yuzu/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
ReinUsesLisp 35df1d1864 vk_staging_buffer_pool: Add stream buffer for small uploads
This uses a ring buffer similar to OpenGL's stream buffer for small
uploads. This stops us from allocating several small buffers, reducing
memory fragmentation and cache locality.

It uses dedicated allocations when possible.
2021-02-13 02:17:24 -03:00

276 lines
10 KiB
C++

// Copyright 2019 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <algorithm>
#include <utility>
#include <vector>
#include <fmt/format.h>
#include "common/alignment.h"
#include "common/assert.h"
#include "common/bit_util.h"
#include "common/common_types.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_staging_buffer_pool.h"
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
namespace Vulkan {
namespace {
// Maximum potential alignment of a Vulkan buffer
constexpr VkDeviceSize MAX_ALIGNMENT = 256;
// Maximum size to put elements in the stream buffer
constexpr VkDeviceSize MAX_STREAM_BUFFER_REQUEST_SIZE = 8 * 1024 * 1024;
// Stream buffer size in bytes
constexpr VkDeviceSize STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
constexpr VkDeviceSize REGION_SIZE = STREAM_BUFFER_SIZE / StagingBufferPool::NUM_SYNCS;
constexpr VkMemoryPropertyFlags HOST_FLAGS =
VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT;
constexpr VkMemoryPropertyFlags STREAM_FLAGS = VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT | HOST_FLAGS;
bool IsStreamHeap(VkMemoryHeap heap) noexcept {
return STREAM_BUFFER_SIZE < (heap.size * 2) / 3;
}
std::optional<u32> FindMemoryTypeIndex(const VkPhysicalDeviceMemoryProperties& props, u32 type_mask,
VkMemoryPropertyFlags flags) noexcept {
for (u32 type_index = 0; type_index < props.memoryTypeCount; ++type_index) {
if (((type_mask >> type_index) & 1) == 0) {
// Memory type is incompatible
continue;
}
const VkMemoryType& memory_type = props.memoryTypes[type_index];
if ((memory_type.propertyFlags & flags) != flags) {
// Memory type doesn't have the flags we want
continue;
}
if (!IsStreamHeap(props.memoryHeaps[memory_type.heapIndex])) {
// Memory heap is not suitable for streaming
continue;
}
// Success!
return type_index;
}
return std::nullopt;
}
u32 FindMemoryTypeIndex(const VkPhysicalDeviceMemoryProperties& props, u32 type_mask) {
// Try to find a DEVICE_LOCAL_BIT type, Nvidia and AMD have a dedicated heap for this
std::optional<u32> type = FindMemoryTypeIndex(props, type_mask, STREAM_FLAGS);
if (type) {
return *type;
}
// Otherwise try without the DEVICE_LOCAL_BIT
type = FindMemoryTypeIndex(props, type_mask, HOST_FLAGS);
if (type) {
return *type;
}
// This should never happen, and in case it does, signal it as an out of memory situation
throw vk::Exception(VK_ERROR_OUT_OF_DEVICE_MEMORY);
}
size_t Region(size_t iterator) noexcept {
return iterator / REGION_SIZE;
}
} // Anonymous namespace
StagingBufferPool::StagingBufferPool(const Device& device_, MemoryAllocator& memory_allocator_,
VKScheduler& scheduler_)
: device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_} {
const vk::Device& dev = device.GetLogical();
stream_buffer = dev.CreateBuffer(VkBufferCreateInfo{
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.size = STREAM_BUFFER_SIZE,
.usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_INDEX_BUFFER_BIT,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
});
if (device.HasDebuggingToolAttached()) {
stream_buffer.SetObjectNameEXT("Stream Buffer");
}
VkMemoryDedicatedRequirements dedicated_reqs{
.sType = VK_STRUCTURE_TYPE_MEMORY_DEDICATED_REQUIREMENTS,
.pNext = nullptr,
