mewin/libpng
1
0
Fork 0
mirror of https://git.code.sf.net/p/libpng/code.git synced 2025-07-10 18:04:09 +02:00
Code Issues Projects Releases Wiki Activity

[master] Imported from libpng-1.6.31.tar

Browse Source
This commit is contained in:
Glenn Randers-Pehrson
2017-07-27 05:25:33 -05:00
parent 8c552ddf5b
commit fbff8986e7
53 changed files with 1014 additions and 233 deletions
Download Patch File Download Diff File Expand all files Collapse all files

110
intel/filter_sse2_intrinsics.c
View File

@@ -5,7 +5,7 @@
* Written by Mike Klein and Matt Sarett
* Derived from arm/filter_neon_intrinsics.c
*
* Last changed in libpng 1.6.29 [March 16, 2017]
* Last changed in libpng 1.6.31 [July 27, 2017]
*
* This code is released under the libpng license.
* For conditions of distribution and use, see the disclaimer
@@ -40,7 +40,7 @@ static __m128i load3(const void* p) {
/* We'll load 2 bytes, then 1 byte,
* then mask them together, and finally load into SSE.
*/
const png_uint_16* p01 = p;
const png_uint_16* p01 = (png_const_uint_16p)p;
const png_byte* p2 = (const png_byte*)(p01+1);
png_uint_32 v012 = (png_uint_32)(*p01)
@@ -53,12 +53,15 @@ static void store3(void* p, __m128i v) {
* its bottom two bytes, then its third byte.
*/
png_uint_32 v012;
png_uint_16* p01;
png_byte* p2;
store4(&v012, v);
png_uint_16* p01 = p;
png_byte* p2 = (png_byte*)(p01+1);
*p01 = v012;
*p2 = v012 >> 16;
p01 = (png_uint_16p)p;
p2 = (png_byte*)(p01+1);
*p01 = (png_uint_16)v012;
*p2 = (png_byte)(v012 >> 16);
}
void png_read_filter_row_sub3_sse2(png_row_infop row_info, png_bytep row,
@@ -68,10 +71,13 @@ void png_read_filter_row_sub3_sse2(png_row_infop row_info, png_bytep row,
* There is no pixel to the left of the first pixel. It's encoded directly.
* That works with our main loop if we just say that left pixel was zero.
*/
png_debug(1, "in png_read_filter_row_sub3_sse2");
png_size_t rb;
__m128i a, d = _mm_setzero_si128();
int rb = row_info->rowbytes;
png_debug(1, "in png_read_filter_row_sub3_sse2");
rb = row_info->rowbytes;
while (rb >= 4) {
a = d; d = load4(row);
d = _mm_add_epi8(d, a);
@@ -88,6 +94,7 @@ void png_read_filter_row_sub3_sse2(png_row_infop row_info, png_bytep row,
row += 3;
rb -= 3;
}
PNG_UNUSED(prev)
}
void png_read_filter_row_sub4_sse2(png_row_infop row_info, png_bytep row,
@@ -97,11 +104,14 @@ void png_read_filter_row_sub4_sse2(png_row_infop row_info, png_bytep row,
* There is no pixel to the left of the first pixel. It's encoded directly.
* That works with our main loop if we just say that left pixel was zero.
*/
png_debug(1, "in png_read_filter_row_sub4_sse2");
png_size_t rb;
__m128i a, d = _mm_setzero_si128();
int rb = row_info->rowbytes;
while (rb > 0) {
png_debug(1, "in png_read_filter_row_sub4_sse2");
rb = row_info->rowbytes+4;
while (rb > 4) {
a = d; d = load4(row);
d = _mm_add_epi8(d, a);
store4(row, d);
@@ -109,6 +119,7 @@ void png_read_filter_row_sub4_sse2(png_row_infop row_info, png_bytep row,
row += 4;
rb -= 4;
}
PNG_UNUSED(prev)
}
void png_read_filter_row_avg3_sse2(png_row_infop row_info, png_bytep row,
@@ -119,18 +130,23 @@ void png_read_filter_row_avg3_sse2(png_row_infop row_info, png_bytep row,
* predicted to be half of the pixel above it. So again, this works
* perfectly with our loop if we make sure a starts at zero.
