/** VBX Motion Estimation, using vbx_3d ops.
* vbw_mtx_motest_3D_byte_setup should be run prior to running this function.
* Using bytes as input data. block_height must be an even number.
*
* @param[out] result
* @param[in] x
* @param[in] y
* @param[in] m
* @returns negative on error condition. See vbw_exit_codes.h
*/
int vbw_mtx_motest_3d_byte(output_type *result, input_type* x, input_type *y, vbw_motest_t *m)
{
int l,j;
int sub_block_width = m->block_width+m->search_width;
for( j = 0; j < m->block_height; j++ ) {
vbx_dma_to_vector( m->v_block+j*m->block_width, x+j*m->image_width, m->block_width*sizeof(input_type) );
}
for( j = 0; j < m->block_height+m->search_height; j++ ) {
vbx_dma_to_vector( m->v_img+j*sub_block_width, y+j*m->image_width, sub_block_width*sizeof(input_type) );
}
vbx_set_3D( m->search_width, m->block_height*sizeof(intermediate_type), sizeof(input_type), 0 );
for( l = 0; l < m->search_height; l++ ) {
//Accumulate each row into a vbx of row SADs
vbx_set_vl( m->block_width );
vbx_set_2D( m->block_height, sizeof(intermediate_type), sub_block_width*sizeof(input_type), m->block_width*sizeof(input_type) );
vbx_acc_3D( VVBHU, VABSDIFF, m->v_row_sad, m->v_img+l*sub_block_width, m->v_block );
//Accumulate the SADs
vbx_set_vl( m->block_height/2 );
vbx_set_2D( m->search_width, sizeof(output_type), m->block_height*sizeof(intermediate_type), m->block_height*sizeof(intermediate_type) );
vbx_acc_2D( VVHWU, VADD, (vbx_uword_t*)m->v_result+l*m->search_width, m->v_row_sad, m->v_row_sad+(m->block_height/2) );
//Transfer the line to host
vbx_dma_to_host( result+l*m->search_width, m->v_result+l*m->search_width, m->search_width*sizeof(output_type) );
}
return VBW_SUCCESS;
}
/// Apply [1 2 1] low-pass filter to raw input row
/// NB: Last two output pixels are not meaningful
inline static void vbw_sobel_3x3_row(vbx_uhalf_t *lpf, vbx_uhalf_t *raw, const short image_width)
{
vbx_set_vl(image_width-1);
vbx(VVHU, VADD, lpf, raw, raw+1);
vbx_set_vl(image_width-2);
vbx(VVHU, VADD, lpf, lpf, lpf+1);
}
int compare_vbx_lbp_ci_to_scalar_patterns(unsigned short* img, int width, int height, int max_print_errors)
{
int j, errors = 0;
unsigned char** scalar_patterns = test_scalar_patterns(img, 0, width, height);
vbx_ubyte_t* v_in = (vbx_ubyte_t*)vbx_sp_malloc(3*width*sizeof(vbx_word_t));
vbx_ubyte_t* v_top = (vbx_byte_t*)vbx_sp_malloc(width*sizeof(vbx_byte_t));
vbx_ubyte_t* v_bot = (vbx_byte_t*)vbx_sp_malloc(width*sizeof(vbx_byte_t));
vbx_ubyte_t* v_lbp = v_bot;
unsigned char* lbp = (unsigned char*)vbx_shared_malloc(width*sizeof(unsigned char));
vbx_set_vl(width);
for(j=0; j < height - 2; j++){
vbx_dma_to_vector(v_in, img+j*width, 3*width*sizeof(unsigned char));
vbx(VVHU, VCUSTOM1, v_top, v_in, v_in+width);
vbx(VVHU, VCUSTOM1, v_bot, v_in+width, v_in+2*width);
vbx(SVHBU, VAND, v_top, 0xf0, v_top);
vbx(SVHBU, VAND, v_bot, 0x0f, v_bot);
vbx(VVBU, VADD, v_lbp, v_bot, v_top);
vbx_dma_to_host(lbp, v_lbp, width*sizeof(unsigned char));
vbx_sync();
errors = match_array_byte(lbp, scalar_patterns[0]+j*width, "custom_lbp", width-2, 1, max_print_errors, 1, j);
}
vbx_sp_free();
vbx_shared_free(lbp);
return errors;
}
int main(void)
{
vbx_test_init();
vbx_mxp_print_params();
int errors=0;
unsigned instr_cycles,instr_count, dma_cycles,dma_count;
vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
int lanes= this_mxp->vector_lanes;
int dma_width=this_mxp->dma_alignment_bytes /4;
debug(lanes);
debug(dma_width);
vbx_set_vl(-1);
VBX_COUNTER_RESET();
vbx(SVW,VMOV,0,0,0);
vbx_sync();
if(VBX_SIMULATOR)
printf("simulator\n");
else
printf("not simulator\n");
instr_cycles=VBX_GET_WRITEBACK_CYCLES();
dma_cycles=VBX_GET_DMA_CYCLES();
dma_count=VBX_GET_DMAS();
instr_count=VBX_GET_INSTRUCTIONS();
debug(instr_cycles);
debug(dma_cycles);
debug(dma_count);
debug(instr_count );
VBX_TEST_END(errors);
return 0;
}
void vector_rectangle_to_luma(
pixel *input_buffer,
vbx_uhalf_t *v_luma_buffer, vbx_uhalf_t *v_row_temp, vbx_uword_t *v_row,
int startx, int starty, int width, int height, const int image_pitch )
{
int y;
vbx_uhalf_t *v_luma;
vbx_set_vl(width);
for(y = 0; y < height; y++){
v_luma = v_luma_buffer+(y*width);
vbx_dma_to_vector(v_row, input_buffer+((y+starty)*image_pitch)+startx, width*sizeof(vbx_uword_t));
//Move the b component into v_luma
vbx(SVWHU, VAND, v_luma, 0xFF, v_row);
vbx(SVHU, VMUL, v_luma, 25, v_luma);
//Move g into v_row_temp and add it to v_luma
vbx(SVWHU, VAND, v_row_temp, 0xFF, (vbx_uword_t*)(((vbx_ubyte_t*)v_row)+1));
vbx(SVHU, VMUL, v_row_temp, 129, v_row_temp);
vbx(VVHU, VADD, v_luma, v_luma, v_row_temp);
//Move r into v_row_temp and add it to v_luma
vbx(SVWHU, VAND, v_row_temp, 0xFF, (vbx_uword_t*)(((vbx_ubyte_t*)v_row)+2));
vbx(SVHU, VMUL, v_row_temp, 66, v_row_temp);
vbx(VVHU, VADD, v_luma, v_luma, v_row_temp);
//divide by 2^8
vbx(SVHU, VSHR, v_luma, 8, v_luma);
}
}
int vbw_mtx_xp(vbx_sp_t *v_dst, vbx_sp_t *v_src, const int INROWS, const int INCOLS )
{
vbx_set_vl( 1 );
vbx_set_2D( INCOLS, INROWS*sizeof(vbx_sp_t), sizeof(vbx_sp_t), 0 );
vbx_set_3D( INROWS, sizeof(vbx_sp_t), INCOLS*sizeof(vbx_sp_t), 0 );
vbxx_3D( VMOV, v_dst, v_src);
return VBW_SUCCESS;
}
int compare_vbx_lut_to_vbx_lut_ci(int sz, int max_print_errors)
{
int f, n, errors;
vbx_byte_t* v_pass = (vbx_byte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_pattern = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_lutc = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_group = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_sel = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_lut = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_word_t));
vbx_ubyte_t* v_idx = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_word_t));
unsigned char* lut = (unsigned char*)vbx_shared_malloc(sz*sizeof(unsigned char));
unsigned char* lut_c = (unsigned char*)vbx_shared_malloc(sz*sizeof(unsigned char));
for (n = 0; n < sz; n++) {
v_pattern[n] = n & 0xff;
}
int s, stage = 11;
for (f = 0; f < face_lbp[stage].count; f++) {
lbp_feat_t feat = face_lbp[stage].feats[f];
vbx_set_vl(sz);
int total = f;
s = 0;
while(s < stage){
total += face_lbp[s].count;
s++;
}
vbx(SVBU, VCUSTOM0, v_lutc, total, v_pattern);
vbx(SVB, VMOV, v_pass, feat.fail, 0);
/* check if pattern is in lut */
vbx(SVBU, VSHR, v_group, 5, v_pattern);
for (n = 0; n < 8; n++) {
vbx(SVB, VADD, v_sel, -n, v_group);
vbx(SVBW, VCMV_Z, v_lut, feat.lut[n], v_sel);
}
vbx(SVBWU, VAND, v_idx, 0x1f, v_pattern);
vbx(VVWB, VSHR, v_lut, v_idx, v_lut);
vbx(SVB, VAND, v_lut, 1, v_lut);
vbx(SVB, VCMV_LEZ, v_pass, feat.pass, v_lut);
vbx_dma_to_host(lut_c, v_lutc, sz*sizeof(unsigned char));
vbx_dma_to_host(lut, v_pass, sz*sizeof(unsigned char));
vbx_sync();
errors = match_array_byte(lut_c, lut, "custom_lut", sz, 1, max_print_errors, 0, 0);
}
vbx_sp_free();
vbx_shared_free(lut);
vbx_shared_free(lut_c);
return errors;
}
int deep_vector_copy_test()
{
int retval;
int num_test;
int total_errors = 0;
const int NUM_TESTS = TEST_DEEP_SP_NUM_TESTS;
const int NB = vbx_sp_getfree();
int NT = NB / sizeof(vbx_sp_t);
vbx_sp_push();
vbx_sp_t *v = vbx_sp_malloc( NB );
srand( 0x1a84c92a );
for( num_test=0; num_test < NUM_TESTS ; num_test++ ) {
// initialize entire available scratchpad
vbx_set_vl( NT );
vbx( SE(T), VAND, v, MSK, 0 );
// choose random src/dest/length:
// -- randomly pick the dest
// -- set a window size of 2*K around the dest
// -- randomly pick the src within the window
// -- randomly pick the length, subject to end-of-scratchpad
// -- this 'window' rule increases probability of overlaps
// -- rough distribution: 30% short (pipeline) overlaps, 20% long overlaps, 50% no overlap
int K, N1, N2, NN;
N1 = rand() % NT;
K = 1 + rand() % ((N1 > 0)? min(min(N1, NT-N1), 1024): min(NT, 1024));
N2 = N1 - K + rand() % (2*K);
NN = rand() % (NT - max(N1,N2));
vbx_sp_t *dst = v + N1;
vbx_sp_t *src = v + N2;
printf("test:%d src:0x%08x dst:0x%08x len:%08d", num_test, N1, N2, NN );
// do the copy
retval = VBX_T(vbw_vec_copy)( dst, src, NN );
vbx_sync();
printf(" retval:0x%04x\n",retval);
// ensure the copy was done properly
int errors = verify_copy((vbx_mm_t *)v, 0, N1, 0, "head")
+ verify_copy((vbx_mm_t *)v, N1, NN+N1, (N2-N1), "copy")
+ verify_copy((vbx_mm_t *)v, NN+N1, NT, 0, "tail");
total_errors += errors;
if( errors ) {
//break;
}
}
vbx_sp_pop();
return total_errors;
}
vbx_mtx_fdct_t *
vbx_mtx_fdct_init( dt *coeff_v, dt *image )
{
const int BIG_TILE_SIZE = NUM_TILE_X * NUM_TILE_Y * DCT_SIZE;
const int num_bytes = BIG_TILE_SIZE * sizeof(dt);
const int co_bytes = NUM_TILE_X* DCT_SIZE *sizeof(dt);
//compute coeffs matrix in double and truncated to dt
int i, j;
double s;
for (i = 0; i < BLOCK_SIZE; i++) {
s = (i == 0) ? sqrt(0.125) : 0.5;
for (j = 0; j < BLOCK_SIZE; j++) {
c2[i][j] = s * cos((double) ((PI / 8.0) * i * j + 0.5));
cs[i][j] = (dt) (c2[i][j] * SHIFT_DOUBLE + 0.499999);
}
}
vbx_sp_push();
vbx_mtx_fdct_t *v = vbx_shared_malloc( sizeof(vbx_mtx_fct_t) );
v->vcoeff = (vbx_half_t *)vbx_sp_malloc( co_bytes );
v->vprods = (vbx_half_t *)vbx_sp_malloc( num_bytes );
#if USE_ACCUM_FLAGS
v->vaccum = (vbx_half_t *)vbx_sp_malloc( num_bytes );
v->vflags = (vbx_half_t *)vbx_sp_malloc( num_bytes );
#endif
// interleave ordering to ensure no false hazards
v->vblock[2] = (vbx_half_t *)vbx_sp_malloc( num_bytes );
v->vimage[0] = (vbx_half_t *)vbx_sp_malloc( num_bytes );
v->vblock[0] = (vbx_half_t *)vbx_sp_malloc( num_bytes );
v->vimage[1] = (vbx_half_t *)vbx_sp_malloc( num_bytes );
v->vblock[1] = (vbx_half_t *)vbx_sp_malloc( num_bytes );
if( !v->vblock[1] ) {
VBX_PRINTF( "ERROR: out of memory.\n" );
VBX_EXIT(-1);
}
vbx_dma_to_vector( v->vcoeff, coeff_v, co_bytes );
int row;
for( row=0; row < BLOCK_SIZE; row++ ) {
getBigTileImageY(v->vimage[v->db],image,row);
}
#if USE_ACCUM_FLAGS
// create a flag vector first element 0, next 'BLOCK_SIZE-1' element non-zero, etc
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE * NUM_TILE_Y * BLOCK_SIZE - (BLOCK_SIZE-1) );
vbx( SEH, VAND, v->vflags, BLOCK_SIZE-1, 0 );
#endif
return v;
}
/** Internal helper function to reverse and optionally rotate a vector of words *in the scratchpad*.
* This function uses a merge reverse algorithm that is faster on large vectors.
* @pre v_src contains the elements to reverse.
* @pre v_src, v_scratch0, and v_scratch1 must all be the same length.
* @pre v_scratch1 and v_src must not overlap.
* @pre v_src *may* overlap v_scratch0 (will clobber v_src).
* @pre MXP must be 2 lanes or more.
* @pre N is a multiple of SP_WIDTH_B.
* @pre NUM_ROWS == N*4 / SP_WIDTH_B.
* @pre v_mask must be SP_WIDTH_B bytes long.
* @post v_scratch0 and v_scratch1 contents are modified, with one containing the result.
* @post v_src clobbered only if v_src overlaps v_scratch0.
*
* @param[in] v_scratch1 *in scratch*.
