void fDCT8x8_32f(const float* s, float* d, float* temp)
{
//for (int j = 0; j < 8; j ++)
//{
// for (int i = 0; i < 8; i ++)
// {
// temp[8*i+j] =s[8*j+i];
// }
//}
transpose8x8(s, temp);
fDCT2D8x4_32f(temp, d);
fDCT2D8x4_32f(temp + 4, d + 4);
//for (int j = 0; j < 8; j ++)
//{
// for (int i = 0; i < 8; i ++)
// {
// temp[8*i+j] =d[8*j+i];
// }
//}
transpose8x8(d, temp);
fDCT2D8x4_32f(temp, d);
fDCT2D8x4_32f(temp + 4, d + 4);
}
void fDCT8x8_32f_and_threshold(const float* s, float* d, float threshold, float* temp)
{
transpose8x8(s, temp);
/*for (int j = 0; j < 8; j ++)
{
for (int i = 0; i < 8; i ++)
{
temp[8*i+j] =s[8*j+i];
}
}*/
fDCT2D8x4_32f(temp, d);
fDCT2D8x4_32f(temp + 4, d + 4);
transpose8x8(d, temp);
/*for (int j = 0; j < 8; j ++)
{
for (int i = 0; i < 8; i ++)
{
temp[8*i+j] =d[8*j+i];
}
}*/
fDCT2D8x4_and_threshold_32f(temp, d, threshold);
fDCT2D8x4_and_threshold_32f(temp + 4, d + 4, threshold);
}
void vpx_hadamard_8x8_neon(const int16_t *src_diff, int src_stride,
int16_t *coeff) {
int16x8_t a0 = vld1q_s16(src_diff);
int16x8_t a1 = vld1q_s16(src_diff + src_stride);
int16x8_t a2 = vld1q_s16(src_diff + 2 * src_stride);
int16x8_t a3 = vld1q_s16(src_diff + 3 * src_stride);
int16x8_t a4 = vld1q_s16(src_diff + 4 * src_stride);
int16x8_t a5 = vld1q_s16(src_diff + 5 * src_stride);
int16x8_t a6 = vld1q_s16(src_diff + 6 * src_stride);
int16x8_t a7 = vld1q_s16(src_diff + 7 * src_stride);
hadamard8x8_one_pass(&a0, &a1, &a2, &a3, &a4, &a5, &a6, &a7);
transpose8x8(&a0, &a1, &a2, &a3, &a4, &a5, &a6, &a7);
hadamard8x8_one_pass(&a0, &a1, &a2, &a3, &a4, &a5, &a6, &a7);
// Skip the second transpose because it is not required.
vst1q_s16(coeff + 0, a0);
vst1q_s16(coeff + 8, a1);
vst1q_s16(coeff + 16, a2);
vst1q_s16(coeff + 24, a3);
vst1q_s16(coeff + 32, a4);
vst1q_s16(coeff + 40, a5);
vst1q_s16(coeff + 48, a6);
vst1q_s16(coeff + 56, a7);
}
void fDCT8x8_32f_and_threshold_and_iDCT8x8_32f(float* s, float threshold)
{
fDCT2D8x4_32f(s, s);
fDCT2D8x4_32f(s + 4, s + 4);
transpose8x8(s);
#ifdef _KEEP_00_COEF_
fDCT2D8x4_and_threshold_keep00_32f(s, s, threshold);
#else
fDCT2D8x4_and_threshold_32f(s, s,threshold);
#endif
fDCT2D8x4_and_threshold_32f(s + 4, s + 4, threshold);
//ommiting transform
//transpose8x8(s);
//transpose8x8(s);
iDCT2D8x4_32f(s, s);
iDCT2D8x4_32f(s + 4, s + 4);
transpose8x8(s);
iDCT2D8x4_32f(s, s);
iDCT2D8x4_32f(s + 4, s + 4);
return;
}
void iDCT8x8_32f(const float* s, float* d, float* temp)
{
transpose8x8((float*)s, temp);
//for (int j = 0; j < 8; j ++)
//{
// for (int i = 0; i < 8; i ++)
// {
// temp[8*i+j] =s[8*j+i];
// }
//}
iDCT2D8x4_32f(temp, d);
iDCT2D8x4_32f(temp + 4, d + 4);
transpose8x8(d, temp);
/*for (int j = 0; j < 8; j ++)
{
for (int i = 0; i < 8; i ++)
{
temp[8*i+j] =d[8*j+i];
}
}*/
iDCT2D8x4_32f(temp, d);
iDCT2D8x4_32f(temp + 4, d + 4);
}
