42 |
* * |
* * |
43 |
* Revision history: * |
* Revision history: * |
44 |
* * |
* * |
45 |
|
* 29.03.2002 interlacing speedup - used transfer strides instead of |
46 |
|
* manual field-to-frame conversion |
47 |
* 26.03.2002 interlacing support - moved transfers outside loops |
* 26.03.2002 interlacing support - moved transfers outside loops |
48 |
* 22.12.2001 get_dc_scaler() moved to common.h |
* 22.12.2001 get_dc_scaler() moved to common.h |
49 |
* 19.11.2001 introduced coefficient thresholding (Isibaar) * |
* 19.11.2001 introduced coefficient thresholding (Isibaar) * |
82 |
|
|
83 |
{ |
{ |
84 |
|
|
85 |
const uint32_t stride = pParam->edged_width; |
uint32_t stride = pParam->edged_width; |
86 |
|
uint32_t stride2 = stride / 2; |
87 |
|
uint32_t next_block = stride * 8; |
88 |
uint32_t i; |
uint32_t i; |
89 |
uint32_t iQuant = pParam->quant; |
uint32_t iQuant = pParam->quant; |
90 |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
91 |
|
|
92 |
pY_Cur = pCurrent->y + (y_pos << 4) * stride + (x_pos << 4); |
pY_Cur = pCurrent->y + (y_pos << 4) * stride + (x_pos << 4); |
93 |
pU_Cur = pCurrent->u + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pU_Cur = pCurrent->u + (y_pos << 3) * stride2 + (x_pos << 3); |
94 |
pV_Cur = pCurrent->v + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pV_Cur = pCurrent->v + (y_pos << 3) * stride2 + (x_pos << 3); |
95 |
|
|
96 |
start_timer(); |
start_timer(); |
97 |
transfer_8to16copy(&data[0*64], pY_Cur, stride); |
transfer_8to16copy(&data[0*64], pY_Cur, stride); |
98 |
transfer_8to16copy(&data[1*64], pY_Cur + 8, stride); |
transfer_8to16copy(&data[1*64], pY_Cur + 8, stride); |
99 |
transfer_8to16copy(&data[2*64], pY_Cur + 8 * stride, stride); |
transfer_8to16copy(&data[2*64], pY_Cur + next_block, stride); |
100 |
transfer_8to16copy(&data[3*64], pY_Cur + 8 * stride + 8, stride); |
transfer_8to16copy(&data[3*64], pY_Cur + next_block + 8,stride); |
101 |
transfer_8to16copy(&data[4*64], pU_Cur, stride / 2); |
transfer_8to16copy(&data[4*64], pU_Cur, stride2); |
102 |
transfer_8to16copy(&data[5*64], pV_Cur, stride / 2); |
transfer_8to16copy(&data[5*64], pV_Cur, stride2); |
103 |
stop_transfer_timer(); |
stop_transfer_timer(); |
104 |
|
|
105 |
start_timer(); |
start_timer(); |
144 |
stop_idct_timer(); |
stop_idct_timer(); |
145 |
} |
} |
146 |
|
|
|
start_timer(); |
|
147 |
if (pMB->field_dct) |
if (pMB->field_dct) |
148 |
{ |
{ |
149 |
MBFieldToFrame(data); |
next_block = stride; |
150 |
|
stride *= 2; |
151 |
} |
} |
|
stop_interlacing_timer(); |
|
152 |
|
|
153 |
start_timer(); |
start_timer(); |
154 |
transfer_16to8copy(pY_Cur, &data[0*64], stride); |
transfer_16to8copy(pY_Cur, &data[0*64], stride); |
155 |
transfer_16to8copy(pY_Cur + 8, &data[1*64], stride); |
transfer_16to8copy(pY_Cur + 8, &data[1*64], stride); |
156 |
transfer_16to8copy(pY_Cur + 8 * stride, &data[2*64], stride); |
transfer_16to8copy(pY_Cur + next_block, &data[2*64], stride); |
157 |
transfer_16to8copy(pY_Cur + 8 + 8 * stride, &data[3*64], stride); |
transfer_16to8copy(pY_Cur + next_block + 8, &data[3*64], stride); |
158 |
transfer_16to8copy(pU_Cur, &data[4*64], stride / 2); |
transfer_16to8copy(pU_Cur, &data[4*64], stride2); |
159 |
transfer_16to8copy(pV_Cur, &data[5*64], stride / 2); |
transfer_16to8copy(pV_Cur, &data[5*64], stride2); |
160 |
stop_transfer_timer(); |
stop_transfer_timer(); |
161 |
|
|
162 |
} |
} |
171 |
|
|
172 |
{ |
{ |
173 |
|
|
174 |
const uint32_t stride = pParam->edged_width; |
uint32_t stride = pParam->edged_width; |
175 |
|
uint32_t stride2 = stride / 2; |
176 |
|
uint32_t next_block = stride * 8; |
177 |
uint32_t i; |
uint32_t i; |
178 |
uint32_t iQuant = pParam->quant; |
uint32_t iQuant = pParam->quant; |