* @param[in] v_src *in scratch*.
* @param[in] N is the number of words to reverse.
* @param[in] v_scratch0 *in scratch*.
* @param[in] v_mask *in scratch*.
* @param[in] SP_WIDTH_B typically the scratchpad width in bytes, it is the length of the data to be worked on at a time.
* @param[in] NUM_ROWS is the number of rows of length SP_WIDTH_B bytes.
* @param[in] rot16 TRUE to swap upper and lower half-words of each word in result.
* @returns the scratchpad address where the result resides. This will be equal to either v_scratch0 or v_scratch1,
* and will depend on log2(MXP vector lanes).
*/
static vbx_word_t *vec_rev_merge_w( vbx_word_t *v_scratch1, vbx_word_t *v_src, const unsigned int N, vbx_word_t *v_scratch0,
vbx_word_t *v_mask, const unsigned int SP_WIDTH_B, const unsigned int NUM_ROWS, const unsigned int rot16 )
{
#if !VBX_SKIP_ALL_CHECKS
if( !N || !v_scratch0 || !v_src || !v_scratch1 || !v_mask || SP_WIDTH_B < 8) {
VBX_PRINTF("Helper function vec_rev_merge_w: null pointer or row length (vector lanes) too short.");
VBX_EXIT(-1);
}
#endif
vbx_word_t *v_scratch[2] = { v_scratch0, v_scratch1 };
unsigned int W = SP_WIDTH_B/4/2; // half the number of words in a row
unsigned int sel = 1;
if( rot16 ) {
vbx_set_vl( W );
vbx_set_2D( NUM_ROWS, -SP_WIDTH_B, 0, SP_WIDTH_B );
vbx_2D( SVWU, VROTL, (vbx_uword_t *)(v_scratch[sel]+N-W), 16, (vbx_uword_t *)v_src );
vbx_2D( SVWU, VROTL, (vbx_uword_t *)(v_scratch[sel]+N-(W*2)), 16, (vbx_uword_t *)(v_src+W) );
} else {
vbx_set_vl( W );
vbx_set_2D( NUM_ROWS, -SP_WIDTH_B, SP_WIDTH_B, 0 );
vbx_2D( VVW, VMOV, v_scratch[sel]+N-W, v_src, 0 );
vbx_2D( VVW, VMOV, v_scratch[sel]+N-(W*2), v_src+W, 0 );
}
vbx_set_vl( SP_WIDTH_B/4 );
vbx_set_2D( NUM_ROWS, SP_WIDTH_B, SP_WIDTH_B, 0 );
while( W > 1 ) {
// set up odd/even mask register
W /= 2;
vbx( SEW, VAND, v_mask, W, 0 );
vbx_2D( VVW, VCMV_NZ, v_scratch[!sel], v_scratch[sel]-W, v_mask );
vbx_2D( VVW, VCMV_Z , v_scratch[!sel], v_scratch[sel]+W, v_mask );
sel = !sel;
}
return v_scratch[sel];
}
/** VBX Motion Estimation.
* Similar to the scalar version but scans vertically as it makes it easier to align vectors.
* vbw_mtx_motest_byte_setup should be run prior to running this function.
*
* @param[out] result
* @param[in] x
* @param[in] y
* @param[in] m
* @returns negative on error condition. See vbw_exit_codes.h
*/
int vbw_mtx_motest_byte(output_type *result, input_type *x, input_type *y, vbw_motest_t *m)
{
int j;
int sub_block_width = m->block_width+m->search_width;
for( j = 0; j < m->block_height; j++ ) {
vbx_dma_to_vector( m->v_block+j*sub_block_width, x+j*m->image_width, sub_block_width*sizeof(input_type) );
}
for( j = 0; j < m->block_height+m->search_height; j++ ) {
vbx_dma_to_vector( m->v_img +j*sub_block_width, y+j*m->image_width, sub_block_width*sizeof(input_type) );
}
// column-ize the reference block
vbx_set_vl( m->block_width );
vbx_set_2D( m->block_height, m->block_width*sizeof(input_type), sub_block_width*sizeof(input_type), 0 );
vbx_2D( VVB, VMOV, (vbx_byte_t*)m->v_block, (vbx_byte_t*)m->v_block, 0 );
//Do column by column
for( j=0; j < m->search_width; j++ )
{
// column-ize the search image
vbx_set_vl( m->block_width );
vbx_set_2D( m->block_height+m->search_height, m->block_width*sizeof(input_type), sub_block_width*sizeof(input_type), 0 );
vbx_2D( VVBU, VMOV, m->v_img_sub, m->v_img+j, 0 );
// search the image columnwise
vbx_set_vl( m->block_width*m->block_height );
vbx_set_2D( m->search_height, m->search_width*sizeof(output_type), 0, m->block_width*sizeof(input_type) );
vbx_acc_2D( VVBWU, VABSDIFF, (vbx_uword_t*)m->v_result+j, m->v_block, m->v_img_sub );
}
// Write back result
vbx_dma_to_host( result, m->v_result, m->result_size );
return VBW_SUCCESS;
}
int vbw_sobel_argb32_3x3(unsigned *output, unsigned *input, const short image_width, const short image_height, const short image_pitch, const short renorm)
{
size_t free_sp=vbx_sp_getfree();
size_t vectors_needed=8;
size_t partial_width=free_sp/(vectors_needed*sizeof(vbx_uword_t));
if(partial_width>image_width){
vbw_sobel_argb32_3x3_partial(output, input, image_width, image_height, image_pitch,renorm);
}else{
//can do entire row at a time, so do partial_width at a time
size_t partial_step=partial_width-2;
int i;
for(i=0;;i+=partial_step){
//account for last tile being smaller
if(i+partial_width > image_width){
partial_width=image_width-i;
}
vbw_sobel_argb32_3x3_partial(output+i, input+i, partial_width, image_height, image_pitch,renorm);
if(i+partial_width == image_width){
//that was the last tile, so break,
//I don't believe that this can be in the for statement
break;
}
}
}
vbx_sp_push();
vbx_word_t* side=vbx_sp_malloc(sizeof(vbx_word_t));
vbx_set_vl(1);
vbx(SVW,VMOV,side,0,0);
vbx_dma_to_host_2D(output,/*host_ptr*/
side,/*sp_ptr*/
sizeof(vbx_word_t),/*row len*/
image_height,/*num rows*/
image_pitch*sizeof(vbx_word_t),/*host_incr*/
0);/*sp incr*/
vbx_dma_to_host_2D(output+image_width-1,/*host_ptr*/
side,/*sp_ptr*/
sizeof(vbx_word_t),/*row len*/
image_height,/*num rows*/
image_pitch*sizeof(vbx_word_t),/*host_incr*/
0);/*sp incr*/
vbx_sp_pop();
vbx_sync();
}
/// Convert a row of aRGB pixels into luma values
/// v_luma should not equal v_row_in
/// Trashes v_temp
static void vbw_rgb2luma(vbx_uhalf_t *v_luma, vbx_uword_t *v_row_in, vbx_uhalf_t *v_temp, const int image_width)
{
vbx_set_vl(image_width);
// Move weighted B into v_luma
vbx(SVWHU, VAND, v_temp, 0xFF, v_row_in);
vbx(SVHU, VMUL, v_luma, 25, v_temp);
// Move weighted G into v_temp and add it to v_luma
vbx(SVWHU, VAND, v_temp, 0xFF, (vbx_uword_t*)(((vbx_ubyte_t *)v_row_in)+1));
vbx(SVHU, VMUL, v_temp, 129, v_temp);
vbx(VVHU, VADD, v_luma, v_luma, v_temp);
// Move weighted R into v_temp and add it to v_luma
vbx(SVWHU, VAND, v_temp, 0xFF, (vbx_uword_t*)(((vbx_ubyte_t *)v_row_in)+2));
vbx(SVHU, VMUL, v_temp, 66, v_temp);
vbx(VVHU, VADD, v_luma, v_luma, v_temp);
vbx(SVHU, VADD, v_luma, 128, v_luma); // for rounding
vbx(SVHU, VSHR, v_luma, 8, v_luma);
}
static int isAbsOutOfRangeV( vptr_half v_src_r, vptr_half v_src_i, vptr_half v_temp, int n )
{
//used for inverse only
vbx_set_vl(n);
vbx(SVH, VABSDIFF, v_temp, 0, v_src_r ); // get abs value of real
vbx(SVH, VSUB, v_temp, 16383, v_temp ); // if (16383 - v_src) < 0, needs scaling
vbx_acc(SVH, VCMV_LTZ, v_temp, 1, v_temp ); // accum # of neg values to see if scaling required
vbx_sync();
if( v_temp[0] ){
return 1;
}
vbx(SVH, VABSDIFF, v_temp, 0, v_src_i ); // get abs value of imag
vbx(SVH, VSUB, v_temp, 16383, v_temp ); // if (16383 - v_src) < 0, needs scaling
vbx_acc(SVH, VCMV_LTZ, v_temp, 1, v_temp ); // accum # of neg values to see if scaling required
vbx_sync();
if( v_temp[0] ){
return 1;
}
return 0;
}
int compare_vbx_lbp_ci_to_scalar_patterns(unsigned short* img, int log, int width, int height, int max_print_errors)
{
int j, l, cell, max_cell, errors = 0;
unsigned char** scalar_patterns = test_scalar_patterns(img, log, width, height);
max_cell = 1<<log;
vbx_uhalf_t* v_in = (vbx_uhalf_t*)vbx_sp_malloc((1+2*max_cell)*width*sizeof(vbx_half_t));
vbx_uhalf_t* v_top = (vbx_half_t*)vbx_sp_malloc(width*sizeof(vbx_half_t));
vbx_uhalf_t* v_bot = (vbx_half_t*)vbx_sp_malloc(width*sizeof(vbx_half_t));
vbx_ubyte_t* v_lbp = (vbx_ubyte_t*)v_bot;
unsigned char* lbp = (unsigned char*)vbx_shared_malloc(width*sizeof(unsigned char));
vbx_set_vl(width);
for(l = 0; l < 1; l++){
cell = 1<<l;
for(j=0; j < height - 2*cell; j++){
vbx_dma_to_vector(v_in, img+j*width, (1+2*cell)*width*sizeof(unsigned short));
vbx(VVHU, VCUSTOM1, v_top, v_in, v_in+(1*cell)*width);
vbx(VVHU, VCUSTOM1, v_bot, v_in+(1*cell)*width, v_in+(2*cell)*width);
vbx(SVHBU, VAND, (vbx_ubyte_t*)v_top, 0xf0, v_top);
vbx(SVHBU, VAND, (vbx_ubyte_t*)v_bot, 0x0f, v_bot);
vbx(VVBU, VADD, v_lbp, v_bot, v_top);
vbx_dma_to_host(lbp, v_lbp, width*sizeof(unsigned char));
vbx_sync();
errors += match_array_byte(lbp, scalar_patterns[l]+j*width, "custom_lbp", width-2*cell, 1, 0, max_print_errors, 1, j);
if (errors > max_print_errors){
max_print_errors = 0;
}
}
}
vbx_sp_free();
vbx_shared_free(lbp);
return errors;
}
int test_lbp_ci(unsigned short* img, int width, int height)
{
vbx_uhalf_t* v_a1 = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_b1 = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_1h = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_a2 = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_b2 = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_2h = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_a4 = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_b4 = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_uhalf_t* v_4h = (vbx_uhalf_t*)vbx_sp_malloc(width*sizeof(vbx_uhalf_t));
vbx_ubyte_t* v_1b = (vbx_ubyte_t*)vbx_sp_malloc(width*sizeof(vbx_ubyte_t));
vbx_ubyte_t* v_2b = (vbx_ubyte_t*)vbx_sp_malloc(width*sizeof(vbx_ubyte_t));
vbx_ubyte_t* v_4b = (vbx_ubyte_t*)vbx_sp_malloc(width*sizeof(vbx_ubyte_t));
unsigned short* lbp1h = (unsigned short*)vbx_shared_malloc(width*sizeof(unsigned short));
unsigned short* lbp2h = (unsigned short*)vbx_shared_malloc(width*sizeof(unsigned short));
unsigned short* lbp4h = (unsigned short*)vbx_shared_malloc(width*sizeof(unsigned short));
unsigned char* lbp1b = (unsigned char*)vbx_shared_malloc(width*sizeof(unsigned char));
unsigned char* lbp2b = (unsigned char*)vbx_shared_malloc(width*sizeof(unsigned char));
unsigned char* lbp4b = (unsigned char*)vbx_shared_malloc(width*sizeof(unsigned char));
img = img + width;
vbx_dma_to_vector(v_a1, img, width*sizeof(unsigned short));
vbx_dma_to_vector(v_b1, img + width, width*sizeof(unsigned short));
vbx_dma_to_vector(v_a2, img, width*sizeof(unsigned short));
vbx_dma_to_vector(v_b2, img + width, width*sizeof(unsigned short));
vbx_dma_to_vector(v_a4, img, width*sizeof(unsigned short));
vbx_dma_to_vector(v_b4, img + width, width*sizeof(unsigned short));
vbx_sync();
int i;
int m = 48;
for(i=0; i<m; i++){
v_a1[i] = 0;
v_b1[i] = 0;
v_a2[i] = 0;
v_b2[i] = 0;
v_a4[i] = 0;
v_b4[i] = 0;
}
int n = 12;
int src_a1[] = {0, 2, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
int src_b1[] = {0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
int src_a2[] = {0, 0, 0, 2, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
int src_b2[] = {0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
int src_a4[] = {0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 2, 0, 0, 0, 0};
int src_b4[] = {0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0};
for(i=0; i<16; i++){
v_a1[i] = src_a1[i];
v_b1[i] = src_b1[i];
v_a2[i] = src_a2[i];
