generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge "Giving consistent names to IDCT 32x32 functions."

Browse Source
This commit is contained in:
Dmitry Kovalev 2013-10-10 15:31:25 -07:00 committed by Gerrit Code Review
commit 2be3b84aed
8 changed files with 23 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
test/dct32x32_test.cc
View File

@ -247,16 +247,16 @@ using std::tr1::make_tuple;
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

8
vp9/common/arm/neon/vp9_short_idct32x32_add_neon.asm
View File

@ -43,7 +43,7 @@ cospi_30_64 EQU 1606
cospi_31_64 EQU 804
EXPORT |vp9_short_idct32x32_add_neon|
EXPORT |vp9_idct32x32_1024_add_neon|
ARM
REQUIRE8
PRESERVE8
@ -288,7 +288,7 @@ cospi_31_64 EQU 804
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,
@ -303,7 +303,7 @@ cospi_31_64 EQU 804
; 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
@ -1295,5 +1295,5 @@ idct32_bands_end_2nd_pass
vpop {d8-d15}
pop {r4-r11}
bx lr
ENDP ; |vp9_short_idct32x32_add_neon|
ENDP ; |vp9_idct32x32_1024_add_neon|
END

10
vp9/common/vp9_idct.c
View File

@ -1245,7 +1245,7 @@ static void idct32_1d(int16_t *input, int16_t *output) {
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;
@ -1282,7 +1282,7 @@ void vp9_short_idct32x32_add_c(int16_t *input, uint8_t *dest, int dest_stride) {
}
}
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;
@ -1347,12 +1347,12 @@ 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) {
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);
}
}

2
vp9/common/vp9_idct.h
View File

@ -92,7 +92,7 @@ void vp9_idct4x4_add(int16_t *input, uint8_t *dest, int stride, int eob);
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);

8
vp9/common/vp9_rtcd_defs.sh
View File

@ -291,11 +291,11 @@ specialize vp9_idct16x16_256_add sse2 neon
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

4
vp9/common/x86/vp9_idct_intrin_sse2.c
View File

@ -2819,7 +2819,7 @@ void vp9_idct16x16_10_add_sse2(int16_t *input, uint8_t *dest,
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);
@ -3550,7 +3550,7 @@ void vp9_short_idct32x32_add_sse2(int16_t *input, uint8_t *dest, int stride) {
}
} //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;

2
vp9/decoder/vp9_decodframe.c
View File

@ -113,7 +113,7 @@ static void decode_block(int plane, int block, BLOCK_SIZE plane_bsize,
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");

4
vp9/encoder/vp9_encodemb.c
View File

@ -442,7 +442,7 @@ static void encode_block(int plane, int block, BLOCK_SIZE plane_bsize,
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]);
@ -539,7 +539,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
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);