Drop the extra full-screen copy at render scale < 1

The SDL_GPU pipeline renders into its own view-sized colour target and
already composites it over the whole render target, so routing it through
gpu_scene_texture cost one extra same-size copy per frame. The scaled
intermediate is now used only by the SDL geometry fallback; the hardware
composite upscales with linear filtering.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
shenlei
2026-10-12 17:19:51 +09:00
co-authored by Claude Opus 5.5
parent b57f2fba99
commit df87bc64eb
4 changed files with 21 additions and 2 deletions
+2
View File
@@ -96,6 +96,8 @@ public:
void RenderSceneGPU(SDL_Renderer* renderer, const std::vector<Render3DDraw>& draws);
void RenderSceneGPU(SDL_Renderer* renderer, const std::vector<const Render3DDraw*>& draws);
void SetGPUDevice(SDL_GPUDevice* device);
// The SDL_GPU pipeline composites its own (smaller) colour target to the full render target.
bool HasHardwarePipeline() const { return m_hardware != nullptr; }
void DestroyGPUTextures();
SDL_Texture* GetGPUTexture(SDL_Renderer* renderer, const std::string& texturePath);