v_b2[i] = src_b2[i];
v_a4[i] = src_a4[i];
v_b4[i] = src_b4[i];
}
vbx_set_vl(width);
vbx(VVHU, VCUSTOM1, v_1h, v_a1, v_b1);
vbx(VVHU, VCUSTOM2, v_2h, v_a2, v_b2);
vbx(VVHU, VCUSTOM3, v_4h, v_a4, v_b4);
vbx(VVHB, VADD, v_1b, v_1h, ((vbx_byte_t*)v_1h) + 1);
vbx(VVHB, VADD, v_2b, v_2h, ((vbx_byte_t*)v_2h) + 1);
vbx(VVHB, VADD, v_4b, v_4h, ((vbx_byte_t*)v_4h) + 1);
vbx_dma_to_host(lbp1h, v_1h, width*sizeof(unsigned short));
vbx_dma_to_host(lbp2h, v_2h, width*sizeof(unsigned short));
vbx_dma_to_host(lbp4h, v_4h, width*sizeof(unsigned short));
vbx_dma_to_host(lbp1b, v_1b, width*sizeof(unsigned char));
vbx_dma_to_host(lbp2b, v_2b, width*sizeof(unsigned char));
vbx_dma_to_host(lbp4b, v_4b, width*sizeof(unsigned char));
vbx_sync();
test_print_array_half(v_a1, n);
test_print_array_half(v_b1, n);
test_print_hex_array_half(lbp1h, n);
test_print_hex_array_byte(lbp1b, n);
test_print_array_half(v_a2, n);
test_print_array_half(v_b2, n);
test_print_hex_array_half(lbp2h, n);
test_print_hex_array_byte(lbp2b, n);
test_print_array_half(v_a4, n);
test_print_array_half(v_b4, n);
test_print_hex_array_half(lbp4h, n);
test_print_hex_array_byte(lbp4b, n);
vbx_sp_free();
vbx_shared_free(lbp1h);
vbx_shared_free(lbp2h);
vbx_shared_free(lbp4h);
vbx_shared_free(lbp1b);
vbx_shared_free(lbp2b);
vbx_shared_free(lbp4b);
return 0;
}
//FIXME stride for match not implemented
int compare_LBPPassStage_to_restricted(unsigned short *vbx_img, int log, lbp_stage_t lbp_stage, int window, int width, int height, int max_print_errors)
{
int l, i, j, cell, errors = 0;
unsigned char** scalar_patterns = test_scalar_patterns(vbx_img, log, width, height);
unsigned char *pass, *vbx_pass;
pass = (unsigned char*)vbx_shared_malloc(width*height*sizeof(unsigned char));
vbx_pass = (unsigned char*)vbx_shared_malloc(width*height*sizeof(unsigned char));
vbx_byte_t** v_lbp =(vbx_byte_t**)vbx_shared_malloc((log+1)*sizeof(vbx_byte_t*));
for (l=0; l<log+1; l++) {
v_lbp[l] = (vbx_byte_t*)vbx_sp_malloc((window+1)*width*sizeof(vbx_byte_t));
}
vbx_byte_t* v_lut = (vbx_byte_t*)vbx_sp_malloc(width*sizeof(vbx_byte_t));
vbx_byte_t* v_stage = (vbx_byte_t*)vbx_sp_malloc(width*sizeof(vbx_byte_t));
vbx_byte_t* v_pattern;
lbp_feat_t feat;
int dx, dy, dw, f;
for (l=0; l<log+1; l++) {
vbx_dma_to_vector(v_lbp[l]+width, scalar_patterns[l], (window)*width*sizeof(unsigned char));
}
vbx_sync();
for(j=0; j < height-(window+1); j++) {
for (l=0; l<log+1; l++) {
vbx_set_vl(width * window);
vbx(VVB, VMOV, v_lbp[l], v_lbp[l]+width, NULL);
vbx_dma_to_vector(v_lbp[l] + window*width, scalar_patterns[l]+(j+window)*width, width*sizeof(unsigned char));
}
vbx_set_vl(width-(window+1));
vbx(SVB, VMOV, v_stage, 0, NULL);
for (f = 0; f < lbp_stage.count; f++) {
feat = lbp_stage.feats[f];
dx = feat.pos.src.x;
dy = feat.pos.src.y;
dw = feat.pos.size.x;
v_pattern = v_lbp[dw>>1]+(dy*width+dx);
vbx(SVBU, VLBPLUT, v_lut, f, v_pattern);
vbx(VVB, VADD, v_stage, v_stage, v_lut);
}
vbx(SVB, VMOV, v_lut, 0, NULL);
vbx(SVB, VCMV_GEZ, v_lut, 1, v_stage);
vbx_dma_to_host(vbx_pass + j*width, v_lut, (width-(window+1))*sizeof(unsigned char));
vbx_sync();
}
unsigned int *iImg, *iiImg;
iImg = (unsigned int *)vbx_shared_malloc(width*height*sizeof(unsigned int));
iiImg = (unsigned int *)vbx_shared_malloc(width*height*sizeof(unsigned int));
gen_integrals(vbx_img, iImg, iiImg, width, height);
image_t lbp_img = {iImg, {width, height}};
for (j = 0; j < height - (window + 1); j++) {
for (i = 0; i < width - (window + 1); i++) {
pair_t lbp_p = {i, j};
pass[j*width+i] = LBPPassStage(lbp_img, lbp_stage, lbp_p);
}
}
/* test pass vs vbx pass */
for (j = 0; j < height - (window + 1); j++) {
errors += match_array_byte(vbx_pass + j*width, pass + j*width, "pass stage", width - (window + 1), 1, 0, max_print_errors, 1, j);
if (errors > max_print_errors){
max_print_errors = 0;
}
}
return errors;
}
int VectorBlox_MXP_Initialize(const char* mxp_dev,const char* cma_dev)
{
PAGE_SIZE=sysconf(_SC_PAGESIZE);
PAGE_SHIFT=0;
int page_size=PAGE_SIZE;
while((page_size>>=1)){
PAGE_SHIFT++;
}
char filename[256];
sprintf(filename,"/dev/%s",mxp_dev);
the_mxp.mxp_fd=open(filename,O_RDWR);
assert(the_mxp.mxp_fd);
the_mxp.scratchpad_size = get_attr_from_file(mxp_dev,"SCRATCHPAD_KB") * 1024;
the_mxp.scratchpad_addr = (void*)get_attr_from_file(mxp_dev,"C_S_AXI_BASEADDR");
void* scratchpad_mmap = mmap(the_mxp.scratchpad_addr, //for the mapping to be the same as the physical mapping
the_mxp.scratchpad_size,
PROT_READ|PROT_WRITE,MAP_SHARED|MAP_FIXED,
the_mxp.mxp_fd,4096);
assert(scratchpad_mmap == the_mxp.scratchpad_addr);
the_mxp.scratchpad_end = (void*)get_attr_from_file(mxp_dev,"C_S_AXI_HIGHADDR")+1;
//M_AXI_DATA_WIDTH is in bits, convert to bytes
the_mxp.dma_alignment_bytes = get_attr_from_file(mxp_dev,"C_M_AXI_DATA_WIDTH")/8;
the_mxp.vector_lanes = get_attr_from_file(mxp_dev,"VECTOR_LANES");
the_mxp.scratchpad_alignment_bytes = the_mxp.vector_lanes * 4;
the_mxp.vcustom0_lanes = get_attr_from_file(mxp_dev, "VCI_0_LANES");
the_mxp.vcustom1_lanes = get_attr_from_file(mxp_dev, "VCI_1_LANES");
the_mxp.vcustom2_lanes = get_attr_from_file(mxp_dev, "VCI_2_LANES");
the_mxp.vcustom3_lanes = get_attr_from_file(mxp_dev, "VCI_3_LANES");
the_mxp.vcustom4_lanes = get_attr_from_file(mxp_dev, "VCI_4_LANES");
the_mxp.vcustom5_lanes = get_attr_from_file(mxp_dev, "VCI_5_LANES");
the_mxp.vcustom6_lanes = get_attr_from_file(mxp_dev, "VCI_6_LANES");
the_mxp.vcustom7_lanes = get_attr_from_file(mxp_dev, "VCI_7_LANES");
the_mxp.vcustom8_lanes = get_attr_from_file(mxp_dev, "VCI_8_LANES");
the_mxp.vcustom9_lanes = get_attr_from_file(mxp_dev, "VCI_9_LANES");
the_mxp.vcustom10_lanes = get_attr_from_file(mxp_dev, "VCI_10_LANES");
the_mxp.vcustom11_lanes = get_attr_from_file(mxp_dev, "VCI_11_LANES");
the_mxp.vcustom12_lanes = get_attr_from_file(mxp_dev, "VCI_12_LANES");
the_mxp.vcustom13_lanes = get_attr_from_file(mxp_dev, "VCI_13_LANES");
the_mxp.vcustom14_lanes = get_attr_from_file(mxp_dev, "VCI_14_LANES");
the_mxp.vcustom15_lanes = get_attr_from_file(mxp_dev, "VCI_15_LANES");
the_mxp.mask_partitions = get_attr_from_file(mxp_dev,"MASK_PARTITIONS");
the_mxp.max_masked_vector_length = get_attr_from_file(mxp_dev,"MAX_MASKED_WAVES")* the_mxp.vector_lanes * 4;
the_mxp.fxp_word_frac_bits = get_attr_from_file(mxp_dev,"MULFXP_WORD_FRACTION_BITS");
the_mxp.fxp_half_frac_bits = get_attr_from_file(mxp_dev,"MULFXP_HALF_FRACTION_BITS");
the_mxp.fxp_byte_frac_bits = get_attr_from_file(mxp_dev,"MULFXP_BYTE_FRACTION_BITS");
the_mxp.core_freq = get_attr_from_file(mxp_dev,"CLOCK_FREQ_HZ");
the_mxp.instr_port_addr = mmap(NULL,PAGE_SIZE,PROT_READ|PROT_WRITE,MAP_SHARED,the_mxp.mxp_fd,0);
sprintf(filename,"/dev/%s",cma_dev);
the_mxp.cma_fd = open(filename,O_RDWR);
assert(the_mxp.cma_fd);
the_mxp.init = 0;
the_mxp.sp = the_mxp.scratchpad_addr;
the_mxp.spstack = (vbx_void_t **) NULL;
the_mxp.spstack_top = (int) 0;
the_mxp.spstack_max = (int) 0;
_vbx_init(&the_mxp);
//clear scratchpad
vbx_set_vl(the_mxp.scratchpad_size);
vbx(SVB,VMOV,(vbx_byte_t*)the_mxp.scratchpad_addr,0,0);
return 0;
}
int VBX_T(vbw_vec_reverse_test)()
{
unsigned int aN[] = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 12, 15, 16, 17, 20, 25, 31, 32, 33, 35, 40, 48, 60, 61, 62, 63, 64, 64, 65,
66, 67, 68, 70, 80, 90, 99, 100, 101, 110, 128, 128, 144, 144, 160, 160, 176, 176, 192, 192, 224, 224,
256, 256, 288, 288, 320, 320, 352, 352, 384, 384, 400, 450, 512, 550, 600, 650, 700, 768, 768, 900,
900, 1023, 1024, 1200, 1400, 1600, 1800, 2048, 2048, 2100, 2200, 2300, 2400, 2500, 2600, 2700, 2800,
2900, 3000, 3100, 3200, 3300, 3400, 3500, 3600, 3700, 3800, 3900, 4000, 4096, 4096, 4100, 4200, 4300,
4400, 4500, 4600, 4700, 4800, 4900, 5000, 6000, 7000, 8000, 8192, 8192, 9000, 10000, 11000, 12000,
13000, 14000, 15000, 16000, 16384, 16384, 20000, 25000, 30000, 32767, 32768, 32768, 35000, 40000,
45000, 50000, 55000, 60000, 65000, 65535, 65536, 65536 };
int retval;
unsigned int N;
unsigned int NBYTES;
unsigned int NREPS = 100;
unsigned int i,k;
vbx_timestamp_t start=0,finish=0;
vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
const unsigned int VBX_SCRATCHPAD_SIZE = this_mxp->scratchpad_size;
for( i=0; i<sizeof(aN)/4; i++ ) {
N = aN[i];
//printf( "testing with vector size %d\n", N );
NBYTES = sizeof(vbx_sp_t)*N;
if( 2*NBYTES > VBX_SCRATCHPAD_SIZE ) continue;
vbx_sp_t *vsrc = vbx_sp_malloc( NBYTES );
vbx_sp_t *vdst = vbx_sp_malloc( NBYTES );
//printf("bytes alloc: %d\n", NBYTES );
if( !vsrc ) VBX_EXIT(-1);
if( !vdst ) VBX_EXIT(-1);
#if ( VBX_TEMPLATE_T == BYTESIZE_DEF | VBX_TEMPLATE_T == UBYTESIZE_DEF )
unsigned int mask = 0x007F;
#elif ( VBX_TEMPLATE_T == HALFSIZE_DEF | VBX_TEMPLATE_T == UHALFSIZE_DEF )
unsigned int mask = 0x7FFF;
#else
unsigned int mask = 0xFFFF;
#endif
vbx_set_vl( N );
vbx( SV(T), VMOV, vdst, -1, 0 ); // Fill the destination vector with -1
vbx( SE(T), VAND, vsrc, mask, 0 ); // Fill the source vector with enumerated values
//VBX_T(print_vector)( "vsrcInit", vsrc, N );
//VBX_T(print_vector)( "vdstInit", vdst, N );
/** measure performance of function call **/
vbx_sync();
start = vbx_timestamp();
for(k=0; k<NREPS; k++ ) {
retval = VBX_T(vbw_vec_reverse)( vdst, vsrc, N );
vbx_sync();
}
finish = vbx_timestamp();
printf( "length %d (%s):\tvbware sp f():\t%llu", N, VBX_EXPAND_AND_QUOTE(BYTEHALFWORD), (unsigned long long) vbx_mxp_cycles((finish-start)/NREPS) );
//VBX_T(print_vector)( "vsrcPost", vsrc, N );
//VBX_T(print_vector)( "vdstPost", vdst, N );
#if VERIFY_VBWARE_ALGORITHM
VBX_T(verify_vector)( vsrc, vdst, N );
#else
printf(" [VERIFY OFF]");
#endif
printf("\treturn value: %X", retval);
vbx_set_vl( N );
vbx( SE(T), VAND, vsrc, mask, 0 ); // Reset the source vector
/** measure performance of simple algorithm **/
vbx_sync();
vbx_set_vl( 1 );
vbx_set_2D( N, -sizeof(vbx_sp_t), sizeof(vbx_sp_t), 0 );
start = vbx_timestamp();
for(k=0; k<NREPS; k++ ) {
vbx_2D( VV(T), VMOV, vdst+N-1, vsrc, 0 );
vbx_sync();
}
finish = vbx_timestamp();
printf( "\tsimple (vl=1):\t%llu", (unsigned long long) vbx_mxp_cycles((finish-start)/NREPS) );
#if VERIFY_SIMPLE_ALGORITHM
VBX_T(verify_vector)( vsrc, vdst, N );
#else
printf(" [VERIFY OFF]");
#endif
printf("\tcycles\n");
vbx_sp_free();
}
vbx_sp_free();
printf("All tests passed successfully.\n");
return 0;
}
int vbw_mtx_median_ext_argb32( unsigned *output, unsigned *input, const int filter_height, const int filter_width,
const int image_height, const int image_width, const int image_pitch )
{
const int FREE_BYTES = vbx_sp_getfree();
int l,k;
int filter_mid, filter_size;
int rows_per_l,vl,temp_vl, temp_vl_byte;
