godotengine/godot
1
0
Fork 0
mirror of https://github.com/godotengine/godot.git synced 2025-11-04 12:00:25 +00:00
Code Releases Wiki Activity

Rewrote how barriers work for faster rendering

Browse Source 
-Added more finegrained control in RenderingDevice API
-Optimized barriers (use less ones for thee same)
-General optimizations
-Shadows render all together unbarriered
-GI can render together with shadows.
-SDFGI can render together with depth-preoass.
-General fixes
-Added GPU detection
This commit is contained in:
reduz
2021-02-02 16:51:36 -03:00
parent 1d1310f3d7
commit f20999f6fe
25 changed files with 2102 additions and 1164 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
servers/rendering/renderer_rd/shader_rd.cpp
View File

@@ -301,6 +301,7 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) {
builder.append(compute_codev.get_data()); // version info (if exists)
builder.append("\n"); //make sure defines begin at newline
builder.append(base_compute_defines.get_data());
builder.append(general_defines.get_data());
builder.append(variant_defines[p_variant].get_data());
@@ -401,7 +402,6 @@ RS::ShaderNativeSourceCode ShaderRD::version_get_native_source_code(RID p_versio
builder.append(fragment_codev.get_data()); // version info (if exists)
builder.append("\n"); //make sure defines begin at newline
builder.append(general_defines.get_data());
builder.append(variant_defines[i].get_data());
for (int j = 0; j < version->custom_defines.size(); j++) {
@@ -440,6 +440,7 @@ RS::ShaderNativeSourceCode ShaderRD::version_get_native_source_code(RID p_versio
builder.append(compute_codev.get_data()); // version info (if exists)
builder.append("\n"); //make sure defines begin at newline
builder.append(base_compute_defines.get_data());
builder.append(general_defines.get_data());
builder.append(variant_defines[i].get_data());
@@ -596,6 +597,22 @@ bool ShaderRD::is_variant_enabled(int p_variant) const {
return variants_enabled[p_variant];
}
ShaderRD::ShaderRD() {
// Do not feel forced to use this, in most cases it makes little to no difference.
bool use_32_threads = false;
if (RD::get_singleton()->get_device_vendor_name() == "NVIDIA") {
use_32_threads = true;
}
String base_compute_define_text;
if (use_32_threads) {
base_compute_define_text = "\n#define NATIVE_LOCAL_GROUP_SIZE 32\n#define NATIVE_LOCAL_SIZE_2D_X 8\n#define NATIVE_LOCAL_SIZE_2D_Y 4\n";
} else {
base_compute_define_text = "\n#define NATIVE_LOCAL_GROUP_SIZE 64\n#define NATIVE_LOCAL_SIZE_2D_X 8\n#define NATIVE_LOCAL_SIZE_2D_Y 8\n";
}
base_compute_defines = base_compute_define_text.ascii();
}
void ShaderRD::initialize(const Vector<String> &p_variant_defines, const String &p_general_defines) {
ERR_FAIL_COND(variant_defines.size());
ERR_FAIL_COND(p_variant_defines.size() == 0);