179 |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
181 |
uint32_t sum; |
uint32_t sum; |
182 |
|
|
183 |
pY_Cur = pCurrent->y + (y_pos << 4) * stride + (x_pos << 4); |
pY_Cur = pCurrent->y + (y_pos << 4) * stride + (x_pos << 4); |
184 |
pU_Cur = pCurrent->u + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pU_Cur = pCurrent->u + (y_pos << 3) * stride2 + (x_pos << 3); |
185 |
pV_Cur = pCurrent->v + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pV_Cur = pCurrent->v + (y_pos << 3) * stride2 + (x_pos << 3); |
186 |
|
|
187 |
start_timer(); |
start_timer(); |
188 |
pMB->field_dct = 0; |
pMB->field_dct = 0; |
238 |
} |
} |
239 |
} |
} |
240 |
|
|
|
start_timer(); |
|
241 |
if (pMB->field_dct) |
if (pMB->field_dct) |
242 |
{ |
{ |
243 |
MBFieldToFrame(data); |
next_block = stride; |
244 |
|
stride *= 2; |
245 |
} |
} |
|
stop_interlacing_timer(); |
|
246 |
|
|
247 |
start_timer(); |
start_timer(); |
248 |
if (cbp & 32) |
if (cbp & 32) |
250 |
if (cbp & 16) |
if (cbp & 16) |
251 |
transfer_16to8add(pY_Cur + 8, &data[1*64], stride); |
transfer_16to8add(pY_Cur + 8, &data[1*64], stride); |
252 |
if (cbp & 8) |
if (cbp & 8) |
253 |
transfer_16to8add(pY_Cur + 8 * stride, &data[2*64], stride); |
transfer_16to8add(pY_Cur + next_block, &data[2*64], stride); |
254 |
if (cbp & 4) |
if (cbp & 4) |
255 |
transfer_16to8add(pY_Cur + 8 + 8 * stride, &data[3*64], stride); |
transfer_16to8add(pY_Cur + next_block + 8, &data[3*64], stride); |
256 |
if (cbp & 2) |
if (cbp & 2) |
257 |
transfer_16to8add(pU_Cur, &data[4*64], stride / 2); |
transfer_16to8add(pU_Cur, &data[4*64], stride2); |
258 |
if (cbp & 1) |
if (cbp & 1) |
259 |
transfer_16to8add(pV_Cur, &data[5*64], stride / 2); |
transfer_16to8add(pV_Cur, &data[5*64], stride2); |
260 |
stop_transfer_timer(); |
stop_transfer_timer(); |
261 |
|
|
262 |
return cbp; |
return cbp; |
371 |
MOVLINE(LINE(3,5), LINE(3,3)); |
MOVLINE(LINE(3,5), LINE(3,3)); |
372 |
MOVLINE(LINE(3,3), tmp); |
MOVLINE(LINE(3,3), tmp); |
373 |
} |
} |
|
|
|
|
|
|
|
/* interlace Y blocks vertically */ |
|
|
|
|
|
void MBFieldToFrame(int16_t data[6*64]) |
|
|
{ |
|
|
uint16_t tmp[8]; |
|
|
|
|
|
/* left blocks */ |
|
|
|
|
|
// 1=8, 8=4, 4=2, 2=1 |
|
|
MOVLINE(tmp, LINE(0,1)); |
|
|
MOVLINE(LINE(0,1), LINE(2,0)); |
|
|
MOVLINE(LINE(2,0), LINE(0,4)); |
|
|
MOVLINE(LINE(0,4), LINE(0,2)); |
|
|
MOVLINE(LINE(0,2), tmp); |
|
|
|
|
|
// 3=9, 9=12, 12=6, 6=3 |
|
|
MOVLINE(tmp, LINE(0,3)); |
|
|
MOVLINE(LINE(0,3), LINE(2,1)); |
|
|
MOVLINE(LINE(2,1), LINE(2,4)); |
|
|
MOVLINE(LINE(2,4), LINE(0,6)); |
|
|
MOVLINE(LINE(0,6), tmp); |
|
|
|
|
|
// 5=10, 10=5 |
|
|
MOVLINE(tmp, LINE(0,5)); |
|
|
MOVLINE(LINE(0,5), LINE(2,2)); |
|
|
MOVLINE(LINE(2,2), tmp); |
|
|
|
|
|
// 7=11, 11=13, 13=14, 14=7 |
|
|
MOVLINE(tmp, LINE(0,7)); |
|
|
MOVLINE(LINE(0,7), LINE(2,3)); |
|
|
MOVLINE(LINE(2,3), LINE(2,5)); |
|
|
MOVLINE(LINE(2,5), LINE(2,6)); |
|
|
MOVLINE(LINE(2,6), tmp); |
|
|
|
|
|
/* right blocks */ |
|
|
|
|
|
// 1=8, 8=4, 4=2, 2=1 |
|
|
MOVLINE(tmp, LINE(1,1)); |
|
|
MOVLINE(LINE(1,1), LINE(3,0)); |
|
|
MOVLINE(LINE(3,0), LINE(1,4)); |
|
|
MOVLINE(LINE(1,4), LINE(1,2)); |
|
|
MOVLINE(LINE(1,2), tmp); |
|
|
|
|
|
// 3=9, 9=12, 12=6, 6=3 |
|
|
MOVLINE(tmp, LINE(1,3)); |
|
|
MOVLINE(LINE(1,3), LINE(3,1)); |
|
|
MOVLINE(LINE(3,1), LINE(3,4)); |
|
|
MOVLINE(LINE(3,4), LINE(1,6)); |
|
|
MOVLINE(LINE(1,6), tmp); |
|
|
|
|
|
// 5=10, 10=5 |
|
|
MOVLINE(tmp, LINE(1,5)); |
|
|
MOVLINE(LINE(1,5), LINE(3,2)); |
|
|
MOVLINE(LINE(3,2), tmp); |
|
|
|
|
|
// 7=11, 11=13, 13=14, 14=7 |
|
|
MOVLINE(tmp, LINE(1,7)); |
|
|
MOVLINE(LINE(1,7), LINE(3,3)); |
|
|
MOVLINE(LINE(3,3), LINE(3,5)); |
|
|
MOVLINE(LINE(3,5), LINE(3,6)); |
|
|
MOVLINE(LINE(3,6), tmp); |
|
|
} |
|