int j,i;
int partial_row = 0;
filter_size = filter_height*filter_width;
filter_mid = filter_size/2;
vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
const int VBX_WIDTH_BYTES = this_mxp->scratchpad_alignment_bytes;
// Could possibly check for low SP here (less than 6*VBX_WIDTH_BYTES) and assign vl differently
// During allocation, max additional SP bytes needed due to alignment is one VBX_WIDTH_BYTES per vector
// Taking that off the top simplifies calculation and will always be correct, but sacrifices a little SP space
vl = (FREE_BYTES-3*VBX_WIDTH_BYTES)/((filter_size+2)*sizeof(vbx_uword_t));
if( vl < 1 ) {
return VBW_ERROR_SP_ALLOC_FAILED;
}
if(vl < image_width){
rows_per_l = 1;
partial_row = 1;
} else {
rows_per_l = vl/image_width;
vl = image_width*rows_per_l;
}
vbx_sp_push();
vbx_uword_t *v_input = (vbx_uword_t *)vbx_sp_malloc(filter_size*vl*sizeof(vbx_uword_t));
vbx_ubyte_t *v_sub = (vbx_ubyte_t *)vbx_sp_malloc(vl*sizeof(vbx_uword_t));
vbx_ubyte_t *v_temp = (vbx_ubyte_t *)vbx_sp_malloc(vl*sizeof(vbx_uword_t));
vbx_ubyte_t *v_min, *v_max;
vbx_ubyte_t *v_input_byte = (vbx_ubyte_t *)v_input;
if( v_temp == NULL ){
vbx_sp_pop();
return VBW_ERROR_SP_ALLOC_FAILED;
}
for(l = 0; l < image_height-filter_height; l+= rows_per_l){
// detect last pass
if(l+rows_per_l > image_height-filter_height){
rows_per_l = (image_height-filter_height)-l;
vl = image_width*rows_per_l;
}
temp_vl = vl;
for(k = 0; k < image_width; k += temp_vl){
if(partial_row){
if(k + temp_vl > image_width){
temp_vl = image_width - k;
}
}
for(j = 0; j < filter_height; j++){
vbx_dma_to_vector_2D(v_input+temp_vl*j,
input+(l+j)*image_pitch+k,
temp_vl/rows_per_l*sizeof(vbx_uword_t),
rows_per_l,
image_width*sizeof(vbx_uword_t),
image_pitch*sizeof(vbx_uword_t));
}
// arrange all pixels within a filter window into single columns, seperated by temp_vl
//
// ex. vl = 5, filter = 3
// vinput before vinput after
//
// a00 a01 a02 a03 a04 | a00 a01 a02 a03 a04 |
// a10 a11 a12 a13 a14 | a10 a11 a12 a13 a14 |
// a20 a21 a22 a23 a24 | a20 a21 a22 a23 a24 |
// ??? ??? ??? ??? ??? | a01 a02 a03 a04 a10 |
// ??? ??? ??? ??? ??? | a11 a12 a13 a14 a20 |
// ??? ??? ??? ??? ??? | a21 a22 a23 a24 a30 |
// ??? ??? ??? ??? ??? | a02 a03 a04 a10 a11 |
// ??? ??? ??? ??? ??? | a12 a13 a14 a20 a21 |
// ??? ??? ??? ??? ??? | a22 a23 a24 a30 a31 |
//
vbx_set_vl(temp_vl);
for(j = 1; j < filter_height; j++){
for(i = 0; i < filter_width; i++){
vbx(VVWU, VMOV, v_input+(j*filter_height+i)*temp_vl,
v_input+i*temp_vl+j,
0);
}
}
//Do the bubble sort up to the filter_size/2^th element on each vbx
// work on individual color channels
temp_vl_byte = temp_vl*sizeof(vbx_uword_t)/sizeof(vbx_ubyte_t);
vbx_set_vl(temp_vl_byte);
// sort lower half of the values in the window
for(j = 0; j < filter_mid; j++){
v_min = v_input_byte+j*temp_vl_byte;
for(i = j+1; i < filter_size; i++){
v_max = v_input_byte+i*temp_vl_byte;
vbx(VVBU, VMOV, v_temp, v_min, 0);
vbx(VVBU, VSUB, v_sub, v_max, v_min);
vbx(VVBU, VCMV_LTZ, v_min, v_max, v_sub);
vbx(VVBU, VCMV_LTZ, v_max, v_temp, v_sub);
}
}
// grab next smallest value, the median, don't sort the rest
v_min = v_input_byte+filter_mid*temp_vl_byte;
for(i = filter_mid+1; i < filter_size; i++){
v_max = v_input_byte+i*temp_vl_byte;
vbx(VVBU, VSUB, v_sub, v_max, v_min);
vbx(VVBU, VCMV_LTZ, v_min, v_max, v_sub);
}
// dma out median value
// back to pixels
vbx_dma_to_host_2D(output+(l*image_pitch)+k,
v_input+temp_vl*filter_mid,
temp_vl/rows_per_l*sizeof(vbx_uword_t),
rows_per_l,
image_pitch*sizeof(vbx_uword_t),
image_width*sizeof(vbx_uword_t));
}
}
vbx_sp_pop();
vbx_sync();
return VBW_SUCCESS;
}
int vbw_bifilt_argb32_3x3(unsigned *output, unsigned *input, short image_width, const short image_height, const short image_pitch, const short renorm)
{
//return vbw_sobel_argb32_3x3( output, input, image_width, image_height, image_pitch, renorm);
int y;
int xx, yy, sharp;
vbx_uword_t *v_row_in;
vbx_ubyte_t *v_luma_top, *v_luma_mid, *v_luma_bot;
vbx_ubyte_t *v_luma_hii, *v_luma_low;
vbx_ubyte_t *v_src[W][W];
vbx_uword_t *v_row_out;
vbx_ubyte_t *v00, *v01, *v02, *v10, *v11, *v12, *v20, *v21, *v22;
#if W==5
vbx_ubyte_t *v03, *v04, *v13, *v14, *v23, *v24;
vbx_ubyte_t *v30, *v31, *v32, *v40, *v41, *v42;
vbx_ubyte_t *v33, *v34, *v43, *v44;
#endif
vbx_ubyte_t *v[W][W];
vbx_uhalf_t *vI, *vW, *vT; // vT== temporary
vbx_sp_push();
// Allocate space in scratchpad for vectors
struct rotating_prefetcher_t v_row_db=rotating_prefetcher(1,image_width*sizeof(vbx_uword_t),
input,input+image_height*image_pitch,
image_pitch*sizeof(vbx_uword_t));
v_row_out = (vbx_uword_t*)vbx_sp_malloc(image_width*sizeof(vbx_uword_t));
vT = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
#if 1
// save some space by overlapping with v_row_out
vW = (vbx_uhalf_t*)v_row_out;
vI = (vbx_uhalf_t*)v_row_out + image_width;
#else
vW = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
vI = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
#endif
#if W==3
v_luma_top = (vbx_ubyte_t*)vbx_sp_malloc( 3 * image_width*sizeof(vbx_ubyte_t));
v_luma_mid = v_luma_top + 1 * image_width*sizeof(vbx_ubyte_t) ;
v_luma_bot = v_luma_top + 2 * image_width*sizeof(vbx_ubyte_t) ;
#else
v_luma_top = (vbx_ubyte_t*)vbx_sp_malloc( 5 * image_width*sizeof(vbx_ubyte_t));
v_luma_hii = v_luma_top + 1 * image_width*sizeof(vbx_ubyte_t) ;
v_luma_mid = v_luma_top + 2 * image_width*sizeof(vbx_ubyte_t) ;
v_luma_low = v_luma_top + 3 * image_width*sizeof(vbx_ubyte_t) ;
v_luma_bot = v_luma_top + 4 * image_width*sizeof(vbx_ubyte_t) ;
#endif
if(v_luma_bot==NULL){
vbx_sp_pop();
return VBW_ERROR_SP_ALLOC_FAILED;
}
// Transfer the first 3 input rows and interleave first 2 rgb2luma and first 2 sobel row calculations
#if W==3
rp_fetch(&v_row_db);
v_row_in = rp_get_buffer(&v_row_db,0);
vbw_rgb2luma(vW, v_row_in, vT, image_width); // 1st luma row
vbx( SVHBU, VSHR, v_luma_top, 8, vW ); // convert to byte
v_row_in = rp_fetch(&v_row_db);
v_row_in = rp_get_buffer(&v_row_db,0);
vbw_rgb2luma( vW, v_row_in, vT, image_width); // 2nd luma row
vbx( SVHBU, VSHR, v_luma_mid, 8, vW ); // convert to byte
#else
rp_fetch(&v_row_db);
v_row_in = rp_get_buffer(&v_row_db,0);
vbw_rgb2luma(vW, v_row_in, vT, image_width); // 1st luma row
vbx( SVHBU, VSHR, v_luma_top, 8, vW ); // convert to byte
rp_fetch(&v_row_db);
v_row_in = rp_get_buffer(&v_row_db,0);
vbw_rgb2luma( vW, v_row_in, vT, image_width); // 2nd luma row
vbx( SVHBU, VSHR, v_luma_hii, 8, vW ); // convert to byte
rp_fetch(&v_row_db);
v_row_in = rp_get_buffer(&v_row_db,0);
vbw_rgb2luma( vW, v_row_in, vT, image_width); // 2nd luma row
vbx( SVHBU, VSHR, v_luma_mid, 8, vW ); // convert to byte
rp_fetch(&v_row_db);
v_row_in = rp_get_buffer(&v_row_db,0);
vbw_rgb2luma( vW, v_row_in, vT, image_width); // 2nd luma row
vbx( SVHBU, VSHR, v_luma_low, 8, vW ); // convert to byte
#endif
// blank out the top and bottom rows
unsigned *out;
vbx_set_vl(image_width);
unsigned COLOUR = ( 200 | (128<<8) | (244<<16) );
vbx(SVWU, VMOV, v_row_out, COLOUR, 0);
for( y=0; y<W/2; y++ ) {
// Set top output rows to 0
out = output + image_width*y;
vbx_dma_to_host( out, v_row_out, image_width*sizeof(vbx_uword_t) );
// Set bottom rows to 0
out = output + image_width*(image_height-1-y);
vbx_dma_to_host( out, v_row_out, image_width*sizeof(vbx_uword_t) );
}
// Calculate edges
for (y = 0; y < image_height-(W-1); y++) {
vbx_set_vl(image_width);
// Transfer the next input row while processing
rp_fetch(&v_row_db);
v_row_in = rp_get_buffer(&v_row_db,0);
// Convert aRGB input to luma
vbw_rgb2luma( vW, v_row_in, vT, image_width);
vbx( SVHBU, VSHR, v_luma_bot, 8, vW ); // convert to byte
vbx_sp_push();
image_width=image_width/2;
vbx_set_vl(image_width);
v[0][0] = v00 = (vbx_ubyte_t*)vbx_sp_malloc( 25 * image_width*sizeof(vbx_ubyte_t));
v[0][1] = v01 = v00 + 1 * image_width*sizeof(vbx_ubyte_t) ;
v[0][2] = v02 = v00 + 2 * image_width*sizeof(vbx_ubyte_t) ;
v[1][0] = v10 = v00 + 3 * image_width*sizeof(vbx_ubyte_t) ;
v[1][1] = v11 = v00 + 4 * image_width*sizeof(vbx_ubyte_t) ;
v[1][2] = v12 = v00 + 5 * image_width*sizeof(vbx_ubyte_t) ;
v[2][0] = v20 = v00 + 6 * image_width*sizeof(vbx_ubyte_t) ;
v[2][1] = v21 = v00 + 7 * image_width*sizeof(vbx_ubyte_t) ;
v[2][2] = v22 = v00 + 8 * image_width*sizeof(vbx_ubyte_t) ;
#if W==5
v[0][3] = v03 = v00 + 9 * image_width*sizeof(vbx_ubyte_t) ;
v[0][4] = v04 = v00 + 10 * image_width*sizeof(vbx_ubyte_t) ;
v[1][3] = v13 = v00 + 11 * image_width*sizeof(vbx_ubyte_t) ;
v[1][4] = v14 = v00 + 12 * image_width*sizeof(vbx_ubyte_t) ;
v[2][3] = v23 = v00 + 13 * image_width*sizeof(vbx_ubyte_t) ;
v[2][4] = v24 = v00 + 14 * image_width*sizeof(vbx_ubyte_t) ;
v[3][0] = v30 = v00 + 15 * image_width*sizeof(vbx_ubyte_t) ;
v[3][1] = v31 = v00 + 16 * image_width*sizeof(vbx_ubyte_t) ;
v[3][2] = v32 = v00 + 17 * image_width*sizeof(vbx_ubyte_t) ;
v[3][3] = v33 = v00 + 18 * image_width*sizeof(vbx_ubyte_t) ;
v[3][4] = v34 = v00 + 19 * image_width*sizeof(vbx_ubyte_t) ;
v[4][0] = v40 = v00 + 20 * image_width*sizeof(vbx_ubyte_t) ;
v[4][1] = v41 = v00 + 22 * image_width*sizeof(vbx_ubyte_t) ;
v[4][2] = v42 = v00 + 22 * image_width*sizeof(vbx_ubyte_t) ;
v[4][3] = v43 = v00 + 23 * image_width*sizeof(vbx_ubyte_t) ;
v[4][4] = v44 = v00 + 24 * image_width*sizeof(vbx_ubyte_t) ;
#endif
if(v00==NULL){
printf("mem alloc failed\n"); fflush(stdout);
vbx_sp_pop();
vbx_sp_pop();
return VBW_ERROR_SP_ALLOC_FAILED;
}
//FIXME -- how to manage row buffers with 5 rows? 3 rows are shown below:
#if W==3
for( xx=0; xx<W; xx++ ) v_src[0][xx] = v_luma_top+xx;
for( xx=0; xx<W; xx++ ) v_src[1][xx] = v_luma_mid+xx;
for( xx=0; xx<W; xx++ ) v_src[2][xx] = v_luma_bot+xx;
#else
for( xx=0; xx<W; xx++ ) v_src[0][xx] = v_luma_top+xx;
for( xx=0; xx<W; xx++ ) v_src[1][xx] = v_luma_hii+xx;
for( xx=0; xx<W; xx++ ) v_src[2][xx] = v_luma_mid+xx;
for( xx=0; xx<W; xx++ ) v_src[3][xx] = v_luma_low+xx;
for( xx=0; xx<W; xx++ ) v_src[4][xx] = v_luma_bot+xx;
#endif
vbx_set_vl( image_width - W + 1 );
// compute error (absdiff) in pixel colour with neighbours
for( yy=0; yy<W; yy++ ) {
for( xx=0; xx<W; xx++ ) {
vbx( VVBU, VABSDIFF, v[yy][xx], v_luma_mid+(W/2), v_src[yy][xx] );
}
}
// v[][] holds the errors (differences) between pixels
// efficiently compute a function that looks approximately something like exp(-x):
// large value for small errors, small value for big errors
for( yy=0; yy<W; yy++ ) {
for( xx=0; xx<W; xx++ ) {
vbx( SVBU, VABSDIFF, v[yy][xx], 255, v[yy][xx] ); // 255 - img_err
// 11 or more iterations is mathematically equivalent to a pure gaussian blur // FIXME is this true?
