_uint8((-src[4]  + 9*src[5] + 9*src[6] - src[7] + 8 - rnd) >> 4);
       dst[6]=av_clip_uint8((-src[5]  + 9*src[6] + 9*src[7] - src[8] + 8 - rnd) >> 4);
       dst[7]=av_clip_uint8((-src[6]  + 9*src[7] + 9*src[8] - src[9] + 8 - rnd) >> 4);
       dst += stride;
       src += stride;
   }

}
#endif
 
//PUT_VC1_MSPEL(3, 0)
/*static void put_vc1_mspel_mc30_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{
  vc1_mspel_mc(dst,src,stride,3,0,rnd);
} */

#ifdef JZ4740_MXU_OPT
static void put_vc1_mspel_mc30_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{
  uint32_t  i,j;
  uint32_t src_aln0,src_aln1,src_rs0,src_rs1,src_rs2,src_rs3;
  int mul_ct0,mul_ct1,mul_ct2,mul_ct3;
  int ct0,rnd1;
  mul_ct0=0x12121212;   //18 18 18 18
  mul_ct1=0x35353535;   //53 53 53 53
  mul_ct2=0x03030303;   //3  3  3  3
  mul_ct3=0x04040404;   //4  4  4  4
  ct0 = 0x00200020;     //32    32
  rnd1   = rnd * 0x00010001;
  src_aln0 = (((uint32_t)src-1) & 0xFFFFFFFC);
  src_aln1 = (((uint32_t)src+1) & 0xFFFFFFFC);
  src_rs0 = 4-(((uint32_t)src - 1) & 3);
  src_rs1 = src_rs0-1;
  src_rs2 = 4-(((int)src + 1) & 3);
  src_rs3 = src_rs2-1; 
  S32I2M(xr15,mul_ct0); //xr15:18 18 18 18
  S32I2M(xr14,mul_ct1); //xr14:53 53 53 53
  S32I2M(xr11,mul_ct2); //xr11:3  3  3  3
  S32I2M(xr10,mul_ct3); //xr10:4  4  4  4
  S32I2M(xr13,ct0);     //xr13:  32  32
  S32I2M(xr12,rnd1);    //xr12:rnd rnd
  for(j=0;j<8;j++){
   S32LDD(xr1,src_aln0,0);
   S32LDD(xr2,src_aln0,4);
   
   S32LDD(xr5,src_aln1,0);       
   S32LDD(xr6,src_aln1,4);      
  
   S32ALN(xr3,xr2,xr1,src_rs0);  //xr3:src[2] src[1] src[0] src[-1]
   S32ALN(xr4,xr2,xr1,src_rs1);  //xr4:src[3] src[2] src[1] src[0]
   S32ALN(xr7,xr6,xr5,src_rs2);  //xr7:src[4] src[3] src[2] src[1]
   S32ALN(xr8,xr6,xr5,src_rs3);  //xr8:src[5] src[4] src[3] src[2]
   Q8MUL(xr5,xr4,xr15,xr6);   //xr5:18*src[3] 18*src[2] xr6:18*src[1] 18*src[0]
   Q8MAC_SS(xr5,xr3,xr11,xr6);//xr5:18*src[3]-3*src[2] 18*src[2]-3*src[1] xr6:18*src[1]-3*src[0] 18*src[0]-3*src[-1]
   Q8MUL(xr9,xr7,xr14,xr1);  //xr9:53*src[4] 53*src[3] xr1:53*src[2] 53*src[1]
   Q8MAC_SS(xr9,xr8,xr10,xr1); //xr9:53*src[4]-4*src[5] 53*src[3]-4*src[4] xr1:53*src[2]-4*src[3] 53*src[1]-4*src[2]
   Q16ADD_AA_WW(xr0,xr5,xr9,xr5); //xr5:dst[3] 18*src[3]-3*src[2]+53*src[4]-4*src[5] dst[2] 18*src[2]-3*src[1]+53*src[3]-4*src[4]
   Q16ADD_AA_WW(xr0,xr6,xr1,xr6); //xr6:dst[1] 18*src[1]-3*src[0]+53*src[2]-4*src[3] dst[0] 18*src[0]-3*src[-1]+53*src[1]-4src[2]
   Q16ACC_SS(xr5,xr13,xr12,xr6); //two instructions with one
   Q16SAR(xr5,xr5,xr6,xr6,6);
   Q16SAT(xr5,xr5,xr6);      //xr5:dst[3]dst[2]dst[1]dst[0]
   S32LDI(xr1,src_aln0,4);
   S32LDD(xr2,src_aln0,4);
   S32ALN(xr3,xr2,xr1,src_rs0);  //xr3:src[6] src[5] src[4] src[3]
   S32ALN(xr4,xr2,xr1,src_rs1);  //xr4:src[7] src[6] src[5] src[4]

