claude_code/procedural-3d-engine
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

No more per-thread vertex and index buffer copy, more objects, random radial distribution (Fixes #168)

Browse source
This commit is contained in:
saschawillems 2016-09-21 18:52:27 +02:00
parent 87dd4142fb
commit 195dbec797
1 changed files with 17 additions and 60 deletions
Download patch file Download diff file Expand all files Collapse all files

77
multithreading/multithreading.cpp
View file

@ -12,6 +12,7 @@
#include <assert.h> #include <assert.h>
#include <vector> #include <vector>
#include <thread> #include <thread>
#include <random>
#define GLM_FORCE_RADIANS #define GLM_FORCE_RADIANS
#define GLM_FORCE_DEPTH_ZERO_TO_ONE #define GLM_FORCE_DEPTH_ZERO_TO_ONE
@ -94,7 +95,6 @@ public:
}; };
struct ThreadData { struct ThreadData {
vkMeshLoader::MeshBuffer mesh;
VkCommandPool commandPool; VkCommandPool commandPool;
// One command buffer per render object // One command buffer per render object
std::vector<VkCommandBuffer> commandBuffer; std::vector<VkCommandBuffer> commandBuffer;
@ -138,7 +138,7 @@ public:
threadPool.setThreadCount(numThreads); threadPool.setThreadCount(numThreads);
numObjectsPerThread = 256 / numThreads; numObjectsPerThread = 512 / numThreads;
} }
~VulkanExample() ~VulkanExample()
@ -160,7 +160,6 @@ public:
{ {
vkFreeCommandBuffers(device, thread.commandPool, thread.commandBuffer.size(), thread.commandBuffer.data()); vkFreeCommandBuffers(device, thread.commandPool, thread.commandBuffer.size(), thread.commandBuffer.data());
vkDestroyCommandPool(device, thread.commandPool, nullptr); vkDestroyCommandPool(device, thread.commandPool, nullptr);
vkMeshLoader::freeMeshBufferResources(device, &thread.mesh);
} }
vkDestroyFence(device, renderFence, nullptr); vkDestroyFence(device, renderFence, nullptr);
@ -196,6 +195,9 @@ public:
uint32_t posX = 0; uint32_t posX = 0;
uint32_t posZ = 0; uint32_t posZ = 0;
std::mt19937 rndGenerator((unsigned)time(NULL));
std::uniform_real_distribution<float> uniformDist(0.0f, 1.0f);
for (uint32_t i = 0; i < numThreads; i++) for (uint32_t i = 0; i < numThreads; i++)
{ {
ThreadData *thread = &threadData[i]; ThreadData *thread = &threadData[i];
@ -216,61 +218,14 @@ public:
thread->commandBuffer.size()); thread->commandBuffer.size());
VK_CHECK_RESULT(vkAllocateCommandBuffers(device, &secondaryCmdBufAllocateInfo, thread->commandBuffer.data())); VK_CHECK_RESULT(vkAllocateCommandBuffers(device, &secondaryCmdBufAllocateInfo, thread->commandBuffer.data()));
// Unique vertex and index buffers per thread
createBuffer(
VK_BUFFER_USAGE_VERTEX_BUFFER_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT,
meshes.ufo.vertices.size,
nullptr,
&thread->mesh.vertices.buf,
&thread->mesh.vertices.mem);
createBuffer(
VK_BUFFER_USAGE_INDEX_BUFFER_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT,
meshes.ufo.indices.size,
nullptr,
&thread->mesh.indices.buf,
