dany/libwebp
1
0
Fork 0
mirror of https://github.com/webmproject/libwebp.git synced 2024-11-20 12:28:26 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity

enc_sse2: drop SSE2 suffix from local functions

Browse Source 
Change-Id: I5d61605a9d410761d50b689b046114f0ab3ba24e
This commit is contained in:
James Zern 2014-04-02 19:49:16 -07:00
parent d038e6193b
commit 2ca42a4fb7
1 changed files with 46 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

93
src/dsp/enc_sse2.c
View File

@ -52,9 +52,9 @@ static void PrintReg(const __m128i r, const char* const name, int size) {
// Compute susceptibility based on DCT-coeff histograms: // Compute susceptibility based on DCT-coeff histograms:
// the higher, the "easier" the macroblock is to compress. // the higher, the "easier" the macroblock is to compress.
static void CollectHistogramSSE2(const uint8_t* ref, const uint8_t* pred, static void CollectHistogram(const uint8_t* ref, const uint8_t* pred,
int start_block, int end_block, int start_block, int end_block,
VP8Histogram* const histo) { VP8Histogram* const histo) {
const __m128i max_coeff_thresh = _mm_set1_epi16(MAX_COEFF_THRESH); const __m128i max_coeff_thresh = _mm_set1_epi16(MAX_COEFF_THRESH);
int j; int j;
for (j = start_block; j < end_block; ++j) { for (j = start_block; j < end_block; ++j) {
@ -98,8 +98,8 @@ static void CollectHistogramSSE2(const uint8_t* ref, const uint8_t* pred,
// Transforms (Paragraph 14.4) // Transforms (Paragraph 14.4)
// Does one or two inverse transforms. // Does one or two inverse transforms.
static void ITransformSSE2(const uint8_t* ref, const int16_t* in, uint8_t* dst, static void ITransform(const uint8_t* ref, const int16_t* in, uint8_t* dst,
int do_two) { int do_two) {
// This implementation makes use of 16-bit fixed point versions of two // This implementation makes use of 16-bit fixed point versions of two
// multiply constants: // multiply constants:
// K1 = sqrt(2) * cos (pi/8) ~= 85627 / 2^16 // K1 = sqrt(2) * cos (pi/8) ~= 85627 / 2^16
@ -318,8 +318,7 @@ static void ITransformSSE2(const uint8_t* ref, const int16_t* in, uint8_t* dst,
} }
} }
static void FTransformSSE2(const uint8_t* src, const uint8_t* ref, static void FTransform(const uint8_t* src, const uint8_t* ref, int16_t* out) {
int16_t* out) {
const __m128i zero = _mm_setzero_si128(); const __m128i zero = _mm_setzero_si128();
const __m128i seven = _mm_set1_epi16(7); const __m128i seven = _mm_set1_epi16(7);
const __m128i k937 = _mm_set1_epi32(937); const __m128i k937 = _mm_set1_epi32(937);
@ -451,7 +450,7 @@ static void FTransformSSE2(const uint8_t* src, const uint8_t* ref,
} }
} }
static void FTransformWHTSSE2(const int16_t* in, int16_t* out) { static void FTransformWHT(const int16_t* in, int16_t* out) {
int32_t tmp[16]; int32_t tmp[16];
int i; int i;
for (i = 0; i < 4; ++i, in += 64) { for (i = 0; i < 4; ++i, in += 64) {
@ -487,8 +486,8 @@ static void FTransformWHTSSE2(const int16_t* in, int16_t* out) {
//------------------------------------------------------------------------------ //------------------------------------------------------------------------------
// Metric // Metric
static int SSE_Nx4SSE2(const uint8_t* a, const uint8_t* b, static int SSE_Nx4(const uint8_t* a, const uint8_t* b,
int num_quads, int do_16) { int num_quads, int do_16) {
const __m128i zero = _mm_setzero_si128(); const __m128i zero = _mm_setzero_si128();
__m128i sum1 = zero; __m128i sum1 = zero;
__m128i sum2 = zero; __m128i sum2 = zero;
@ -565,19 +564,19 @@ static int SSE_Nx4SSE2(const uint8_t* a, const uint8_t* b,
} }
} }
static int SSE16x16SSE2(const uint8_t* a, const uint8_t* b) { static int SSE16x16(const uint8_t* a, const uint8_t* b) {
return SSE_Nx4SSE2(a, b, 4, 1); return SSE_Nx4(a, b, 4, 1);
} }
static int SSE16x8SSE2(const uint8_t* a, const uint8_t* b) { static int SSE16x8(const uint8_t* a, const uint8_t* b) {
return SSE_Nx4SSE2(a, b, 2, 1); return SSE_Nx4(a, b, 2, 1);
} }
static int SSE8x8SSE2(const uint8_t* a, const uint8_t* b) { static int SSE8x8(const uint8_t* a, const uint8_t* b) {
return SSE_Nx4SSE2(a, b, 2, 0); return SSE_Nx4(a, b, 2, 0);
} }
static int SSE4x4SSE2(const uint8_t* a, const uint8_t* b) { static int SSE4x4(const uint8_t* a, const uint8_t* b) {
const __m128i zero = _mm_setzero_si128(); const __m128i zero = _mm_setzero_si128();