#define NUM_SHARPEN_ITERATIONS 3 // 0 to 10 iterations, practical max is 7 or 8
for( sharp=0; sharp < NUM_SHARPEN_ITERATIONS; sharp++ ) {
vbx( VVBU, VMULHI, v[yy][xx], v[yy][xx], v[yy][xx] ); // v*v;
}
}
}
// with right decimal place, could do the next two instructions using MULFXP and do as BYTES
// convolve errors with gaussian blur kernel
for( yy=0; yy<W; yy++ ) {
for( xx=0; xx<W; xx++ ) {
vbx( SVBU, VMULHI, v[yy][xx], gauss[yy][xx], v[yy][xx] );
}
}
// sum up the weights for normalization later
vbx( VVBHU, VADD, vW, v[0][0], v[0][1] );
vbx( VVBHU, VADD, vT, v[0][2], v[1][0] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VADD, vT, v[1][1], v[1][2] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VADD, vT, v[2][0], v[2][1] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VMOV, vT, v[2][2], 0 );
vbx( VVHU, VADD, vW, vW, vT );
#if (W==5)
vbx( VVBHU, VADD, vT, v[3][0], v[3][1] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VADD, vT, v[3][2], v[4][0] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VADD, vT, v[4][1], v[4][2] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VMOV, vT, v[0][3], v[0][4] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VMOV, vT, v[1][3], v[1][4] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VMOV, vT, v[2][3], v[2][4] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VMOV, vT, v[3][3], v[3][4] );
vbx( VVHU, VADD, vW, vW, vT );
vbx( VVBHU, VMOV, vT, v[4][3], v[4][4] );
vbx( VVHU, VADD, vW, vW, vT );
#endif
// convolve image with new weights
for( yy=0; yy<W; yy++ ) {
for( xx=0; xx<W; xx++ ) {
vbx( VVBU, VMULHI, v[yy][xx], v_src[yy][xx], v[yy][xx] );
//vbx( SVBU, VMULHI, v[yy][xx], gauss[yy][xx], v_src[yy][xx] );
//vbx( SVBU, VMUL , v[yy][xx], 1 , v_src[yy][xx] );
}
}
// sum up the weighted pixels
vbx( VVBHU, VADD, vI, v[0][0], v[0][1] );
vbx( VVBHU, VADD, vT, v[0][2], v[1][0] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VADD, vT, v[1][1], v[1][2] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VADD, vT, v[2][0], v[2][1] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VMOV, vT, v[2][2], 0 );
vbx( VVHU, VADD, vI, vI, vT );
#if (W==5)
vbx( VVBHU, VADD, vT, v[3][0], v[3][1] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VADD, vT, v[3][2], v[4][0] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VADD, vT, v[4][1], v[4][2] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VMOV, vT, v[0][3], v[0][4] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VMOV, vT, v[1][3], v[1][4] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VMOV, vT, v[2][3], v[2][4] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VMOV, vT, v[3][3], v[3][4] );
vbx( VVHU, VADD, vI, vI, vT );
vbx( VVBHU, VMOV, vT, v[4][3], v[4][4] );
vbx( VVHU, VADD, vI, vI, vT );
#endif
// keep RHS of image as original grayscale
image_width=image_width*2;
vbx_set_vl( image_width/2 );
//vbx( VVWHU, VMOV, vT+image_width/2, (v_row_in ) + image_width/2+1, 0 );
vbx( VVBHU, VMOV, vT+image_width/2, (v_src[ 0 ][ 0 ]) + image_width/2+1, 0 );
vbx_sp_pop(); // don't need v[][] data any more
// compute LHS of image
#if 0
vbx( VVBHU, VMOV, vT, v_src[2][2], 0 );
//vbx( SVHU, VSHR, vI, 3, vI );
//vbx( SVHU, VSHR, vW, 3, vW );
//vbx( VVHU, VMUL, vT, vI, vW );
//vbx( SVHU, VSHR, vT, 8, vT );
#else
uint32_t h = image_width/2;
vbx( SVHU, VADD, vW, 0x80, vW ); // round
vbx( SVHU, VSHR, vW, 8, vW );
vbw_vec_divide_uhalf( vT , vI , vW , h );
//vbw_vec_divide_uhalf( vT+h, vI+h, vW+h, image_width-W+1-h );
#endif
// ensure LHS doesn't overflow
vbx( SVHU, VAND, vT, 0xff, vT );
// Copy the result to the low byte of the output row
// Trick to copy the low byte (b) to the middle two bytes as well
// Note that first and last columns are 0
vbx_set_vl(image_width-W+1);
vbx(SVHWU, VMULLO, v_row_out+W/2, 0x00010101, vT);
// blank out left and right edges
// then DMA the result to the output
vbx_set_vl(W/2);
vbx(SVWU, VMOV, v_row_out, COLOUR, 0 );
vbx(SVWU, VMOV, v_row_out + image_width - (W/2), COLOUR, 0 );
vbx_dma_to_host( output+(y+1)*image_pitch, v_row_out, image_width*sizeof(vbx_uword_t) );
// Rotate luma buffers
vbx_ubyte_t *tmp_ptr;
tmp_ptr = v_luma_top;
#if W==3
v_luma_top = v_luma_mid;
v_luma_mid = v_luma_bot;
v_luma_bot = tmp_ptr;
#else
v_luma_top = v_luma_hii;
v_luma_hii = v_luma_mid;
v_luma_mid = v_luma_low;
v_luma_low = v_luma_bot;
v_luma_bot = tmp_ptr;
#endif
}
vbx_sync();
vbx_sp_pop();
return VBW_SUCCESS;
}
/** Luma Edge Detection.
*
* @brief 3x3 Sobel edge detection with 8-bit luma image
*
* @param[out] output 32-bit aRGB edge-intensity output
* @param[in] input 8-bit luma input
* @param[in] image_width Image width in pixels
* @param[in] image_height Image height in pixels
* @param[in] image_pitch Distance in pixels between the start of subsequent rows. usually equal to image_width
* @param[in] renorm Number of bits to shift the final intensity by to the right
* @returns Negative on error condition. See vbw_exit_codes.h
*/
int vbw_sobel_luma8_3x3(unsigned *output, unsigned char *input, const short image_width, const short image_height, const short image_pitch, const short renorm)
{
int y;
vbx_ubyte_t *v_luma_top, *v_luma_mid, *v_luma_bot;
vbx_uword_t *v_row_out;
vbx_uhalf_t *v_sobel_row_top, *v_sobel_row_mid, *v_sobel_row_bot;
vbx_uhalf_t *v_gradient_x, *v_gradient_y;
vbx_uhalf_t *v_tmp;
void *tmp_ptr;
vbx_sp_push();
// Allocate space in scratchpad for vectors
rotating_prefetcher_t v_luma=rotating_prefetcher(3,image_width*sizeof(vbx_ubyte_t),
input,input+image_height*image_pitch,
image_pitch*sizeof(vbx_ubyte_t));
v_sobel_row_top = (vbx_uhalf_t *)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_sobel_row_mid = (vbx_uhalf_t *)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_sobel_row_bot = (vbx_uhalf_t *)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_gradient_x = (vbx_uhalf_t *)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_gradient_y = (vbx_uhalf_t *)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_row_out = (vbx_uword_t *)vbx_sp_malloc(image_width*sizeof(vbx_uword_t));
if(v_row_out==NULL) {
vbx_sp_pop();
return VBW_ERROR_SP_ALLOC_FAILED;
}
// Transfer the first 3 input rows and interleave first 2 sobel row calculations
rp_fetch(&v_luma);
rp_fetch(&v_luma);
v_luma_top=rp_get_buffer(&v_luma, 0);
vbw_sobel_3x3_row(v_sobel_row_top, v_luma_top,image_width);
rp_fetch(&v_luma);
v_luma_mid=rp_get_buffer(&v_luma, 1);
vbw_sobel_3x3_row(v_sobel_row_mid, v_luma_mid, image_width);
// Set top output row to 0
vbx_set_vl(image_width);
vbx(SVWU, VMOV, v_row_out, 0, 0);
vbx_dma_to_host(output, v_row_out, image_width*sizeof(vbx_uword_t));
// Calculate edges
for (y = 0; y < image_height-(FILTER_HEIGHT-1); y++) {
// Transfer the next input row while processing
rp_fetch(&v_luma);
v_luma_top=rp_get_buffer(&v_luma,0);
v_luma_mid=rp_get_buffer(&v_luma,1);
v_luma_bot=rp_get_buffer(&v_luma,2);
// Start calculating gradient_x
vbx_set_vl(image_width);
vbx(SVBHU, VSHL, v_gradient_x, 1, v_luma_mid); // multiply by 2
// Calculate gradient_y
// Apply [1 2 1] matrix to last row in window and calculate absolute difference with pre-computed first row
vbw_sobel_3x3_row(v_sobel_row_bot, v_luma_bot, image_width);
vbx(VVH, VABSDIFF, (vbx_half_t*)v_gradient_y, (vbx_half_t*)v_sobel_row_top, (vbx_half_t*)v_sobel_row_bot);
// Re-use v_sobel_row_top
v_tmp = v_sobel_row_top;
// Finish calculating gradient_x
// Apply [1 2 1]T matrix to all columns
vbx_set_vl(image_width);
vbx(VVBHU, VADD, v_tmp, v_luma_top, v_luma_bot);
vbx(VVHU, VADD, v_tmp, v_tmp, v_gradient_x);
// For each column, calculate absolute difference with 2nd column to the right
vbx_set_vl(image_width-2);
vbx(VVH, VABSDIFF, (vbx_half_t*)v_gradient_x, (vbx_half_t*)v_tmp, (vbx_half_t*)v_tmp+2);
// sum of absoute gradients
//vbx_set_vl(image_width-2);
vbx(VVHU, VADD, v_tmp, v_gradient_x, v_gradient_y);
vbx(SVHU, VSHR, v_tmp, renorm, v_tmp);
// Threshold
vbx(SVHU, VSUB, v_gradient_y, 255, v_tmp);
vbx(SVHU, VCMV_LTZ, v_tmp, 255, v_gradient_y);
// Copy the result to the low byte of the output row
// Trick to copy the low byte (b) to the middle two bytes as well
// Note that first and last columns are 0
//vbx_set_vl(image_width-2);
vbx(SVHWU, VMULLO, v_row_out+1, 0x00010101, v_tmp);
// DMA the result to the output
vbx_dma_to_host(output+(y+1)*image_pitch, v_row_out, image_width*sizeof(vbx_uword_t));
// Rotate v_sobel_row buffers (for gradient_y)
tmp_ptr = (void *)v_sobel_row_top;
v_sobel_row_top = v_sobel_row_mid;
v_sobel_row_mid = v_sobel_row_bot;
v_sobel_row_bot = (vbx_uhalf_t *)tmp_ptr;
}
// Set bottom row to 0
vbx_set_vl(image_width);
vbx(SVWU, VMOV, v_row_out, 0, 0);
vbx_dma_to_host(output+(image_height-1)*image_pitch, v_row_out, image_width*sizeof(vbx_uword_t));
vbx_sync();
vbx_sp_pop();
return VBW_SUCCESS;
}
int vbw_vec_reverse( vbx_sp_t *v_dst, vbx_sp_t *v_src, const unsigned int N )
{
const int VBW_ROT16= sizeof(vbx_sp_t) <=sizeof(vbx_half_t);
const int VBW_ROT8= sizeof(vbx_sp_t)== sizeof(vbx_byte_t);
const int VBW_RSHIFT_T_TO_W= (sizeof(vbx_sp_t)==sizeof(vbx_word_t)? 0:
sizeof(vbx_sp_t)==sizeof(vbx_half_t)? 1:/*byte_sized*/2);
const int VBW_LSHIFT_W_TO_T= VBW_RSHIFT_T_TO_W;
vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
const unsigned int NUM_LANES = this_mxp->vector_lanes;
//printf("\n%d\n",VBX_SKIP_ALL_CHECKS);
// Can the whole vector fit in the scratchpad width?
