782e2e5f9a
In Blender a context should not be shared between threads. In Vulkan a command pool must not be shared between threads. In the current implementation the command pool are stored on device level and could therefore be shared between multiple context which made the implementation not matching these rules. This PR moves the command pool from device to command buffers where it would not conflict between other contexts. This PR doesn't make the Vulkan backend fully multithreaded. The access to the queue is still missing. Pull Request: https://projects.blender.org/blender/blender/pulls/114977
66 lines
1.9 KiB
C++
66 lines
1.9 KiB
C++
/* SPDX-FileCopyrightText: 2023 Blender Authors
|
|
*
|
|
* SPDX-License-Identifier: GPL-2.0-or-later */
|
|
|
|
/** \file
|
|
* \ingroup gpu
|
|
*/
|
|
|
|
#pragma once
|
|
|
|
#include "BLI_vector.hh"
|
|
|
|
#include "vk_descriptor_set.hh"
|
|
|
|
namespace blender::gpu {
|
|
class VKDevice;
|
|
|
|
/**
|
|
* List of VkDescriptorPools.
|
|
*
|
|
* In Vulkan a pool is constructed with a fixed size per resource type. When more resources are
|
|
* needed it a next pool should be created. VKDescriptorPools will keep track of those pools and
|
|
* construct new pools when the previous one is exhausted.
|
|
*
|
|
* At the beginning of a new frame the descriptor pools are reset. This will start allocating
|
|
* again from the first descriptor pool in order to use freed space from previous pools.
|
|
*/
|
|
class VKDescriptorPools {
|
|
/**
|
|
* Pool sizes to use. When one descriptor pool is requested to allocate a descriptor but isn't
|
|
* able to do so, it will fail.
|
|
*
|
|
* Better defaults should be set later on, when we know more about our resource usage.
|
|
*/
|
|
static constexpr uint32_t POOL_SIZE_STORAGE_BUFFER = 1000;
|
|
static constexpr uint32_t POOL_SIZE_DESCRIPTOR_SETS = 1000;
|
|
static constexpr uint32_t POOL_SIZE_STORAGE_IMAGE = 1000;
|
|
static constexpr uint32_t POOL_SIZE_COMBINED_IMAGE_SAMPLER = 1000;
|
|
static constexpr uint32_t POOL_SIZE_UNIFORM_BUFFER = 1000;
|
|
static constexpr uint32_t POOL_SIZE_UNIFORM_TEXEL_BUFFER = 1000;
|
|
|
|
Vector<VkDescriptorPool> pools_;
|
|
int64_t active_pool_index_ = 0;
|
|
|
|
public:
|
|
VKDescriptorPools();
|
|
~VKDescriptorPools();
|
|
|
|
void init(const VKDevice &vk_device);
|
|
|
|
std::unique_ptr<VKDescriptorSet> allocate(const VkDescriptorSetLayout &descriptor_set_layout);
|
|
|
|
/**
|
|
* Reset the pools to start looking for free space from the first descriptor pool.
|
|
*/
|
|
void reset();
|
|
|
|
private:
|
|
VkDescriptorPool active_pool_get();
|
|
void activate_next_pool();
|
|
void activate_last_pool();
|
|
bool is_last_pool_active();
|
|
void add_new_pool(const VKDevice &device);
|
|
};
|
|
} // namespace blender::gpu
|