*/
png_debug(1, "in png_read_filter_row_avg3_sse2");
png_size_t rb;
const __m128i zero = _mm_setzero_si128();
__m128i b;
__m128i a, d = zero;
int rb = row_info->rowbytes;
png_debug(1, "in png_read_filter_row_avg3_sse2");
rb = row_info->rowbytes;
while (rb >= 4) {
__m128i avg;
b = load4(prev);
a = d; d = load4(row );
/* PNG requires a truncating average, so we can't just use _mm_avg_epu8 */
__m128i avg = _mm_avg_epu8(a,b);
avg = _mm_avg_epu8(a,b);
/* ...but we can fix it up by subtracting off 1 if it rounded up. */
avg = _mm_sub_epi8(avg, _mm_and_si128(_mm_xor_si128(a,b),
_mm_set1_epi8(1)));
@@ -142,11 +158,12 @@ void png_read_filter_row_avg3_sse2(png_row_infop row_info, png_bytep row,
rb -= 3;
}
if (rb > 0) {
__m128i avg;
b = load3(prev);
a = d; d = load3(row );
/* PNG requires a truncating average, so we can't just use _mm_avg_epu8 */
__m128i avg = _mm_avg_epu8(a,b);
avg = _mm_avg_epu8(a,b);
/* ...but we can fix it up by subtracting off 1 if it rounded up. */
avg = _mm_sub_epi8(avg, _mm_and_si128(_mm_xor_si128(a,b),
_mm_set1_epi8(1)));
@@ -168,18 +185,21 @@ void png_read_filter_row_avg4_sse2(png_row_infop row_info, png_bytep row,
* predicted to be half of the pixel above it. So again, this works
* perfectly with our loop if we make sure a starts at zero.
*/
png_debug(1, "in png_read_filter_row_avg4_sse2");
png_size_t rb;
const __m128i zero = _mm_setzero_si128();
__m128i b;
__m128i a, d = zero;
int rb = row_info->rowbytes;
while (rb > 0) {
png_debug(1, "in png_read_filter_row_avg4_sse2");
rb = row_info->rowbytes+4;
while (rb > 4) {
__m128i avg;
b = load4(prev);
a = d; d = load4(row );
/* PNG requires a truncating average, so we can't just use _mm_avg_epu8 */
__m128i avg = _mm_avg_epu8(a,b);
avg = _mm_avg_epu8(a,b);
/* ...but we can fix it up by subtracting off 1 if it rounded up. */
avg = _mm_sub_epi8(avg, _mm_and_si128(_mm_xor_si128(a,b),
_mm_set1_epi8(1)));
@@ -237,38 +257,42 @@ void png_read_filter_row_paeth3_sse2(png_row_infop row_info, png_bytep row,
* Here we zero b and d, which become c and a respectively at the start of
* the loop.
*/
png_debug(1, "in png_read_filter_row_paeth3_sse2");
png_size_t rb;
const __m128i zero = _mm_setzero_si128();
__m128i c, b = zero,
a, d = zero;
int rb = row_info->rowbytes;
png_debug(1, "in png_read_filter_row_paeth3_sse2");
rb = row_info->rowbytes;
while (rb >= 4) {
/* It's easiest to do this math (particularly, deal with pc) with 16-bit
* intermediates.