   S32LDI(xr1,src_aln1,4);
   S32LDD(xr2,src_aln1,4);
   S32ALN(xr7,xr2,xr1,src_rs2);  //xr7:src[8] src[7] src[6] src[5]
   S32ALN(xr8,xr2,xr1,src_rs3);  //xr8:src[9] src[8] src[7] src[6]
   Q8MUL(xr1,xr4,xr15,xr2);   //xr1:18*src[7] 18*src[6] xr2:18*src[5] 18*src[4]
   Q8MAC_SS(xr1,xr3,xr11,xr2);//xr1:18*src[7]-3*src[6] 18*src[6]-3*src[5] xr2:18*src[5]-3*src[4] 18*src[4]-3*src[3]

   Q8MUL(xr9,xr7,xr14,xr3);  //xr9:53*src[8] 53*src[7] xr3:53*src[6] 53*src[5]
   Q8MAC_SS(xr9,xr8,xr10,xr3); //xr9:53*src[8]-4*src[9] 53*src[7]-4*src[8] xr3:53*src[6]-4*src[7] 53*src[5]-4*src[6]
   Q16ADD_AA_WW(xr0,xr1,xr9,xr1); //xr1:dst[3] 18*src[7]-3*src[6]+53*src[8]-4*src[9]  dst[2] 18*src[6]-3*src[5]+53*src[7]-4*src[8]

   Q16ADD_AA_WW(xr0,xr2,xr3,xr2); //xr2:dst[1] 18*src[5]-3*src[4]+53*src[6]-4*src[7]  dst[0] 18*src[4]-3*src[3]53+*src[5]-4*src[6]
   Q16ACC_SS(xr1,xr13,xr12,xr2);
   Q16SAR(xr1,xr1,xr2,xr2,6);
   Q16SAT(xr1,xr1,xr2);      //xr1:dst[7]dst[6]dst[5]dst[4]
   
   S32STD(xr5,dst,0);
   S32STD(xr1,dst,4);
   dst+=stride;
   src_aln0+=(stride-4);
   src_aln1+=(stride-4);
}
}
#else
static void put_vc1_mspel_mc30_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{
int j;
for(j = 0; j < 8; j++) 
   {
       dst[0]=av_clip_uint8((-3*src[-1]+ 18*src[0] + 53*src[1] - 4*src[2] + 32 - rnd) >> 6);
       dst[1]=av_clip_uint8((-3*src[0] + 18*src[1] + 53*src[2] - 4*src[3] + 32 - rnd) >> 6);
       dst[2]=av_clip_uint8((-3*src[1] + 18*src[2] + 53*src[3] - 4*src[4] + 32 - rnd) >> 6);
       dst[3]=av_clip_uint8((-3*src[2] + 18*src[3] + 53*src[4] - 4*src[5] + 32 - rnd) >> 6);
       dst[4]=av_clip_uint8((-3*src[3] + 18*src[4] + 53*src[5] - 4*src[6] + 32 - rnd) >> 6);
       dst[5]=av_clip_uint8((-3*src[4] + 18*src[5] + 53*src[6] - 4*src[7] + 32 - rnd) >> 6);
       dst[6]=av_clip_uint8((-3*src[5] + 18*src[6] + 53*src[7] - 4*src[8] + 32 - rnd) >> 6);
       dst[7]=av_clip_uint8((-3*src[6] + 18*src[7] + 53*src[8] - 4*src[9] + 32 - rnd) >> 6);
       dst += stride;
       src += stride;
   }
}
#endif
 