&thread->mesh.indices.mem);
// Copy from mesh buffer
VkBufferCopy copyRegion = {};
// Vertex buffer
copyRegion.size = meshes.ufo.vertices.size;
vkCmdCopyBuffer(
setupCmdBuffer,
meshes.ufo.vertices.buf,
thread->mesh.vertices.buf,
1,
&copyRegion);
// Index buffer
copyRegion.size = meshes.ufo.indices.size;
vkCmdCopyBuffer(
setupCmdBuffer,
meshes.ufo.indices.buf,
thread->mesh.indices.buf,
1,
&copyRegion);
// todo : staging
thread->mesh.indexCount = meshes.ufo.indexCount;
thread->pushConstBlock.resize(numObjectsPerThread); thread->pushConstBlock.resize(numObjectsPerThread);
thread->objectData.resize(numObjectsPerThread); thread->objectData.resize(numObjectsPerThread);
float step = 360.0f / (float)(numThreads * numObjectsPerThread);
for (uint32_t j = 0; j < numObjectsPerThread; j++) for (uint32_t j = 0; j < numObjectsPerThread; j++)
{ {
float radius = 8.0f + rnd(8.0f) - rnd(4.0f); float theta = 2.0f * float(M_PI) * uniformDist(rndGenerator);
float phi = acos(1.0f - 2.0f * uniformDist(rndGenerator));
thread->objectData[j].pos.x = (posX - maxX / 2.0f) * 3.0f + rnd(1.5f) - rnd(1.5f); thread->objectData[j].pos = glm::vec3(sin(phi) * cos(theta), 0.0f, cos(phi)) * 35.0f;
thread->objectData[j].pos.z = (posZ - maxX / 2.0f) * 3.0f + rnd(1.5f) - rnd(1.5f);
posX += 1.0f;
if (posX >= maxX)
{
posX = 0.0f;
posZ += 1.0f;
}
thread->objectData[j].rotation = glm::vec3(0.0f, rnd(360.0f), 0.0f); thread->objectData[j].rotation = glm::vec3(0.0f, rnd(360.0f), 0.0f);
thread->objectData[j].deltaT = rnd(1.0f); thread->objectData[j].deltaT = rnd(1.0f);
@ -281,10 +236,7 @@ public:
thread->pushConstBlock[j].color = glm::vec3(rnd(1.0f), rnd(1.0f), rnd(1.0f)); thread->pushConstBlock[j].color = glm::vec3(rnd(1.0f), rnd(1.0f), rnd(1.0f));
} }
} }
// Submit buffer copies to the queue
flushSetupCommandBuffer();
// todo : fence?
} }
// Builds the secondary command buffer for each thread // Builds the secondary command buffer for each thread
@ -347,9 +299,14 @@ public:
&thread->pushConstBlock[cmdBufferIndex]); &thread->pushConstBlock[cmdBufferIndex]);
VkDeviceSize offsets[1] = { 0 }; VkDeviceSize offsets[1] = { 0 };
vkCmdBindVertexBuffers(cmdBuffer, 0, 1, &thread->mesh.vertices.buf, offsets); //vkCmdBindVertexBuffers(cmdBuffer, 0, 1, &thread->mesh.vertices.buf, offsets);
vkCmdBindIndexBuffer(cmdBuffer, thread->mesh.indices.buf, 0, VK_INDEX_TYPE_UINT32); //vkCmdBindIndexBuffer(cmdBuffer, thread->mesh.indices.buf, 0, VK_INDEX_TYPE_UINT32);
vkCmdDrawIndexed(cmdBuffer, thread->mesh.indexCount, 1, 0, 0, 0); //vkCmdDrawIndexed(cmdBuffer, thread->mesh.indexCount, 1, 0, 0, 0);
vkCmdBindVertexBuffers(cmdBuffer, 0, 1, &meshes.ufo.vertices.buf, offsets);
vkCmdBindIndexBuffer(cmdBuffer, meshes.ufo.indices.buf, 0, VK_INDEX_TYPE_UINT32);
vkCmdDrawIndexed(cmdBuffer, meshes.ufo.indexCount, 1, 0, 0, 0);
VK_CHECK_RESULT(vkEndCommandBuffer(cmdBuffer)); VK_CHECK_RESULT(vkEndCommandBuffer(cmdBuffer));
} }