if( N < (NUM_LANES << VBW_LSHIFT_W_TO_T) ){
vbx_set_vl( 1 );
vbx_set_2D( N, (int)-sizeof(vbx_sp_t), (int)sizeof(vbx_sp_t), 0 );
vbxx_2D(VMOV, v_dst+N-1, v_src);
return VBW_SUCCESS;
}
unsigned int threshold_w = (NUM_LANES >= 32 ? VL1_THRESHOLD_V32_UP :
NUM_LANES == 16 ? VL1_THRESHOLD_V16 :
NUM_LANES == 8 ? VL1_THRESHOLD_V8 : UINT_MAX);
unsigned int N_w = N >> VBW_RSHIFT_T_TO_W; // Equivalent number of words in the vector
if( N_w && N_w <= threshold_w ) {
if( VBW_ROT16){
// remainder of elements that can't add to a whole word
unsigned int stub_t = N - (N_w << VBW_LSHIFT_W_TO_T);
if( stub_t ) {
vbx_set_vl( 1 );
vbx_set_2D( stub_t, (int)-sizeof(vbx_sp_t), sizeof(vbx_sp_t), 0 );
vbxx_2D(VMOV, v_dst+stub_t-1, v_src+N-stub_t);
v_dst += stub_t;
}
vec_rev_rot16_w(v_dst, v_src, N_w);
}else{
vec_rev_w(v_dst, v_src, N_w);
}
if( VBW_ROT8){
vec_rot8_h(v_dst, v_dst, N_w*2);
}
return VBW_SUCCESS;
}
const unsigned int SP_WIDTH_B = this_mxp->scratchpad_alignment_bytes;
const unsigned int FREE_BYTES = vbx_sp_getfree();
const unsigned int ODD_LOG_SEL = NUM_LANES & 0x55555555 ? 1 : 0;
vbx_word_t *v_mask, *v_result;
vbx_word_t *v_scratch[2] = {0,0};
unsigned int num_rows_w = N_w / NUM_LANES;
unsigned int working_set_w = num_rows_w * NUM_LANES;
unsigned int tail_t = N - (working_set_w << VBW_LSHIFT_W_TO_T);
unsigned int remaining_w = working_set_w;
if( tail_t ) {
vbx_set_vl( 1 );
vbx_set_2D( tail_t, (int)-sizeof(vbx_sp_t), sizeof(vbx_sp_t), 0 );
vbxx_2D(VMOV, v_dst+tail_t-1, v_src+N-tail_t);
v_dst += tail_t;
}
vbx_word_t *v_src_w = (vbx_word_t *)v_src;
vbx_word_t *v_dst_w = (vbx_word_t *)v_dst;
if(!num_rows_w) {
return VBW_SUCCESS;
}
remaining_w = working_set_w;
while( remaining_w*sizeof(vbx_word_t) + SP_WIDTH_B > FREE_BYTES ) {
if( remaining_w <= threshold_w*2 ) {
if( VBW_ROT16){
vec_rev_rot16_w(v_dst_w, v_src_w, remaining_w);
}else{
vec_rev_w(v_dst_w, v_src_w, remaining_w);
}
if( VBW_ROT8){
vec_rot8_h(v_dst_w, v_dst_w, remaining_w*2);
}
return VBW_SUCCESS;
}
working_set_w = VBX_PAD_DN( (remaining_w - NUM_LANES)/2, NUM_LANES );
v_mask = v_dst_w + (working_set_w*2);
remaining_w -= working_set_w;
v_scratch[0] = v_dst_w;
v_scratch[1] = v_dst_w + working_set_w;
num_rows_w = working_set_w / NUM_LANES;
v_result = vec_rev_merge_w( v_scratch[ODD_LOG_SEL], v_src_w + remaining_w, working_set_w,
v_scratch[!ODD_LOG_SEL], v_mask, SP_WIDTH_B, num_rows_w, VBW_ROT16 );
#if !VBX_SKIP_ALL_CHECKS
if( v_result != v_dst_w ) {
VBX_PRINTF("Unexpected behavior: merge reverse returned the wrong vector. Parameter order was chosen based on NUM_LANES.");
VBX_EXIT(-1);
}
#endif
if( VBW_ROT8){
vec_rot8_h(v_result, v_result, working_set_w*2);
}
v_dst_w += working_set_w;
}
vbx_sp_push();
v_scratch[0] = v_dst_w;
v_scratch[1] = (vbx_word_t*)vbx_sp_malloc( remaining_w * sizeof(vbx_word_t) );
#if !VBX_SKIP_ALL_CHECKS
if( !v_scratch[1] ) {
VBX_PRINTF("vbx_sp_malloc failed when it was predetermined to have enough space.");
VBX_EXIT(-1);
}
#endif
v_mask = (vbx_word_t*)vbx_sp_malloc( SP_WIDTH_B );
#if !VBX_SKIP_ALL_CHECKS
if( !v_mask ) {
VBX_PRINTF("vbx_sp_malloc failed when it was predetermined to have enough space.");
VBX_EXIT(-1);
}
#endif
num_rows_w = remaining_w / NUM_LANES;
v_result = vec_rev_merge_w( v_scratch[ODD_LOG_SEL], v_src_w, remaining_w, v_scratch[!ODD_LOG_SEL],
v_mask, SP_WIDTH_B, num_rows_w, VBW_ROT16 );
#if !VBX_SKIP_ALL_CHECKS
if( v_result != v_dst_w ) {
VBX_PRINTF("Unexpected behavior: merge reverse returned the wrong vector. Parameter order was chosen based on NUM_LANES.");
VBX_EXIT(-1);
}
#endif
if( VBW_ROT8){
vec_rot8_h(v_result, v_result, remaining_w*2);
}
vbx_sp_pop();
return VBW_SUCCESS;
}
void vbx_mtx_fdct( vbx_mtx_fdct_t *v, dt *block_v, dt *image,
int start_x, int start_y, int end_x, int end_y,int num_tile_x, int num_tile_y )
{
// vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
// const int VBX_SCRATCHPAD_SIZE = this_mxp->scratchpad_size;
const int BIG_TILE_SIZE = num_tile_x * num_tile_y * DCT_SIZE;
int next_x=start_x+1;
int next_y=start_y;
int get_next=1;
if( start_x == end_x && start_y == end_y ) {
get_next=0;
}
if( start_x == end_x ) {
next_x = 0;
next_y++;
}
const vbx_half_t *vimageDMA = v->vimage[!v->db]; // dma
// const vbx_half_t *vblockDMA = v->vblock[!v->db]; // dma // never used directly
const vbx_half_t *vimageVPU = v->vimage[ v->db]; // active
const vbx_half_t *vblockVPU = v->vblock[ v->db]; // active
const vbx_half_t *vblockTMP = v->vblock[ 2 ]; // temp
const vbx_half_t *vcoeff = v->vcoeff;
const vbx_half_t *vprods = v->vprods;
const vbx_half_t *vaccum = v->vaccum;
const vbx_half_t *vflags = v->vflags;
#if DMA
// First, prefetch the next chunk of the next image for a future call to fdct_tile()
#if NUM_TILE_Y > 1
if( get_next ) // get row 0
getBigTileImageY( vimageDMA,
image+next_x*NUM_TILE_X*BLOCK_SIZE+next_y*IMAGE_WIDTH*NUM_TILE_Y*BLOCK_SIZE, 0 );
#else
if( get_next ) // get row 0
getBigTileImage( vimageDMA,
image+next_x*NUM_TILE_X*BLOCK_SIZE+next_y*IMAGE_WIDTH*NUM_TILE_Y*BLOCK_SIZE, 0 );
#endif
#endif
int r;
for( r=0; r < BLOCK_SIZE; r++ ) {
// perform multiply of the whole BIG_TILE with row 'r' of the image matrix -- before had dct matrix switching
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE ); // for the length of tiled rows
vbx_set_2D( BLOCK_SIZE, NUM_TILE_X*BLOCK_SIZE*sizeof(dt), 0, NUM_TILE_X*BLOCK_SIZE*sizeof(dt) ); // for all rows of tiled coeffiencents
vbx_set_3D( NUM_TILE_Y, NUM_TILE_X * DCT_SIZE*sizeof(dt), NUM_TILE_X * DCT_SIZE*sizeof(dt), 0 ); // for all groups Y
vbx_3D( VVH, VMUL, vprods, vimageVPU + r*NUM_TILE_X*BLOCK_SIZE, vcoeff); // for all 'columns' of tiled data
#if ACCUMULATE
// accumulate the multiply operations
#if 0 & USE_ACCUM_FLAGS
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE * NUM_TILE_Y * BLOCK_SIZE - (BLOCK_SIZE-1) );
vbx( VVH, VADD, vaccum, vprods+0, vprods+1 );
vbx_set_2D( BLOCK_SIZE-2, 0, 0, sizeof(dt) );
vbx_2D( VVH, VADD, vaccum, vaccum, vprods+2 );
vbx( VVH, VCMV_Z, vblockTMP+r, vaccum, vflags );
#elif BLOCK4
//case DCT 4
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE * NUM_TILE_Y * BLOCK_SIZE - (BLOCK_SIZE-1) );
vbx( VVH, VADD, vaccum, vprods, vprods+1 );
vbx( VVH, VADD, vaccum, vaccum, vprods+2 );
vbx( VVH, VADD, vaccum, vaccum, vprods+3 );
vbx( VVH, VCMV_Z, vblockTMP+r, vaccum, vflags );
#else
//correct?
vbx_set_vl( BLOCK_SIZE );
vbx_set_2D( BLOCK_SIZE, NUM_TILE_X*BLOCK_SIZE*sizeof(dt), NUM_TILE_X*BLOCK_SIZE*sizeof(dt), NUM_TILE_X*BLOCK_SIZE*sizeof(dt) );
vbx_set_3D( NUM_TILE_X, BLOCK_SIZE*sizeof(dt), BLOCK_SIZE*sizeof(dt), BLOCK_SIZE*sizeof(dt) );
#if NUM_TILE_Y == 1
vbx_acc_3D( VVH, VOR, vblockTMP + r, vprods , vprods );
#else
int y;
for (y=0; y< NUM_TILE_Y; y++){
vbx_acc_3D( VVH, VOR, vblockTMP + r + y*NUM_TILE_X*DCT_SIZE, vprods+ y*NUM_TILE_X*DCT_SIZE, vprods+ y*NUM_TILE_X*DCT_SIZE );
}
#endif
#endif
#endif
#if 0
// dont do DMA READS here yet. a DMA WRITE may still be in progress, give it chance to finish
#if DMA
// every other iteration, prefetch the next row of the next image
// NB: with 2D DMA, we could issue this as a single DMA request at the top of the file
// instead, we must intersperse these 1D DMA strips to ensure they don't block the instruction queue
#if NUM_TILE_Y > 1
if( !(r&1) && get_next )
getBigTileImageY( vimageDMA,
image+next_x*NUM_TILE_X*BLOCK_SIZE+next_y*IMAGE_WIDTH*NUM_TILE_Y*BLOCK_SIZE,
(1+((r-1)>>1)) ); //BLOCK_SIZE/2 rows added
#else
if( !(r&1) && get_next )
getBigTileImage( vimageDMA,
image+next_x*NUM_TILE_X*BLOCK_SIZE+next_y*IMAGE_WIDTH*NUM_TILE_Y*BLOCK_SIZE,
(1+((r-1)>>1)) ); //BLOCK_SIZE/2 rows added
#endif
#endif
#endif
}
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE * NUM_TILE_Y * BLOCK_SIZE );
vbx( SVH, VSHR, vblockTMP, SHIFT_AMOUNT, vblockTMP );
// now do the transposed version
for( r=0; r < BLOCK_SIZE; r++ ) {
// perform multiply of the whole BIG_TILE with row 'r' of the image matrix -- before had dct matrix switching
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE ); // for the length of tiled rows
vbx_set_2D( BLOCK_SIZE, NUM_TILE_X * BLOCK_SIZE*sizeof(dt), NUM_TILE_X * BLOCK_SIZE*sizeof(dt), 0 ); // for all 'columns' of tiled data
vbx_set_3D( NUM_TILE_Y, NUM_TILE_X * DCT_SIZE*sizeof(dt), NUM_TILE_X * DCT_SIZE*sizeof(dt), 0 ); // for all groups Y
vbx_3D( VVH, VMUL, vprods, vblockTMP, vcoeff + r*NUM_TILE_X*BLOCK_SIZE); // for all rows of tiled coeffients
#if ACCUMULATE
// accumulate the multiply operations
#if 0 & USE_ACCUM_FLAGS
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE * NUM_TILE_Y * BLOCK_SIZE - (BLOCK_SIZE-1) );
vbx( VVH, VADD, vaccum, vprods+0, vprods+1 );
vbx_set_2D( BLOCK_SIZE-2, 0, 0, sizeof(dt) );
vbx_2D( VVH, VADD, vaccum, vaccum, vprods+2 );
vbx( VVH, VCMV_Z, vblockVPU+r, vaccum, vflags );
#elif BLOCK4
//case DCT 4
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE * NUM_TILE_Y * BLOCK_SIZE - (BLOCK_SIZE-1) );
vbx( VVH, VADD, vaccum, vprods, vprods+1 );
vbx( VVH, VADD, vaccum, vaccum, vprods+2 );
vbx( VVH, VADD, vaccum, vaccum, vprods+3 );
//vbx( VVH, VCMV_Z, vblockVPU+r, vaccum, vflags );
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE - (BLOCK_SIZE-1) ); // for the length of a tiled row
vbx_set_2D( BLOCK_SIZE, 1*sizeof(dt), NUM_TILE_X*BLOCK_SIZE*sizeof(dt), 0);// for all tiled rows
#if NUM_TILE_Y == 1
vbx_2D(VVH, VCMV_Z, vblockVPU+r*NUM_TILE_X*BLOCK_SIZE, vaccum, vflags ); //
#else
int y;
for (y=0; y< NUM_TILE_Y; y++){
vbx_2D(VVH, VCMV_Z, vblockVPU+r*NUM_TILE_X*BLOCK_SIZE + y*NUM_TILE_X*DCT_SIZE , vaccum+y*NUM_TILE_X*DCT_SIZE, vflags ); //
}
#endif
#else
//correct?