//PUT_VC1_MSPEL(0, 1)
/*static void put_vc1_mspel_mc01_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{
  vc1_mspel_mc(dst,src,stride,0,1,rnd);
} */
#ifdef JZ4740_MXU_OPT
static void put_vc1_mspel_mc01_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{
  uint32_t  j,r;
  r = 1-rnd;
  uint32_t src_aln0,src_rs0;
  int mul_ct0,mul_ct1,mul_ct2,mul_ct3;
  int ct0,rnd1;
  mul_ct1=0x12121212;   //18 18 18 18
  mul_ct0=0x35353535;   //53 53 53 53
  mul_ct3=0x03030303;   //3  3  3  3
  mul_ct2=0x04040404;   //4  4  4  4
  ct0 = 0x00200020;     //32    32
  rnd1 = r * 0x00010001;
  src_aln0 = (((uint32_t)src - stride) & 0xFFFFFFFC);
  src_rs0 = 4-(((uint32_t)src - stride) & 3);
  S32I2M(xr15,mul_ct0); //xr15:53 53 53 53
  S32I2M(xr14,mul_ct1); //xr14:18 18 18 18
  S32I2M(xr11,mul_ct2); //xr11:4  4  4  4
  S32I2M(xr10,mul_ct3); //xr10:3  3  3  3
  S32I2M(xr13,ct0);     //xr13:  32  32
  S32I2M(xr12,rnd1);    //xr12:   r   r
  for(j=0;j<8;j++){
   S32LDD(xr1,src_aln0,0);
   S32LDD(xr2,src_aln0,4);
   
   S32LDIV(xr5,src_aln0,stride,0);       
   S32LDD(xr6,src_aln0,4);      
  
   S32ALN(xr3,xr2,xr1,src_rs0);  
   S32ALN(xr7,xr6,xr5,src_rs0);  
  
   S32LDIV(xr1,src_aln0,stride,0);
   S32LDD(xr2,src_aln0,4);

   S32LDIV(xr5,src_aln0,stride,0);
   S32LDD(xr6,src_aln0,4);
   
   S32ALN(xr4,xr2,xr1,src_rs0);
   S32ALN(xr8,xr6,xr5,src_rs0);
   
   Q8MUL(xr1,xr7,xr15,xr2);   
   Q8MAC_SS(xr1,xr3,xr11,xr2); //xr1:-4*src[-stride+3] + 53*src[3] -4*src[-stride+2] + 53*src[2] xr2:
   Q8MUL(xr5,xr4,xr14,xr6);   
   Q8MAC_SS(xr5,xr8,xr10,xr6); //xr5:18*src[stride+3]-3*src[stride*2+3] 18*src[stride+2]-3*src[stride*2+2] xr6: 
 
   Q16ADD_AA_WW(xr0,xr5,xr1,xr5); 
   Q16ADD_AA_WW(xr0,xr6,xr2,xr6); 
  
   Q16ACC_SS(xr5,xr13,xr12,xr6); //two instructions with one
   Q16SAR(xr5,xr5,xr6,xr6,6);
   Q16SAT(xr5,xr5,xr6);      //xr5:dst[3]dst[2]dst[1]dst[0]
   src_aln0 = src_aln0 - 3*stride;   
   S32LDI(xr1,src_aln0,4);
   S32LDD(xr2,src_aln0,4);
   
