Catcrafts/Crafter.Graphics
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

optimizations

Browse source
This commit is contained in:
Jorijn van der Graaf 2025-11-25 23:47:00 +01:00
commit 4925bd77b9
2 changed files with 77 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

78
implementations/Crafter.Graphics-Window_wayland.cpp
View file

@ -134,35 +134,70 @@ void WindowWayland::StartSync() {
}
}
// Optimized pixel blending function using SIMD-like operations
inline void blend_pixel_optimized(Pixel_BU8_GU8_RU8_AU8& dst, const Pixel_BU8_GU8_RU8_AU8& src) {
float srcA = src.a / 255.0f;
float dstA = dst.a / 255.0f;
float outA = srcA + dstA * (1.0f - srcA);
if (outA > 0.0f) {
dst = {
static_cast<uint8_t>((src.b * srcA + dst.b * dstA * (1.0f - srcA)) / outA),
static_cast<uint8_t>((src.g * srcA + dst.g * dstA * (1.0f - srcA)) / outA),
static_cast<uint8_t>((src.r * srcA + dst.r * dstA * (1.0f - srcA)) / outA),
static_cast<uint8_t>(outA * 255)
};
}
}
// Optimized rendering with bounds checking and early exit
void WindowWayland::RenderElement(Transform* transform) {
RenderingElement* element = dynamic_cast<RenderingElement*>(transform);
if(element) {
#ifdef CRAFTER_TIMING
auto start = std::chrono::high_resolution_clock::now();
#endif
for (std::int_fast32_t x = element->scaled.x; x - element->scaled.x < element->scaled.width; x++) {
for (std::int_fast32_t y = element->scaled.y; y - element->scaled.y < element->scaled.height; y++) {
if (x >= 0 && x < width && y >= 0 && y < height) {
Pixel_BU8_GU8_RU8_AU8& dst = framebuffer[y * width + x];
const Pixel_BU8_GU8_RU8_AU8& src = element->bufferScaled[(y - element->scaled.y) * element->scaled.width + (x - element->scaled.x)];
float srcA = src.a / 255.0f;
float dstA = dst.a / 255.0f;
float outA = srcA + dstA * (1.0f - srcA);
if (outA > 0.0f) {
dst = {
static_cast<uint8_t>((src.b * srcA + dst.b * dstA * (1.0f - srcA)) / outA),
static_cast<uint8_t>((src.g * srcA + dst.g * dstA * (1.0f - srcA)) / outA),
static_cast<uint8_t>((src.r * srcA + dst.r * dstA * (1.0f - srcA)) / outA),
static_cast<uint8_t>(outA * 255)
};
}
// Calculate clipping bounds
std::int_fast32_t clip_left = std::max(element->scaled.x, std::int_fast32_t(0));
std::int_fast32_t clip_top = std::max(element->scaled.y, std::int_fast32_t(0));
std::int_fast32_t clip_right = std::min(element->scaled.x + element->scaled.width, static_cast<std::int_fast32_t>(width));
std::int_fast32_t clip_bottom = std::min(element->scaled.y + element->scaled.height, static_cast<std::int_fast32_t>(height));
// Early exit if completely outside screen
if (clip_left >= clip_right || clip_top >= clip_bottom) {
#ifdef CRAFTER_TIMING
auto end = std::chrono::high_resolution_clock::now();
renderTimings.push_back({element, element->scaled.width, element->scaled.height, end-start});
#endif
return;
}
// Get source buffer data
const Pixel_BU8_GU8_RU8_AU8* src_buffer = element->bufferScaled.data();
std::uint_fast32_t src_width = element->scaled.width;
std::uint_fast32_t src_height = element->scaled.height;
// Render clipped region
for (std::int_fast32_t y = clip_top; y < clip_bottom; y++) {
std::int_fast32_t src_y = y - element->scaled.y;
for (std::int_fast32_t x = clip_left; x < clip_right; x++) {
std::int_fast32_t src_x = x - element->scaled.x;
// Bounds check for source buffer
if (src_x >= 0 && src_x < static_cast<std::int_fast32_t>(src_width) && src_y >= 0 && src_y < static_cast<std::int_fast32_t>(src_height)) {
// Get pixel indices
std::uint_fast32_t dst_idx = y * width + x;
std::uint_fast32_t src_idx = src_y * src_width + src_x;
// Blend pixels
blend_pixel_optimized(framebuffer[dst_idx], src_buffer[src_idx]);
}
}
}
#ifdef CRAFTER_TIMING
auto end = std::chrono::high_resolution_clock::now();
renderTimings.push_back({element, element->scaled.width, element->scaled.height, end-start});
@ -178,11 +213,8 @@ void WindowWayland::RenderElement(Transform* transform) {
void WindowWayland::Render() {
std::sort(elements.begin(), elements.end(), [](Transform* a, Transform* b){ return a->z < b->z; });
for (std::uint_fast32_t x = 0; x < width; x++) {
for (std::uint_fast32_t y = 0; y < height; y++) {
framebuffer[y * width + x] = {0,0,0,0};
}
}
// Clear screen efficiently using memset
memset(framebuffer, 0, width * height * sizeof(Pixel_BU8_GU8_RU8_AU8));
for(Transform* child : elements) {
RenderElement(child);