INSTANTIATE_TEST_CASE_P(
C, Trans32x32Test,
::testing::Values(
- make_tuple(&vp9_short_fdct32x32_c, &vp9_short_idct32x32_add_c, 0),
- make_tuple(&vp9_short_fdct32x32_rd_c, &vp9_short_idct32x32_add_c, 1)));
+ make_tuple(&vp9_short_fdct32x32_c, &vp9_idct32x32_1024_add_c, 0),
+ make_tuple(&vp9_short_fdct32x32_rd_c, &vp9_idct32x32_1024_add_c, 1)));
#if HAVE_SSE2
INSTANTIATE_TEST_CASE_P(
SSE2, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_short_fdct32x32_sse2,
- &vp9_short_idct32x32_add_sse2, 0),
+ &vp9_idct32x32_1024_add_sse2, 0),
make_tuple(&vp9_short_fdct32x32_rd_sse2,
- &vp9_short_idct32x32_add_sse2, 1)));
+ &vp9_idct32x32_1024_add_sse2, 1)));
#endif
} // namespace
cospi_31_64 EQU 804
- EXPORT |vp9_short_idct32x32_add_neon|
+ EXPORT |vp9_idct32x32_1024_add_neon|
ARM
REQUIRE8
PRESERVE8
MEND
; --------------------------------------------------------------------------
-;void vp9_short_idct32x32_add_neon(int16_t *input, uint8_t *dest, int dest_stride);
+;void vp9_idct32x32_1024_add_neon(int16_t *input, uint8_t *dest, int dest_stride);
;
; r0 int16_t *input,
; r1 uint8_t *dest,
; r9 dest + 15 * dest_stride, descending (14, 13, 12, ...)
; r10 dest + 16 * dest_stride, ascending (17, 18, 19, ...)
-|vp9_short_idct32x32_add_neon| PROC
+|vp9_idct32x32_1024_add_neon| PROC
; This function does one pass of idct32x32 transform.
;
; This is done by transposing the input and then doing a 1d transform on
vpop {d8-d15}
pop {r4-r11}
bx lr
- ENDP ; |vp9_short_idct32x32_add_neon|
+ ENDP ; |vp9_idct32x32_1024_add_neon|
END
output[31] = step1[0] - step1[31];
}
-void vp9_short_idct32x32_add_c(int16_t *input, uint8_t *dest, int dest_stride) {
+void vp9_idct32x32_1024_add_c(int16_t *input, uint8_t *dest, int dest_stride) {
int16_t out[32 * 32];
int16_t *outptr = out;
int i, j;
}
}
-void vp9_short_idct32x32_1_add_c(int16_t *input, uint8_t *dest,
+void vp9_idct32x32_1_add_c(int16_t *input, uint8_t *dest,
int dest_stride) {
int i, j;
int a1;
}
}
-void vp9_idct_add_32x32(int16_t *input, uint8_t *dest, int stride, int eob) {
+void vp9_idct32x32_add(int16_t *input, uint8_t *dest, int stride, int eob) {
if (eob) {
if (eob == 1)
- vp9_short_idct32x32_1_add(input, dest, stride);
+ vp9_idct32x32_1_add(input, dest, stride);
else
- vp9_short_idct32x32_add(input, dest, stride);
+ vp9_idct32x32_1024_add(input, dest, stride);
}
}
void vp9_iwht4x4_add(int16_t *input, uint8_t *dest, int stride, int eob);
void vp9_idct8x8_add(int16_t *input, uint8_t *dest, int stride, int eob);
void vp9_idct16x16_add(int16_t *input, uint8_t *dest, int stride, int eob);
-void vp9_idct_add_32x32(int16_t *input, uint8_t *dest, int stride, int eob);
+void vp9_idct32x32_add(int16_t *input, uint8_t *dest, int stride, int eob);
void vp9_iht_add(TX_TYPE tx_type, int16_t *input, uint8_t *dest,
int stride, int eob);
prototype void vp9_idct16x16_10_add "int16_t *input, uint8_t *dest, int dest_stride"
specialize vp9_idct16x16_10_add sse2 neon
-prototype void vp9_short_idct32x32_add "int16_t *input, uint8_t *dest, int dest_stride"
-specialize vp9_short_idct32x32_add sse2 neon
+prototype void vp9_idct32x32_1024_add "int16_t *input, uint8_t *dest, int dest_stride"
+specialize vp9_idct32x32_1024_add sse2 neon
-prototype void vp9_short_idct32x32_1_add "int16_t *input, uint8_t *dest, int dest_stride"
-specialize vp9_short_idct32x32_1_add sse2
+prototype void vp9_idct32x32_1_add "int16_t *input, uint8_t *dest, int dest_stride"
+specialize vp9_idct32x32_1_add sse2
prototype void vp9_short_iht4x4_add "int16_t *input, uint8_t *dest, int dest_stride, int tx_type"
specialize vp9_short_iht4x4_add sse2 neon
input += 8; \
} \
-void vp9_short_idct32x32_add_sse2(int16_t *input, uint8_t *dest, int stride) {
+void vp9_idct32x32_1024_add_sse2(int16_t *input, uint8_t *dest, int stride) {
const __m128i rounding = _mm_set1_epi32(DCT_CONST_ROUNDING);
const __m128i final_rounding = _mm_set1_epi16(1<<5);
}
} //NOLINT
-void vp9_short_idct32x32_1_add_sse2(int16_t *input, uint8_t *dest, int stride) {
+void vp9_idct32x32_1_add_sse2(int16_t *input, uint8_t *dest, int stride) {
__m128i dc_value;
const __m128i zero = _mm_setzero_si128();
int a, i;
break;
case TX_32X32:
tx_type = DCT_DCT;
- vp9_idct_add_32x32(qcoeff, dst, stride, eob);
+ vp9_idct32x32_add(qcoeff, dst, stride, eob);
break;
default:
assert(!"Invalid transform size");
switch (tx_size) {
case TX_32X32:
- vp9_short_idct32x32_add(dqcoeff, dst, pd->dst.stride);
+ vp9_idct32x32_1024_add(dqcoeff, dst, pd->dst.stride);
break;
case TX_16X16:
vp9_idct16x16_add(dqcoeff, dst, pd->dst.stride, pd->eobs[block]);
p->quant, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, p->zbin_extra, eob, scan, iscan);
if (!x->skip_encode && *eob)
- vp9_short_idct32x32_add(dqcoeff, dst, pd->dst.stride);
+ vp9_idct32x32_1024_add(dqcoeff, dst, pd->dst.stride);
break;
case TX_16X16:
tx_type = get_tx_type_16x16(pd->plane_type, xd);