// Load values. Note that we read 8 pixels instead of 4, // Load values. Note that we read 8 pixels instead of 4,
@ -634,8 +633,8 @@ static int SSE4x4SSE2(const uint8_t* a, const uint8_t* b) {
// Hadamard transform // Hadamard transform
// Returns the difference between the weighted sum of the absolute value of // Returns the difference between the weighted sum of the absolute value of
// transformed coefficients. // transformed coefficients.
static int TTransformSSE2(const uint8_t* inA, const uint8_t* inB, static int TTransform(const uint8_t* inA, const uint8_t* inB,
const uint16_t* const w) { const uint16_t* const w) {
int32_t sum[4]; int32_t sum[4];
__m128i tmp_0, tmp_1, tmp_2, tmp_3; __m128i tmp_0, tmp_1, tmp_2, tmp_3;
const __m128i zero = _mm_setzero_si128(); const __m128i zero = _mm_setzero_si128();
@ -782,19 +781,19 @@ static int TTransformSSE2(const uint8_t* inA, const uint8_t* inB,
return sum[0] + sum[1] + sum[2] + sum[3]; return sum[0] + sum[1] + sum[2] + sum[3];
} }
static int Disto4x4SSE2(const uint8_t* const a, const uint8_t* const b, static int Disto4x4(const uint8_t* const a, const uint8_t* const b,
const uint16_t* const w) { const uint16_t* const w) {
const int diff_sum = TTransformSSE2(a, b, w); const int diff_sum = TTransform(a, b, w);
return abs(diff_sum) >> 5; return abs(diff_sum) >> 5;
} }
static int Disto16x16SSE2(const uint8_t* const a, const uint8_t* const b, static int Disto16x16(const uint8_t* const a, const uint8_t* const b,
const uint16_t* const w) { const uint16_t* const w) {
int D = 0; int D = 0;
int x, y; int x, y;
for (y = 0; y < 16 * BPS; y += 4 * BPS) { for (y = 0; y < 16 * BPS; y += 4 * BPS) {
for (x = 0; x < 16; x += 4) { for (x = 0; x < 16; x += 4) {
D += Disto4x4SSE2(a + x + y, b + x + y, w); D += Disto4x4(a + x + y, b + x + y, w);
} }
} }
return D; return D;
@ -805,10 +804,10 @@ static int Disto16x16SSE2(const uint8_t* const a, const uint8_t* const b,
// //
#define QFIX2 0 #define QFIX2 0
static WEBP_INLINE int QuantizeBlock(int16_t in[16], int16_t out[16], static WEBP_INLINE int DoQuantizeBlock(int16_t in[16], int16_t out[16],
int n, int shift, int n, int shift,
const uint16_t* const sharpen, const uint16_t* const sharpen,
const VP8Matrix* const mtx) { const VP8Matrix* const mtx) {
const __m128i max_coeff_2047 = _mm_set1_epi16(MAX_LEVEL); const __m128i max_coeff_2047 = _mm_set1_epi16(MAX_LEVEL);
const __m128i zero = _mm_setzero_si128(); const __m128i zero = _mm_setzero_si128();
__m128i coeff0, coeff8; __m128i coeff0, coeff8;
@ -921,14 +920,14 @@ static WEBP_INLINE int QuantizeBlock(int16_t in[16], int16_t out[16],
return (_mm_movemask_epi8(_mm_cmpeq_epi8(packed_out, zero)) != 0xffff); return (_mm_movemask_epi8(_mm_cmpeq_epi8(packed_out, zero)) != 0xffff);
} }
static int QuantizeBlockSSE2(int16_t in[16], int16_t out[16], static int QuantizeBlock(int16_t in[16], int16_t out[16],
int n, const VP8Matrix* const mtx) { int n, const VP8Matrix* const mtx) {
return QuantizeBlock(in, out, n, 0, &mtx->sharpen_[0], mtx); return DoQuantizeBlock(in, out, n, 0, &mtx->sharpen_[0], mtx);
} }
static int QuantizeBlockWHTSSE2(int16_t in[16], int16_t out[16], static int QuantizeBlockWHT(int16_t in[16], int16_t out[16],
const VP8Matrix* const mtx) { const VP8Matrix* const mtx) {
return QuantizeBlock(in, out, 0, 0, &mtx->sharpen_[0], mtx); return DoQuantizeBlock(in, out, 0, 0, &mtx->sharpen_[0], mtx);
} }
#endif // WEBP_USE_SSE2 #endif // WEBP_USE_SSE2
@ -940,18 +939,18 @@ extern void VP8EncDspInitSSE2(void);
void VP8EncDspInitSSE2(void) { void VP8EncDspInitSSE2(void) {
#if defined(WEBP_USE_SSE2) #if defined(WEBP_USE_SSE2)
VP8CollectHistogram = CollectHistogramSSE2; VP8CollectHistogram = CollectHistogram;
VP8EncQuantizeBlock = QuantizeBlockSSE2; VP8EncQuantizeBlock = QuantizeBlock;
VP8EncQuantizeBlockWHT = QuantizeBlockWHTSSE2; VP8EncQuantizeBlockWHT = QuantizeBlockWHT;
VP8ITransform = ITransformSSE2; VP8ITransform = ITransform;
VP8FTransform = FTransformSSE2; VP8FTransform = FTransform;
VP8FTransformWHT = FTransformWHTSSE2; VP8FTransformWHT = FTransformWHT;
VP8SSE16x16 = SSE16x16SSE2; VP8SSE16x16 = SSE16x16;
VP8SSE16x8 = SSE16x8SSE2; VP8SSE16x8 = SSE16x8;
VP8SSE8x8 = SSE8x8SSE2; VP8SSE8x8 = SSE8x8;
VP8SSE4x4 = SSE4x4SSE2; VP8SSE4x4 = SSE4x4;
VP8TDisto4x4 = Disto4x4SSE2; VP8TDisto4x4 = Disto4x4;
VP8TDisto16x16 = Disto16x16SSE2; VP8TDisto16x16 = Disto16x16;
#endif // WEBP_USE_SSE2 #endif // WEBP_USE_SSE2
} }