3outeille

⚠️ ⚠️ This Gist Has Been Moved to This Repo ⚠️ ⚠️

Table of Contents

• Features of OOP
• Smart Pointers
• Name Mangling and Externs
• Virtual Functions
  • Pure Virtual Functions
  • Virtual Destructors

• Virtual Table

3outeille

 /* Loop of 16x16 blocks */
for (y = 0; y < heightComp[compCtr]; y += 16)
{
    ...
    for (x = 0; x < widthComp[compCtr]; x += 16)
    {
        ...
        for (blkId = 0; blkId < 4; blkId++)
        {
            yOffset8x8 = (blkId >> 1) * 8;

3outeille

int16_t fg_compute_block_avg(int16_t *dstSampleBlk8, uint32_t widthComp, uint16_t *pNumSamples,
                      uint8_t ySize, uint8_t xSize, uint8_t bitDepth)
{
  uint32_t blockAvg   = 0;
  uint16_t numSamples = 0;
  uint8_t  k, l;

  for (k = 0; k < ySize; k++)
  {
    for (l = 0; l < xSize; l++)

3outeille

int16_t fg_compute_block_avg_sse4(int16_t *dstSampleBlk8, uint32_t widthComp, uint16_t *pNumSamples,
                      uint8_t ySize, uint8_t xSize, uint8_t bitDepth)
{
  uint16_t blockAvg   = 0;
  uint16_t numSamples = 0;

  __m128i acc = _mm_setzero_si128();

  for (int i = 0; i < ySize; i+=1, numSamples+=8)
  {

3outeille

void fg_blend_stripe_sse4(int16_t *dstSampleOffsetY, int16_t *srcSampleOffsetY, int32_t *grainStripe, uint32_t widthComp, uint32_t blockHeight, uint8_t bitDepth)
{
  uint32_t  k, l;

  // Prepare SIMD SSE4 ov_clip_uintp2
  __m128i mask = _mm_set1_epi32((1 << bitDepth));
  __m128i not_mask = _mm_xor_si128(mask, mask);
  not_mask  = _mm_sub_epi32(not_mask, mask);
  mask = _mm_sub_epi32(mask, _mm_set1_epi32(1));

3outeille

%define ARCH_X86_64 1
%define private_prefix asm

%include "x86inc.asm"

    section .rodata align=16
zeros_vec:  dd  0, 0, 0, 0
ones_vec:   dd  1, 1, 1, 1
neg_ones_vec:   dd  -1, -1, -1, -1

3outeille

static inline uint32_t ov_clip_uintp2(int32_t val, uint32_t a)
{
    if (val > 0) {
        int32_t mask  = (1 << a) - 1;
        int32_t overflow = !!(val & (~mask));
        return ((-overflow) & mask) | (val & mask);
    } else {
        return 0;
    }
    #if 0

3outeille

void fg_simulate_grain_blk8x8_sse4(int32_t *grainStripe, uint32_t grainStripeOffsetBlk8,
  uint32_t width, uint8_t log2ScaleFactor, int16_t scaleFactor, uint32_t kOffset, uint32_t lOffset, uint8_t h, uint8_t v, uint32_t xSize)
{
    uint32_t idx_offset_l1, idx_offset_l2, idx_offset_l3, idx_offset_l4;
    uint32_t grainStripeOffsetBlk8_l1, grainStripeOffsetBlk8_l2, grainStripeOffsetBlk8_l3, grainStripeOffsetBlk8_l4;

    uint32_t idx_offset = ( h*NUM_CUT_OFF_FREQ +  v ) * DATA_BASE_SIZE  * DATA_BASE_SIZE;

    __m128i scale = _mm_set_epi32(scaleFactor, scaleFactor, scaleFactor, scaleFactor);

3outeille

void fg_simulate_grain_blk8x8(int32_t *grainStripe, uint32_t grainStripeOffsetBlk8,
  uint32_t width, uint8_t log2ScaleFactor, int16_t scaleFactor, uint32_t kOffset, uint32_t lOffset, uint8_t h, uint8_t v, uint32_t xSize)
{
  uint32_t k, l;
  uint32_t idx, idx_offset, idx_offset_l, grainStripeOffsetBlk8_l;

  idx_offset = ( h*NUM_CUT_OFF_FREQ +  v ) * DATA_BASE_SIZE  * DATA_BASE_SIZE;
  for (l = 0; l < 8; l++) /* y direction */
  {
    idx_offset_l            = idx_offset + (l + lOffset) * DATA_BASE_SIZE;

3outeille

struct BenchWrapper
{
    size_t nb_fct_call;
    size_t iter_per_fct;
    // Define function arguments here
    int (*func)(const OVSEI *sei, OVFrame **frame);
};

void bench_decorator(struct BenchWrapper bw, char* name, const OVSEI *sei, OVFrame **frame)
{