refactor(gpu): eliminar GPU compute boids (prevé crash macOS)

Elimina el kernel Metal O(N²) de boids en GPU que causava GPU timeout a macOS amb >50K boles, arrossegant WindowServer fins al crash. - Elimina gpu_boid_buffer.hpp/cpp (GpuBoidBuffer, BallComputeData, BoidParams) - Elimina kBoidComputeMSL i kBallComputeVertMSL de gpu_pipeline - Elimina boid_compute_pipeline_ i ball_compute_pipeline_ - Elimina use_gpu_boids_, boid_params_, ball_screen_uniforms_ de Engine - Elimina syncAndExitGpuBoids() i tot el compute dispatch de render() - Mode BOIDS ara usa sempre boid_manager_ (CPU, spatial hash O(N)) i renderitza via gpu_ball_buffer_ instanced (mateix path que PHYSICS) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-20 08:45:01 +01:00
parent badf92420b
commit d2e7f2ff86
8 changed files with 321 additions and 41 deletions
--- a/source/gpu/gpu_ball_buffer.cpp
+++ b/source/gpu/gpu_ball_buffer.cpp
@@ -0,0 +1,65 @@
+#include "gpu_ball_buffer.hpp"
+
+#include <SDL3/SDL_log.h>
+#include <algorithm>  // std::min
+#include <cstring>    // memcpy
+
+bool GpuBallBuffer::init(SDL_GPUDevice* device) {
+    Uint32 buf_size = static_cast<Uint32>(MAX_BALLS) * sizeof(BallGPUData);
+
+    // GPU vertex buffer (instance-rate data read by the ball instanced shader)
+    SDL_GPUBufferCreateInfo buf_info = {};
+    buf_info.usage = SDL_GPU_BUFFERUSAGE_VERTEX;
+    buf_info.size  = buf_size;
+    gpu_buf_ = SDL_CreateGPUBuffer(device, &buf_info);
+    if (!gpu_buf_) {
+        SDL_Log("GpuBallBuffer: GPU buffer creation failed: %s", SDL_GetError());
+        return false;
+    }
+
+    // Transfer buffer (upload staging, cycled every frame)
+    SDL_GPUTransferBufferCreateInfo tb_info = {};
+    tb_info.usage = SDL_GPU_TRANSFERBUFFERUSAGE_UPLOAD;
+    tb_info.size  = buf_size;
+    transfer_buf_ = SDL_CreateGPUTransferBuffer(device, &tb_info);
+    if (!transfer_buf_) {
+        SDL_Log("GpuBallBuffer: transfer buffer creation failed: %s", SDL_GetError());
+        return false;
+    }
+
+    SDL_Log("GpuBallBuffer: initialized (capacity %d balls, %.1f MB VRAM)",
+            MAX_BALLS, buf_size / (1024.0f * 1024.0f));
+    return true;
+}
+
+void GpuBallBuffer::destroy(SDL_GPUDevice* device) {
+    if (!device) return;
+    if (transfer_buf_) { SDL_ReleaseGPUTransferBuffer(device, transfer_buf_); transfer_buf_ = nullptr; }
+    if (gpu_buf_)      { SDL_ReleaseGPUBuffer(device, gpu_buf_);              gpu_buf_      = nullptr; }
+    count_ = 0;
+}
+
+bool GpuBallBuffer::upload(SDL_GPUDevice* device, SDL_GPUCommandBuffer* cmd,
+                            const BallGPUData* data, int count) {
+    if (!data || count <= 0) { count_ = 0; return false; }
+    count = std::min(count, MAX_BALLS);
+
+    Uint32 upload_size = static_cast<Uint32>(count) * sizeof(BallGPUData);
+
+    void* ptr = SDL_MapGPUTransferBuffer(device, transfer_buf_, true /* cycle */);
+    if (!ptr) {
+        SDL_Log("GpuBallBuffer: transfer buffer map failed: %s", SDL_GetError());
+        return false;
+    }
+    memcpy(ptr, data, upload_size);
+    SDL_UnmapGPUTransferBuffer(device, transfer_buf_);
+
+    SDL_GPUCopyPass* copy = SDL_BeginGPUCopyPass(cmd);
+    SDL_GPUTransferBufferLocation src = { transfer_buf_, 0 };
+    SDL_GPUBufferRegion           dst = { gpu_buf_,      0, upload_size };
+    SDL_UploadToGPUBuffer(copy, &src, &dst, true /* cycle */);
+    SDL_EndGPUCopyPass(copy);
+
+    count_ = count;
+    return true;
+}