   S32LDIV(xr9,src_aln0,stride,0);       
   S32LDD(xr6,src_aln0,4);      
   S32ALN(xr3,xr2,xr1,src_rs0);  
   S32ALN(xr7,xr6,xr9,src_rs0);  
  
   S32LDIV(xr1,src_aln0,stride,0);
   S32LDD(xr2,src_aln0,4);

   S32LDIV(xr9,src_aln0,stride,0);
   S32LDD(xr6,src_aln0,4);
   
   S32ALN(xr4,xr2,xr1,src_rs0);
   S32ALN(xr8,xr6,xr9,src_rs0);
   
   Q8MUL(xr1,xr7,xr15,xr2);   
   Q8MAC_SS(xr1,xr3,xr11,xr2); //xr1:-4*src[-stride+3] + 53*src[3] -4*src[-stride+2] + 53*src[2] xr2:
   Q8MUL(xr9,xr4,xr14,xr6);   
   Q8MAC_SS(xr9,xr8,xr10,xr6); //xr5:18*src[stride+3]-3*src[stride*2+3] 18*src[stride+2]-3*src[stride*2+2] xr6: 
 
   Q16ADD_AA_WW(xr0,xr9,xr1,xr9); 
   Q16ADD_AA_WW(xr0,xr6,xr2,xr6); 
  
   Q16ACC_SS(xr9,xr13,xr12,xr6); //two instructions with one
   Q16SAR(xr9,xr9,xr6,xr6,6);
   Q16SAT(xr9,xr9,xr6);      //xr5:dst[3]dst[2]dst[1]dst[0]
  
   S32STD(xr5,dst,0);
   S32STD(xr9,dst,4);
   dst+=stride;
   src_aln0+=(-2*stride-4);
  }
}
#else 
static void put_vc1_mspel_mc01_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{   int j,r;
    r = 1-rnd;
    for(j = 0; j < 8; j++) 
       {
     dst[0] = av_clip_uint8((-4*src[-stride] + 53*src[0] + 18*src[stride] - 3*src[stride*2] + 32 - r) >> 6);   
     dst[1] = av_clip_uint8((-4*src[-stride+1] + 53*src[1] + 18*src[stride+1] - 3*src[stride*2+1] + 32 - r) >> 6);   
     dst[2] = av_clip_uint8((-4*src[-stride+2] + 53*src[2] + 18*src[stride+2] - 3*src[stride*2+2] + 32 - r) >> 6);       
     dst[3] = av_clip_uint8((-4*src[-stride+3] + 53*src[3] + 18*src[stride+3] - 3*src[stride*2+3] + 32 - r) >> 6);       
     dst[4] = av_clip_uint8((-4*src[-stride+4] + 53*src[4] + 18*src[stride+4] - 3*src[stride*2+4] + 32 - r) >> 6);       
     dst[5] = av_clip_uint8((-4*src[-stride+5] + 53*src[5] + 18*src[stride+5] - 3*src[stride*2+5] + 32 - r) >> 6);       
     dst[6] = av_clip_uint8((-4*src[-stride+6] + 53*src[6] + 18*src[stride+6] - 3*src[stride*2+6] + 32 - r) >> 6);       
     dst[7] = av_clip_uint8((-4*src[-stride+7] + 53*src[7] + 18*src[stride+7] - 3*src[stride*2+7] + 32 - r) >> 6);       
     src += stride;
     dst += stride;
      }
}
#endif
//PUT_VC1_MSPEL(1, 1)
/*
static void put_vc1_mspel_mc11_c(uint8_t *dst,const uint8_t *src,int stride,int rnd)
{
  vc1_mspel_mc(dst,src,stride,1,1,rnd);
} */
 
#ifdef JZ4740_MXU_OPT
#define w1 53
#define w2 4
#define w3 18
#define w4 3