vbx_set_vl( BLOCK_SIZE ); // for the length of a row
vbx_set_2D( BLOCK_SIZE, sizeof(dt), NUM_TILE_X*BLOCK_SIZE*sizeof(dt), NUM_TILE_X*BLOCK_SIZE*sizeof(dt) ); // for all rows in that block
vbx_set_3D( NUM_TILE_X, BLOCK_SIZE*sizeof(dt), BLOCK_SIZE*sizeof(dt), BLOCK_SIZE*sizeof(dt) ); // for all tiled blocks horizontally(x)
#if NUM_TILE_Y == 1
vbx_acc_3D( VVH, VOR, vblockVPU + r*NUM_TILE_X*BLOCK_SIZE , vprods , vprods );
#else
int y;
for (y=0; y< NUM_TILE_Y; y++){
vbx_acc_3D( VVH, VOR, vblockVPU + r*NUM_TILE_X*BLOCK_SIZE + y*NUM_TILE_X*DCT_SIZE, vprods+ y*NUM_TILE_X*DCT_SIZE, vprods+ y*NUM_TILE_X*DCT_SIZE );
}
#endif
#endif
#endif
#if DMA
// every other iteration, prefetch the next row of the next image
// NB: with 2D DMA, we could issue this as a single DMA request at the top of the file
// instead, we must intersperse these 1D DMA strips to ensure they don't block the instruction queue
#if NUM_TILE_Y > 1
//if( !(r&1) && r<(BLOCK_SIZE-1) && get_next )
if( get_next )
getBigTileImageY(
vimageDMA,
image+next_x*NUM_TILE_X*BLOCK_SIZE+next_y*IMAGE_WIDTH*NUM_TILE_Y*BLOCK_SIZE,
r );
//(BLOCK_SIZE/2 +1+((r-1)>>1)) ); // BLOCK/2 -1 rows
#else
//if( !(r&1) && r<(BLOCK_SIZE-1) && get_next )
if( get_next )
getBigTileImage( vimageDMA,
image+next_x*NUM_TILE_X*BLOCK_SIZE+next_y*IMAGE_WIDTH*NUM_TILE_Y*BLOCK_SIZE,
r );
//(BLOCK_SIZE/2 +1+((r-1)>>1)) ); // BLOCK/2 -1 rows
#endif
#endif
}
vbx_set_vl( NUM_TILE_X * BLOCK_SIZE * NUM_TILE_Y * BLOCK_SIZE );
vbx( SVH, VSHR, vblockVPU, SHIFT_AMOUNT, vblockVPU );
#if DMA2
// Write result back to memory as one big block
vbx_dma_to_host( block_v, vblockVPU, BIG_TILE_SIZE*sizeof(dt) );
#endif
v->db = !v->db;
#ifdef DEBUG
{
vbx_sync();
int i,j;
printf("%d\n", !db);
for(i=0;i<BLOCK_SIZE*NUM_TILE_Y;i++){
for(j=0;j<BLOCK_SIZE*NUM_TILE_X;j++){
printf(" %4d", block_v[i*BLOCK_SIZE*NUM_TILE_X+j]);
}
printf("\n");
}
}
#endif
}
/** Luma Edge Detection
*
* @brief 3x3 Sobel edge detection with 32-bit aRGB image
*
* @param[out] output 32-bit aRGB edge-intensity output
* @param[in] input 32-bit aRGB input
* @param[in] image_width Image width in pixels
* @param[in] image_height Image height in pixels
* @param[in] image_pitch Distance in pixels between the start of subsequent rows. usually equal to image_width
* @param[in] renorm Number of bits to shift the final intensity by to the right
* @returns Negative on error condition. See vbw_exit_codes.h
*/
int vbw_sobel_argb32_3x3_partial(unsigned *output, unsigned *input, const short image_width, const short image_height, const short image_pitch, const short renorm)
{
int y;
vbx_uword_t *v_row_in;
vbx_uhalf_t *v_luma_top, *v_luma_mid, *v_luma_bot;
vbx_uword_t *v_row_out;
vbx_uhalf_t *v_sobel_row_top, *v_sobel_row_mid, *v_sobel_row_bot;
vbx_uhalf_t *v_gradient_x, *v_gradient_y;
vbx_uhalf_t *v_tmp;
void *tmp_ptr;
vbx_sp_push();
// Allocate space in scratchpad for vectors
struct rotating_prefetcher_t v_row_db=rotating_prefetcher(1,image_width*sizeof(vbx_uword_t),
input,input+image_pitch*image_width,
image_pitch*sizeof(vbx_uword_t));
v_luma_top = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_luma_mid = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_luma_bot = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_sobel_row_top = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_sobel_row_mid = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_sobel_row_bot = (vbx_uhalf_t*)vbx_sp_malloc(image_width*sizeof(vbx_uhalf_t));
v_row_out = (vbx_uword_t*)vbx_sp_malloc(image_width*sizeof(vbx_uword_t));
if(v_row_out==NULL){
vbx_sp_pop();
return VBW_ERROR_SP_ALLOC_FAILED;
}
// Re-use v_sobel_row_bot as v_tmp
v_tmp = v_sobel_row_bot;
// Transfer the first 3 input rows and interleave first 2 rgb2luma and first 2 sobel row calculations
rp_fetch(&v_row_db);
rp_fetch(&v_row_db);
v_row_in=rp_get_buffer(&v_row_db,0);
vbw_rgb2luma(v_luma_top, v_row_in, v_tmp, image_width); // 1st luma row
vbw_sobel_3x3_row(v_sobel_row_top, v_luma_top, image_width); // 1st partial sobel row
rp_fetch(&v_row_db);
v_row_in=rp_get_buffer(&v_row_db,0);
vbw_rgb2luma(v_luma_mid, v_row_in, v_tmp, image_width); // 2nd luma row
vbw_sobel_3x3_row(v_sobel_row_mid, v_luma_mid, image_width); // 2nd partial sobel row
// Set top output row to 0
vbx_set_vl(image_width);
vbx(SVWU, VMOV, v_row_out, 0, 0);
vbx_dma_to_host(output, v_row_out, image_width*sizeof(vbx_uword_t));
// Calculate edges
for (y = 0; y < image_height-(FILTER_HEIGHT-1); y++) {
// Transfer the next input row while processing
rp_fetch(&v_row_db);
v_row_in=rp_get_buffer(&v_row_db,0);
// Re-use v_sobel_row_bot as v_tmp
v_tmp = v_sobel_row_bot;
// Convert aRGB input to luma
vbw_rgb2luma(v_luma_bot, v_row_in, v_tmp, image_width);
// Done with v_row_in; re-use for v_gradient_x and v_gradient_y (be careful!)
v_gradient_x = (vbx_uhalf_t *)v_row_in;
v_gradient_y = (vbx_uhalf_t *)v_row_in + image_width;
// Calculate gradient_x
// Apply [1 2 1]T matrix to all columns
vbx_set_vl(image_width);
vbx(SVHU, VSHL, v_gradient_x, 1, v_luma_mid); // multiply by 2
vbx(VVHU, VADD, v_tmp, v_luma_top, v_luma_bot);
vbx(VVHU, VADD, v_tmp, v_tmp, v_gradient_x);
// For each column, calculate absolute difference with 2nd column to the right
vbx_set_vl(image_width-2);
vbx(VVH, VABSDIFF, (vbx_half_t*)v_gradient_x, (vbx_half_t*)v_tmp, (vbx_half_t*)v_tmp+2);
// Calculate gradient_y
// Apply [1 2 1] matrix to last row in window and calculate absolute difference with pre-computed first row
vbw_sobel_3x3_row(v_sobel_row_bot, v_luma_bot, image_width);
vbx(VVH, VABSDIFF, (vbx_half_t*)v_gradient_y, (vbx_half_t*)v_sobel_row_top, (vbx_half_t*)v_sobel_row_bot);
// Re-use v_sobel_row_top as v_tmp
v_tmp = v_sobel_row_top;
// sum of absoute gradients
vbx_set_vl(image_width-2);
vbx(VVHU, VADD, v_tmp, v_gradient_x, v_gradient_y);
vbx(SVHU, VSHR, v_tmp, renorm, v_tmp);
// Threshold
vbx(SVHU, VSUB, v_gradient_y, 255, v_tmp);
vbx(SVHU, VCMV_LTZ, v_tmp, 255, v_gradient_y);
// Copy the result to the low byte of the output row
// Trick to copy the low byte (b) to the middle two bytes as well
// Note that first and last columns are 0
vbx_set_vl(image_width-2);
vbx(SVHWU, VMULLO, v_row_out+1, 0x00010101, v_tmp);
// DMA the result to the output (minus the outside two pixels
vbx_dma_to_host(output+(y+1)*image_pitch+1, v_row_out+1, (image_width-2)*sizeof(vbx_uword_t));
// Rotate luma buffers
tmp_ptr = (void *)v_luma_top;
v_luma_top = v_luma_mid;
v_luma_mid = v_luma_bot;
v_luma_bot = (vbx_uhalf_t *)tmp_ptr;
// Rotate v_sobel_row buffers (for gradient_y)
tmp_ptr = (void *)v_sobel_row_top;
v_sobel_row_top = v_sobel_row_mid;
v_sobel_row_mid = v_sobel_row_bot;
v_sobel_row_bot = (vbx_uhalf_t *)tmp_ptr;
}
// Set bottom row to 0
vbx_set_vl(image_width);
vbx(SVWU, VMOV, v_row_out, 0, 0);
vbx_dma_to_host(output+(image_height-1)*image_pitch, v_row_out, image_width*sizeof(vbx_uword_t));
vbx_sp_pop();
return VBW_SUCCESS;
}
int vbw_mtx_xp_ext(vbx_mm_t *out, vbx_mm_t *in, const int INROWS, const int INCOLS )
{
typedef vbx_mm_t vbx_sp_t;
int elements = INROWS * INCOLS;
if(elements < SCALAR_THRESHOLD) {
vbx_sync(); //in case we input is waiting on a DMA transfer
int i,j;
for(i = 0; i < INROWS; i++) {
for(j = 0; j < INCOLS; j++) {
out[j*INROWS+i] = in[i*INCOLS+j];
}
}
return VBW_SUCCESS;
}
vbx_sp_push();
vbx_sp_t *v_in;
vbx_sp_t *v_out;
int tile_height = 0;
int tile_width = 0;
int prev_tile_width = 0;
int tile_y = 0;
int tile_x = 0;
vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
int SP_WIDTH_B = this_mxp->scratchpad_alignment_bytes;
int SP_SIZE = vbx_sp_getfree();
int max_sp_elements = vbx_sp_getfree() / sizeof(vbx_sp_t);
int max_tile_elements = VBX_PAD_DN( SP_SIZE/2, SP_WIDTH_B ) / sizeof(vbx_sp_t);
if( INROWS == 1 || INCOLS == 1 ) { // 1D transpose becomes a simple copy operation
if( elements <= max_sp_elements ) { // We can use the whole scratchpad for this
v_in = (vbx_sp_t*)vbx_sp_malloc( elements * sizeof(vbx_sp_t) );
vbx_dma_to_vector( v_in, in, elements*sizeof(vbx_mm_t) );
v_out = v_in;
vbx_dma_to_host( out, v_out, elements*sizeof(vbx_mm_t) );
} else { // To test this, you'll need a very large 1D matrix (or a small SP)
tile_width = max_sp_elements;
v_in = (vbx_sp_t*)vbx_sp_malloc( tile_width * sizeof(vbx_sp_t) );
for (tile_x = 0; tile_x < elements; tile_x += tile_width) {
if( tile_x + tile_width > elements) tile_width = elements - tile_x;
vbx_dma_to_vector( v_in, in + tile_x, tile_width*sizeof(vbx_mm_t) );
v_out = v_in;
vbx_dma_to_host( out+tile_x, v_out, tile_width*sizeof(vbx_mm_t) );
}
}
} else if( elements < max_tile_elements ) { // Matrix is small enough to handle entirely in SP
v_in = (vbx_sp_t*)vbx_sp_malloc( elements * sizeof(vbx_sp_t) );
v_out = (vbx_sp_t*)vbx_sp_malloc( elements * sizeof(vbx_sp_t) );
vbx_dma_to_vector( v_in, in, elements*sizeof(vbx_mm_t) );
vbw_mtx_xp(v_out,v_in,INROWS,INCOLS);
vbx_dma_to_host( out, v_out, elements*sizeof(vbx_mm_t) );
} else { // At this point we know at least one full tile will be needed
#define QUICK_A_LANES_THRESHOLD 8 // Use merge transpose if there are at least this many lanes
#define QUICK_A_TILE_WIDTH 128
#define QUICK_A_TILE_ELEMENTS (QUICK_A_TILE_WIDTH*QUICK_A_TILE_WIDTH)
#define QUICK_A_VF_ELEMENTS (QUICK_A_TILE_ELEMENTS/2)
#define QUICK_A_REQ_ELEMENTS (2*VBX_PAD_UP(QUICK_A_TILE_ELEMENTS,SP_WIDTH_B/sizeof(vbx_sp_t)) + VBX_PAD_UP(QUICK_A_VF_ELEMENTS,sizeof(vbx_sp_t)))
#define QUICK_B_LANES_THRESHOLD 16 // Use smaller merge transpose tile only if there are a lot of lanes
#define QUICK_B_TILE_WIDTH 64 // and only if larger tile A size cannot be used.
#define QUICK_B_TILE_ELEMENTS (QUICK_B_TILE_WIDTH*QUICK_B_TILE_WIDTH)
#define QUICK_B_VF_ELEMENTS (QUICK_B_TILE_ELEMENTS/2)
#define QUICK_B_REQ_ELEMENTS (2*VBX_PAD_UP(QUICK_B_TILE_ELEMENTS,SP_WIDTH_B/sizeof(vbx_sp_t)) + VBX_PAD_UP(QUICK_B_VF_ELEMENTS,sizeof(vbx_sp_t)))
int NUM_LANES = this_mxp->vector_lanes;
int DMA_BYTES = this_mxp->dma_alignment_bytes;
int min_tile_dim = DMA_BYTES / sizeof(vbx_sp_t);
vbx_sp_t *v_out_sel;
vbx_sp_t *vf = 0;
if( NUM_LANES >= QUICK_A_LANES_THRESHOLD // Check for appropriate conditions to use merge transpose tiles
&& INCOLS >= QUICK_A_TILE_WIDTH
&& INROWS >= QUICK_A_TILE_WIDTH
&& (unsigned)max_sp_elements >= QUICK_A_REQ_ELEMENTS ) {
tile_width = tile_height = QUICK_A_TILE_WIDTH;
vf = (vbx_sp_t *)vbx_sp_malloc( QUICK_A_VF_ELEMENTS * sizeof(vbx_sp_t));
} else if( NUM_LANES >= QUICK_B_LANES_THRESHOLD
&& INCOLS >= QUICK_B_TILE_WIDTH
&& INROWS >= QUICK_B_TILE_WIDTH
&& (unsigned)max_sp_elements >= QUICK_B_REQ_ELEMENTS ) {
tile_width = tile_height = QUICK_B_TILE_WIDTH;
vf = (vbx_sp_t *)vbx_sp_malloc( QUICK_B_VF_ELEMENTS * sizeof(vbx_sp_t));
} else {
findTileSize( &tile_height, &tile_width, INROWS, INCOLS, max_tile_elements, min_tile_dim );
}
prev_tile_width = tile_width;
v_in = (vbx_sp_t*)vbx_sp_malloc( tile_height*tile_width * sizeof(vbx_sp_t) );
v_out = (vbx_sp_t*)vbx_sp_malloc( tile_height*tile_width * sizeof(vbx_sp_t) );
if( v_out==NULL ) {
vbx_sp_pop();
return VBW_ERROR_SP_ALLOC_FAILED;
}
vbx_sp_t *v[2] = { v_in, v_out };
tile_y = 0; // Reset y position for new col
while( tile_y < INROWS ) {
vbx_set_2D( tile_width, tile_height*sizeof(vbx_sp_t), sizeof(vbx_sp_t), sizeof(vbx_sp_t) );
vbx_set_3D( tile_height, sizeof(vbx_sp_t), tile_width*sizeof(vbx_sp_t), tile_width*sizeof(vbx_sp_t) );
tile_x = 0; // Reset x position for new row
while( tile_x < INCOLS ) {
vbx_dma_to_vector_2D(
v_in,
in+(tile_y*INCOLS)+tile_x,
tile_width*sizeof(vbx_mm_t),
tile_height,
tile_width*sizeof(vbx_sp_t),
INCOLS*sizeof(vbx_mm_t) );
v_out_sel = v_out; // select v_out as default vector to DMA to MM
/* *** merge transpose (matrix must be square and a power of 2 wide) *** */
if( vf && tile_width == tile_height
&& (tile_width==QUICK_A_TILE_WIDTH || tile_width==QUICK_B_TILE_WIDTH) ) {
int src = 0;
int n;
for( n=1; n<tile_width; n *= 2 ) { // can't do 1st iteration until entire tile is DMA'd in
const int nn = 2*n;
// copy the destination matrix
vbx_set_vl( tile_width*tile_width ); // use v_in & v_out as working matrices (clobber v_in)
vbxx( VMOV, v[!src], v[src]);
// do the work
vbx_set_vl( n*tile_width );
vbxx( VAND, vf, n, (vbx_enum_t*)0 ); // mask for merging: 0101010... then 00110011...
