static inline void ifft4(buf)
{
uint32_t tm4=(uint32_t)(buf);
S32LDI(xr1, tm4, 0);
S32LDI(xr2, tm4, 8);
S32LDI(xr3, tm4, 8);
S32LDI(xr4, tm4, 8);
D32ADD_AS(xr5, xr1, xr2, xr6);
D32ADD_AS(xr7, xr4, xr3, xr8);
D32ADD_AS(xr9, xr5, xr7, xr10);
S32SDI(xr10,tm4,-8);
S32SDI(xr9,tm4,-16);
S32LDI(xr1, tm4, 4);
S32LDI(xr2, tm4, 8);
S32LDI(xr3, tm4, 8);
S32LDI(xr4, tm4, 8);
D32ADD_AS(xr5, xr1, xr2, xr9);
D32ADD_AS(xr7, xr3, xr4, xr10);
D32ADD_AS(xr1,xr5,xr7,xr2);
D32ADD_AS(xr11,xr6,xr10,xr12);
D32ADD_AS(xr13,xr9,xr8,xr14);
S32SDI(xr14,tm4,0);
S32SDI(xr12,tm4,-4);
S32SDI(xr2, tm4,-4);
S32SDI(xr13,tm4,-8);
S32SDI(xr11,tm4,-4);
S32SDI(xr1, tm4,-4);
}
static void rv40_dequant4x4(DCTELEM *block,uint32_t *dst, int n)
{
int i;
uint32_t src=block-4;
uint32_t dst_t = dst-4;
#if 0
for(i = 0; i < n; i++){
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_LW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
}
#else
/////////////////////
if(n == 1){
S32LDI(xr1,src,0x8);
S32LDI(xr2,src,0x8);
S32LDI(xr7,src,0x8);
S32LDI(xr8,src,0x8);
D16MUL_XW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
S32SDI(xr5,dst_t,0x10);
S32STD(xr6,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
//S32STD(xr15,dst_t,0x8);
//S32STD(xr14,dst_t,0xc);
S32SDI(xr15,dst_t,0x10);
S32STD(xr14,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
}
else if(n==2)
{
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_XW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
S32SDI(xr0,dst_t,0x10);
S32STD(xr0,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
S32SDI(xr0,dst_t,0x10);
S32STD(xr0,dst_t,0x4);
S32STD(xr0,dst_t,0x8);
S32STD(xr0,dst_t,0xc);
}
else
{
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_XW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
S32LDI(xr1,src,0x8);
S32LDD(xr2,src,0x4);
S32LDI(xr7,src,0x8);
S32LDD(xr8,src,0x4);
D16MUL_LW(xr4,xr12,xr1,xr3);
D16MUL_LW(xr10,xr12,xr7,xr9);
D16MUL_LW(xr14,xr12,xr8,xr15);
D32ASUM_AA(xr3,xr13,xr13,xr4);
D16MUL_LW(xr6,xr12,xr2,xr5);
D32SLR(xr3,xr3,xr4,xr4,4);
D32ASUM_AA(xr5,xr13,xr13,xr6);
D32ASUM_AA(xr9,xr13,xr13,xr10);
D32SLR(xr5,xr5,xr6,xr6,4);
D32SLR(xr9,xr9,xr10,xr10,4);
D32ASUM_AA(xr15,xr13,xr13,xr14);
S32SDI(xr3,dst_t,0x10);
S32STD(xr4,dst_t,0x4);
S32STD(xr5,dst_t,0x8);
S32STD(xr6,dst_t,0xc);
D32SLR(xr15,xr15,xr14,xr14,4);
S32SDI(xr9,dst_t,0x10);
S32STD(xr10,dst_t,0x4);
S32STD(xr15,dst_t,0x8);
S32STD(xr14,dst_t,0xc);
}
#endif
}
void ff_init_cabac_states_mxu(CABACContext *c){
uint32_t i = 0;
uint8_t *p_lps_range = lps_range[0]-4;
uint8_t *p_mps_state = mps_state-4;
uint8_t *p_lps_state = lps_state-4;
uint8_t *p_ff_h264_mlps_state1 = &ff_h264_mlps_state[0]+124;
uint8_t *p_ff_h264_mlps_state2 = &ff_h264_mlps_state[0]+128;//
uint8_t *p_ff_h264_lps_range = ff_h264_lps_range-128;
S32I2M(xr15,0x1010101);
for(i=0; i<16; i++){
/* load 32 */
S32LDI(xr1,p_lps_range,4);
S32LDI(xr2,p_lps_range,4);
S32LDI(xr3,p_lps_range,4);
S32LDI(xr4,p_lps_range,4);
S32SFL(xr7,xr2,xr1,xr8,0);
S32SFL(xr9,xr4,xr3,xr10,0);
S32SFL(xr11,xr7,xr7,xr12,0);
S32SFL(xr13,xr8,xr8,xr14,0);
S32SFL(xr1,xr9,xr9,xr2,0);
S32SFL(xr3,xr10,xr10,xr4,0);
S32SDI(xr14,p_ff_h264_lps_range,128);
S32SDI(xr4,p_ff_h264_lps_range,4);
S32SDI(xr13,p_ff_h264_lps_range,124);
S32SDI(xr3,p_ff_h264_lps_range,4);
S32SDI(xr12,p_ff_h264_lps_range,124);
S32SDI(xr2,p_ff_h264_lps_range,4);
S32SDI(xr11,p_ff_h264_lps_range,124);
S32SDI(xr1,p_ff_h264_lps_range,4);
p_ff_h264_lps_range-=128*4-4;
/* part2 */
S32LDI(xr1,p_mps_state,4);
Q8ADD_AA(xr3,xr1,xr1);// 2*mps_state[i]+0
Q8ADD_AA(xr4,xr3,xr15);//2*mps_state[i]+1
S32SFL(xr11,xr4,xr3,xr12,0);
S32SDI(xr12,p_ff_h264_mlps_state1,4);// ff_h264_mlps_state[128+2*i+1]= 2*mps_state[i]+0;
S32SDI(xr11,p_ff_h264_mlps_state1,4);
/*part3 */
S32LDI(xr1,p_lps_state,4);
Q8ADD_AA(xr3,xr1,xr1);// 2*mps_state[i]+0
Q8ADD_AA(xr4,xr3,xr15);//2*mps_state[i]+1
S32SFL(xr11,xr3,xr4,xr12,0);
S32ALN(xr11,xr11,xr11,2);
S32ALN(xr12,xr12,xr12,2);
S32SDI(xr12,p_ff_h264_mlps_state2,-4);
S32SDI(xr11,p_ff_h264_mlps_state2,-4);
}
ff_h264_mlps_state[127]= 1;
ff_h264_mlps_state[126]= 0;
}