Anurag Thakur pushed to branch gsoc-anurag-2023 at FreeType / FreeType
Commits:
2 changed files:
Changes:
src/base/ftobjs.c
... |
... |
@@ -3154,7 +3154,7 @@ int conic_to2(FT_GlyphSlot* slot, FT_Vector *control, FT_Vector *from, FT_Vector |
3154
|
3154
|
face->garray = (FT_GlyphSlot*)malloc(
|
3155
|
3155
|
face->driver->clazz->slot_object_size * face->num_glyphs );
|
3156
|
3156
|
//error = FT_Set_Char_Size( face, 0, 160 * 64, 300, 300 );
|
3157
|
|
- error = FT_Set_Pixel_Sizes( face, 0, 50);
|
|
3157
|
+ error = FT_Set_Pixel_Sizes( face, 0, 500);
|
3158
|
3158
|
// int glyph_index = FT_Get_Char_Index( face, 'A' );
|
3159
|
3159
|
// error = FT_Load_Glyph( face, glyph_index, FT_LOAD_NO_HINTING );
|
3160
|
3160
|
|
src/dense/ftdense.c
... |
... |
@@ -454,6 +454,7 @@ dense_render_glyph( dense_worker* worker, const FT_Bitmap* target, FT_PreLine pl |
454
|
454
|
|
455
|
455
|
__m128i offset = _mm_setzero_si128();
|
456
|
456
|
__m128i mask = _mm_set1_epi32( 0x0c080400 );
|
|
457
|
+ __m128i nzero = _mm_castps_si128(_mm_set1_ps(-0.0));
|
457
|
458
|
|
458
|
459
|
for (int i = 0; i < worker->m_h*worker->m_w; i += 4)
|
459
|
460
|
{
|
... |
... |
@@ -463,18 +464,31 @@ __m128i offset = _mm_setzero_si128(); |
463
|
464
|
|
464
|
465
|
x = _mm_add_epi32( x, _mm_slli_si128( x, 4 ) );
|
465
|
466
|
|
466
|
|
- x = _mm_add_epi32(
|
467
|
|
- x, _mm_castps_si128( _mm_shuffle_ps( _mm_setzero_ps(),
|
468
|
|
- _mm_castsi128_ps( x ), 0x40 ) ) );
|
|
467
|
+ // x = _mm_add_epi32(
|
|
468
|
+ // x, _mm_castps_si128( _mm_shuffle_ps( _mm_setzero_ps(),
|
|
469
|
+ // _mm_castsi128_ps( x ), 0x40 ) ) );
|
|
470
|
+ x = _mm_add_epi32(x, _mm_slli_si128(x,8));
|
469
|
471
|
|
470
|
472
|
// add the prefsum of previous 4 floats to all current floats
|
471
|
473
|
x = _mm_add_epi32( x, offset );
|
472
|
474
|
|
|
475
|
+
|
|
476
|
+
|
|
477
|
+ // __m128 y = _mm_mul_ps(_mm_castsi128_ps(x), _mm_set1_ps(255.9));
|
|
478
|
+
|
|
479
|
+ // y = _mm_andnot_ps(_mm_castsi128_ps(nzero), y);
|
|
480
|
+
|
|
481
|
+ // __m128i z = _mm_cvttps_epi32(y);
|
|
482
|
+ // z = _mm_packus_epi16(_mm_packs_epi32(z, nzero), nzero);
|
|
483
|
+
|
|
484
|
+ // *((int*)dest+i) = _mm_extract_epi16(z, 0);
|
|
485
|
+
|
|
486
|
+
|
473
|
487
|
// take absolute value
|
474
|
488
|
__m128i y = _mm_abs_epi32( x ); // fabs(x)
|
475
|
489
|
|
476
|
490
|
// cap max value to 1
|
477
|
|
- y = _mm_min_epi32( y, _mm_set1_epi32( 4080 ) );
|
|
491
|
+ y = _mm_min_epi32( y, _mm_set1_epi32( 4094 ) );
|
478
|
492
|
|
479
|
493
|
// reduce to 255
|
480
|
494
|
y = _mm_srli_epi32( y, 4 );
|
... |
... |
@@ -488,6 +502,7 @@ __m128i offset = _mm_setzero_si128(); |
488
|
502
|
offset = _mm_castps_si128( _mm_shuffle_ps( _mm_castsi128_ps( x ),
|
489
|
503
|
_mm_castsi128_ps( x ),
|
490
|
504
|
_MM_SHUFFLE( 3, 3, 3, 3 ) ) );
|
|
505
|
+ //offset = _mm_set1_epi32(_mm_extract_epi32(x, 3));
|
491
|
506
|
}
|
492
|
507
|
|
493
|
508
|
#else /* FT_SSE4_1 */
|
|