vbx_set_2D( tile_width/nn, nn*tile_width*sizeof(vbx_sp_t), nn*tile_width*sizeof(vbx_sp_t), 0 );
vbxx_2D( VCMV_Z, v[!src]+n*tile_width, v[src]+n , vf );
vbxx_2D( VCMV_Z, v[!src]+n, v[src]+n*tile_width, vf );
src = !src;
}
v_out_sel = v[src]; // depending on the size of the mtx, the final result may be in v_in or v_out
} else {
vbx_set_vl( 1 ); // 2D and 3D will be set by the x and y edge conditions, even using merge
vbxx_3D(VMOV, v_out, v_in );
}
vbx_dma_to_host_2D(
out+(tile_x*INROWS)+tile_y,
v_out_sel,
tile_height*sizeof(vbx_mm_t),
tile_width,
INROWS*sizeof(vbx_mm_t),
tile_height*sizeof(vbx_sp_t) );
tile_x += tile_width; // Set up width for next tile
if( tile_x + tile_width > INCOLS ) { // Temporarily reduce tile width when reaching right edge of matrix
tile_width = INCOLS - tile_x;
vbx_set_2D( tile_width, tile_height*sizeof(vbx_sp_t), sizeof(vbx_sp_t), sizeof(vbx_sp_t) );
vbx_set_3D( tile_height, sizeof(vbx_sp_t), tile_width*sizeof(vbx_sp_t), tile_width*sizeof(vbx_sp_t) );
}
}
tile_y += tile_height; // Set up width and height for next row of tiles
tile_width = prev_tile_width; // Restore original tile width for next row of tiles
/* *** Permanently reduce tile height when reaching bottom of matrix *** */
tile_height = ( tile_y + tile_height > INROWS ) ? INROWS - tile_y : tile_height; }
}
vbx_sp_pop();
vbx_sync();
return VBW_SUCCESS;
}
//vector version of rgb converter
void vector_blend(
output_pointer img_out, input_pointer img_in1, input_pointer img_in2,
unsigned int num_row, unsigned int num_column, intermediate_type blending_const )
{
intermediate_type *v_img1[2];
input_type *v_img2[2];
intermediate_type *v_temp;
intermediate_type blending_const_bar = 256-blending_const;
int j;
vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
const int VBX_SCRATCHPAD_SIZE = this_mxp->scratchpad_size;
const int VBX_WIDTH_BYTES = this_mxp->vector_lanes * sizeof(int);
const int VBX_DMA_ALIGNMENT = this_mxp->dma_alignment_bytes;
unsigned int chunk_size = VBX_SCRATCHPAD_SIZE/((3*sizeof(intermediate_type))+(2*sizeof(input_type)));
chunk_size = VBX_PAD_UP( chunk_size-(VBX_WIDTH_BYTES-1), VBX_DMA_ALIGNMENT );
unsigned int chunk_size_old = chunk_size;
unsigned int vector_length = chunk_size;
unsigned int vector_length_old = vector_length;
v_img1[0] = (intermediate_type *)vbx_sp_malloc( chunk_size*sizeof(intermediate_type) );
v_img1[1] = (intermediate_type *)vbx_sp_malloc( chunk_size*sizeof(intermediate_type) );
v_img2[0] = (input_type *)vbx_sp_malloc( chunk_size*sizeof(input_type) );
v_img2[1] = (input_type *)vbx_sp_malloc( chunk_size*sizeof(input_type) );
v_temp = (intermediate_type *)vbx_sp_malloc( chunk_size*sizeof(intermediate_type) );
if( v_temp == NULL ) {
VBX_EXIT(0xBADDEAD);
}
int bufselect = 0;
vbx_dma_to_vector( v_img1[bufselect], img_in1, chunk_size*sizeof(input_type) );
vbx_dma_to_vector( v_img2[bufselect], img_in2, chunk_size*sizeof(input_type) );
for( j=0; j<num_row*num_column; j+=vector_length_old ) {
vbx_set_vl(vector_length);
if( j > 0 ) {
vbx_dma_to_host( img_out+j-vector_length_old, v_img1[1-bufselect], chunk_size_old*sizeof(output_type) );
}
if( (j+vector_length_old) < (num_row*num_column-1) ) {
if( (j+vector_length_old*2) >= num_row*num_column ) {
vector_length = num_row*num_column - j - vector_length_old;
chunk_size = vector_length;
}
vbx_dma_to_vector( v_img1[1-bufselect], img_in1+j+vector_length_old, chunk_size*sizeof(input_type) );
vbx_dma_to_vector( v_img2[1-bufselect], img_in2+j+vector_length_old, chunk_size*sizeof(input_type) );
}
vbx( SVBHU, VMULLO, v_temp, blending_const, v_img1[bufselect] );
vbx( SVBHU, VMULLO, v_img1[bufselect], blending_const_bar, v_img2[bufselect] );
vbx( VVHU, VADD, v_img1[bufselect], v_img1[bufselect], v_temp );
vbx( SVHBU, VSHR, v_img1[bufselect], 8, v_img1[bufselect] );
bufselect = 1-bufselect;
}
vbx_dma_to_host( img_out+j-vector_length_old, v_img1[1-bufselect], chunk_size*sizeof(output_type) );
vbx_sp_free();
vbx_sync();
}
inline vector_mask_obj::vector_mask_obj(const VBX::Vector<T>& msk)
{
vbx_set_vl(msk.size);
constructor(msk);
}
int compare_vbx_lut_to_vbx_lut_ci(int stage, int max_print_errors)
{
vbx_mxp_t *this_mxp = VBX_GET_THIS_MXP();
int vci_lanes = this_mxp->vcustom0_lanes;
int sz = this_mxp->scratchpad_size/(16*sizeof(vbx_ubyte_t));
vbx_byte_t* v_pass = (vbx_byte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_pattern = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_lutc = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_group = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_sel = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_byte_t));
vbx_ubyte_t* v_lut = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_word_t));
vbx_ubyte_t* v_idx = (vbx_ubyte_t*)vbx_sp_malloc(sz*sizeof(vbx_word_t));
if(v_idx == NULL) {
printf("failed to allocate in compare_vbx_lut_to_vbx_lut_ci\n");
}
unsigned char* lut = (unsigned char*)vbx_shared_malloc(sz*sizeof(unsigned char));
unsigned char* lut_c = (unsigned char*)vbx_shared_malloc(sz*sizeof(unsigned char));
int f, n, s, errors = 0;
for (n = 0; n < sz; n++) {
v_pattern[n] = (n & 0xff);
}
for (f = 0; f < face_lbp[stage].count; f++) {
lbp_feat_t feat = face_lbp[stage].feats[f];
vbx_set_vl(sz);
int total = f;
s = 0;
while(s < stage){
total += face_lbp[s].count;
s++;
}
if(total < 256) {
vbx(SVBU, VLBPLUT, v_lutc, total, v_pattern);
} else {
vbx(SVBS, VLBPLUT, v_lutc, total-256, v_pattern);
}
vbx(SVB, VMOV, v_pass, feat.fail, 0);
/* check if pattern is in lut */
vbx(SVBU, VSHR, v_group, 5, v_pattern);
for (n = 0; n < 8; n++) {
vbx(SVB, VADD, v_sel, -n, v_group);
vbx(SVBW, VCMV_Z, v_lut, feat.lut[n], v_sel);
}
vbx(SVBWU, VAND, v_idx, 0x1f, v_pattern);
vbx(VVWB, VSHR, v_lut, v_idx, v_lut);
vbx(SVB, VAND, v_lut, 1, v_lut);
vbx(SVB, VCMV_LEZ, v_pass, feat.pass, v_lut);
vbx_dma_to_host(lut_c, v_lutc, sz*sizeof(unsigned char));
vbx_dma_to_host(lut, v_pass, sz*sizeof(unsigned char));
vbx_sync();
errors += match_array_byte(lut, lut_c, "custom_lut", sz, 1, 0, max_print_errors, 0, 0);
}
vbx_sp_free();
vbx_shared_free(lut);
vbx_shared_free(lut_c);
return errors;
}
/* takes in precomputed bfly */
static int vector_fix_fft_dif_long_fly(short fr[], short fi[], short fr2[], short fi2[], short tw_r[], short tw_i[], short m, short inverse, short real)
{
int i, j, l, k, scale, shift, a1,a2,bfly,mul,flight,swap,row_num;
short wr, wi;
vptr_half v_fr, v_fi, v_fr2, v_fi2, v_tmp;
vptr_half v_twr, v_twi;
vptr_half v_arp, v_aip, v_brp, v_bip, v_crp, v_cip;
vptr_half v_temp;
vptr_half v_twr2, v_twi2;
const int n = 1 << m;
const int half = n >> 1;
scale = 0;
mul = 0;
swap = m >> 1;
l = m-1;
flight = 1;
bfly = half;
const int INROWS = 1<<swap;
const int INCOLS = 1<<(m-swap);
if ( !(m%2) ){
swap--;
}
// allocate space in vector memory for vectors
v_fr = (vptr_half)vbx_sp_malloc( n*sizeof(vbx_half_t) );
v_fi = (vptr_half)vbx_sp_malloc( n*sizeof(vbx_half_t) );
v_fr2 = (vptr_half)vbx_sp_malloc( n*sizeof(vbx_half_t) );
v_fi2 = (vptr_half)vbx_sp_malloc( n*sizeof(vbx_half_t) );
v_twr = (vptr_half)vbx_sp_malloc( half*sizeof(vbx_half_t) );
v_twi = (vptr_half)vbx_sp_malloc( half*sizeof(vbx_half_t) );
v_temp = (vptr_half)vbx_sp_malloc( n*sizeof(vbx_half_t) );
if( v_fr == NULL || v_fi == NULL || v_fr2 == NULL || v_fi2== NULL || \
v_twr == NULL || v_twi == NULL || v_temp == NULL) {
VBX_EXIT(-1);
}
v_twr2 = (vptr_half)vbx_sp_malloc( half*sizeof(vbx_half_t) );
v_twi2 = (vptr_half)vbx_sp_malloc( half*sizeof(vbx_half_t) );
if( v_twr2 == NULL || v_twi2 == NULL) {
VBX_EXIT(-1);
}
vbx_dma_to_vector( v_fr, fr, n*sizeof(vbx_half_t) );
vbx_dma_to_vector( v_fi, fi, n*sizeof(vbx_half_t) );
vbx_dma_to_vector( v_twr, tw_r, half*sizeof(vbx_half_t) );
vbx_dma_to_vector( v_twi, tw_i, half*sizeof(vbx_half_t) );
#if 1
if(real){
vector_fix_fft_untangle_real_scratch( v_fr, v_fi, v_fr2, v_fi2, v_twr,v_twi, m, inverse);
}
#endif
while (l > swap) {
if (inverse) {
// variable scaling, depending upon data
shift = 0;
if( isAbsOutOfRangeV(v_fr,v_fi,v_temp,n) ) {
shift = 1;
scale++;
}
} else {
// fixed scaling, for proper normalization
// -- overall factor of 1/n, distributed to maximize arithmetic accuracy
shift = 1;
}
// shift will be performed on each data point exactly once during pass
SWAP( v_fr, v_fr2, v_tmp );
SWAP( v_fi, v_fi2, v_tmp );
if (shift){
vbx_set_vl( n );
vbx(SVH,VSHR, v_fr2, 1, v_fr2 );
vbx(SVH,VSHR, v_fi2, 1, v_fi2 );
}
vbx_set_vl( 1<<l );
vbx_set_2D( flight, sizeof(vbx_half_t)*(1<<l), sizeof(vbx_half_t)*(1<<l+1), sizeof(vbx_half_t)*(1<<l+1) );
vbx_2D( VVH, VADD, v_fr, v_fr2, v_fr2 + (1<<l) );
vbx_2D( VVH, VADD, v_fi, v_fi2, v_fi2 + (1<<l) );
vbx_set_2D( flight, sizeof(vbx_half_t)*(1<<l+1), sizeof(vbx_half_t)*(1<<l+1), sizeof(vbx_half_t)*(1<<l+1) );
vbx_2D( VVH, VSUB, v_fr2, v_fr2, v_fr2 + (1<<l) );
vbx_2D( VVH, VSUB, v_fi2, v_fi2, v_fi2 + (1<<l) );
vbx_set_2D( flight, sizeof(vbx_half_t)*(1<<l), sizeof(vbx_half_t)*(1<<l+1), 0 );
vbx_2D( VVH, VMULFXP, &v_fr[n>>1], v_fr2, v_twr );
vbx_2D( VVH, VMULFXP, v_temp, v_fi2, v_twi );
vbx_set_vl( n>>1 ); // vbx_set_2D( flight, sizeof(vbx_half_t)*(1<<l), sizeof(vbx_half_t)*(1<<l), sizeof(vbx_half_t)*(1<<l) );
vbx( VVH, VSUB, &v_fr[n>>1], &v_fr[n>>1], v_temp );
vbx_set_vl( 1<<l );
vbx_set_2D( flight, sizeof(vbx_half_t)*(1<<l), sizeof(vbx_half_t)*(1<<l+1), 0 );
vbx_2D( VVH, VMULFXP, &v_fi[n>>1], v_fi2, v_twr );
vbx_2D( VVH, VMULFXP, v_temp, v_fr2, v_twi );
vbx_set_vl( n>>1 ); //vbx_set_2D( flight, sizeof(vbx_half_t)*(1<<l), sizeof(vbx_half_t)*(1<<l), sizeof(vbx_half_t)*(1<<l) );
vbx( VVH, VADD, &v_fi[n>>1], &v_fi[n>>1], v_temp );
l--;
mul++;
flight <<= 1;
if( l > swap ) {
vbx_set_vl( 1<<l );
vbx( VVWH, VMOV, v_twr, v_twr, 0 );
vbx( VVWH, VMOV, v_twi, v_twi, 0 );
}
}
if ( !(m%2) ) {
l++;
flight >>=1;
}