*/
__m128i pa,pb,pc,smallest,nearest;
c = b; b = _mm_unpacklo_epi8(load4(prev), zero);
a = d; d = _mm_unpacklo_epi8(load4(row ), zero);
/* (p-a) == (a+b-c - a) == (b-c) */
__m128i pa = _mm_sub_epi16(b,c);
pa = _mm_sub_epi16(b,c);
/* (p-b) == (a+b-c - b) == (a-c) */
__m128i pb = _mm_sub_epi16(a,c);
pb = _mm_sub_epi16(a,c);
/* (p-c) == (a+b-c - c) == (a+b-c-c) == (b-c)+(a-c) */
__m128i pc = _mm_add_epi16(pa,pb);
pc = _mm_add_epi16(pa,pb);
pa = abs_i16(pa); /* |p-a| */
pb = abs_i16(pb); /* |p-b| */
pc = abs_i16(pc); /* |p-c| */
__m128i smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb));
smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb));
/* Paeth breaks ties favoring a over b over c. */
__m128i nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a,
if_then_else(_mm_cmpeq_epi16(smallest, pb), b,
c));
nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a,
if_then_else(_mm_cmpeq_epi16(smallest, pb), b,
c));
/* Note `_epi8`: we need addition to wrap modulo 255. */
d = _mm_add_epi8(d, nearest);
@@ -282,26 +306,27 @@ void png_read_filter_row_paeth3_sse2(png_row_infop row_info, png_bytep row,
/* It's easiest to do this math (particularly, deal with pc) with 16-bit
* intermediates.
*/
__m128i pa,pb,pc,smallest,nearest;
c = b; b = _mm_unpacklo_epi8(load3(prev), zero);
a = d; d = _mm_unpacklo_epi8(load3(row ), zero);
/* (p-a) == (a+b-c - a) == (b-c) */
__m128i pa = _mm_sub_epi16(b,c);
pa = _mm_sub_epi16(b,c);
/* (p-b) == (a+b-c - b) == (a-c) */
__m128i pb = _mm_sub_epi16(a,c);
pb = _mm_sub_epi16(a,c);
/* (p-c) == (a+b-c - c) == (a+b-c-c) == (b-c)+(a-c) */
__m128i pc = _mm_add_epi16(pa,pb);
pc = _mm_add_epi16(pa,pb);
pa = abs_i16(pa); /* |p-a| */
pb = abs_i16(pb); /* |p-b| */
pc = abs_i16(pc); /* |p-c| */
__m128i smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb));
smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb));
/* Paeth breaks ties favoring a over b over c. */
__m128i nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a,
nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a,
if_then_else(_mm_cmpeq_epi16(smallest, pb), b,
c));
@@ -331,13 +356,16 @@ void png_read_filter_row_paeth4_sse2(png_row_infop row_info, png_bytep row,
* Here we zero b and d, which become c and a respectively at the start of
* the loop.
*/
png_debug(1, "in png_read_filter_row_paeth4_sse2");
png_size_t rb;
const __m128i zero = _mm_setzero_si128();
__m128i pa,pb,pc,smallest,nearest;
__m128i c, b = zero,
a, d = zero;
int rb = row_info->rowbytes;
while (rb > 0) {
png_debug(1, "in png_read_filter_row_paeth4_sse2");
rb = row_info->rowbytes+4;
while (rb > 4) {
/* It's easiest to do this math (particularly, deal with pc) with 16-bit
* intermediates.
*/
@@ -345,22 +373,22 @@ void png_read_filter_row_paeth4_sse2(png_row_infop row_info, png_bytep row,
a = d; d = _mm_unpacklo_epi8(load4(row ), zero);
/* (p-a) == (a+b-c - a) == (b-c) */
__m128i pa = _mm_sub_epi16(b,c);
pa = _mm_sub_epi16(b,c);
/* (p-b) == (a+b-c - b) == (a-c) */
__m128i pb = _mm_sub_epi16(a,c);
pb = _mm_sub_epi16(a,c);
/* (p-c) == (a+b-c - c) == (a+b-c-c) == (b-c)+(a-c) */
__m128i pc = _mm_add_epi16(pa,pb);
pc = _mm_add_epi16(pa,pb);
pa = abs_i16(pa); /* |p-a| */
pb = abs_i16(pb); /* |p-b| */
pc = abs_i16(pc); /* |p-c| */
__m128i smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb));
smallest = _mm_min_epi16(pc, _mm_min_epi16(pa, pb));
/* Paeth breaks ties favoring a over b over c. */
__m128i nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a,
nearest = if_then_else(_mm_cmpeq_epi16(smallest, pa), a,
if_then_else(_mm_cmpeq_epi16(smallest, pb), b,
c));