.prefersDedicatedAllocation = VK_FALSE,
.requiresDedicatedAllocation = VK_FALSE,
};
const auto requirements = dev.GetBufferMemoryRequirements(*stream_buffer, &dedicated_reqs);
const bool make_dedicated = dedicated_reqs.prefersDedicatedAllocation == VK_TRUE ||
dedicated_reqs.requiresDedicatedAllocation == VK_TRUE;
const VkMemoryDedicatedAllocateInfo dedicated_info{
.sType = VK_STRUCTURE_TYPE_MEMORY_DEDICATED_ALLOCATE_INFO,
.pNext = nullptr,
.image = nullptr,
.buffer = *stream_buffer,
};
const auto memory_properties = device.GetPhysical().GetMemoryProperties();
stream_memory = dev.AllocateMemory(VkMemoryAllocateInfo{
.sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO,
.pNext = make_dedicated ? &dedicated_info : nullptr,
.allocationSize = requirements.size,
.memoryTypeIndex = FindMemoryTypeIndex(memory_properties, requirements.memoryTypeBits),
});
if (device.HasDebuggingToolAttached()) {
stream_memory.SetObjectNameEXT("Stream Buffer Memory");
}
stream_buffer.BindMemory(*stream_memory, 0);
stream_pointer = stream_memory.Map(0, STREAM_BUFFER_SIZE);
}
StagingBufferPool::~StagingBufferPool() = default;
StagingBufferRef StagingBufferPool::Request(size_t size, MemoryUsage usage) {
if (usage == MemoryUsage::Upload && size <= MAX_STREAM_BUFFER_REQUEST_SIZE) {
return GetStreamBuffer(size);
}
if (const std::optional<StagingBufferRef> ref = TryGetReservedBuffer(size, usage)) {
return *ref;
}
return CreateStagingBuffer(size, usage);
}
void StagingBufferPool::TickFrame() {
current_delete_level = (current_delete_level + 1) % NUM_LEVELS;
ReleaseCache(MemoryUsage::DeviceLocal);
ReleaseCache(MemoryUsage::Upload);
ReleaseCache(MemoryUsage::Download);
}
StagingBufferRef StagingBufferPool::GetStreamBuffer(size_t size) {
for (size_t region = Region(used_iterator), region_end = Region(iterator); region < region_end;
++region) {
sync_ticks[region] = scheduler.CurrentTick();
}
used_iterator = iterator;
for (size_t region = Region(free_iterator) + 1,
region_end = std::min(Region(iterator + size) + 1, NUM_SYNCS);
region < region_end; ++region) {
scheduler.Wait(sync_ticks[region]);
}
if (iterator + size > free_iterator) {
free_iterator = iterator + size;
}
if (iterator + size > STREAM_BUFFER_SIZE) {
for (size_t region = Region(used_iterator); region < NUM_SYNCS; ++region) {
sync_ticks[region] = scheduler.CurrentTick();
}
used_iterator = 0;
iterator = 0;
free_iterator = size;
for (size_t region = 0, region_end = Region(size); region <= region_end; ++region) {
scheduler.Wait(sync_ticks[region]);
}
}
const size_t offset = iterator;
iterator = Common::AlignUp(iterator + size, MAX_ALIGNMENT);
return StagingBufferRef{
.buffer = *stream_buffer,
.offset = static_cast<VkDeviceSize>(offset),
.mapped_span = std::span<u8>(stream_pointer + offset, size),
};
}
std::optional<StagingBufferRef> StagingBufferPool::TryGetReservedBuffer(size_t size,
MemoryUsage usage) {
StagingBuffers& cache_level = GetCache(usage)[Common::Log2Ceil64(size)];
const auto is_free = [this](const StagingBuffer& entry) {
return scheduler.IsFree(entry.tick);
};
auto& entries = cache_level.entries;
const auto hint_it = entries.begin() + cache_level.iterate_index;
auto it = std::find_if(entries.begin() + cache_level.iterate_index, entries.end(), is_free);
if (it == entries.end()) {
it = std::find_if(entries.begin(), hint_it, is_free);
if (it == hint_it) {
return std::nullopt;
}
}
cache_level.iterate_index = std::distance(entries.begin(), it) + 1;
it->tick = scheduler.CurrentTick();
return it->Ref();
}
StagingBufferRef StagingBufferPool::CreateStagingBuffer(size_t size, MemoryUsage usage) {
const u32 log2 = Common::Log2Ceil64(size);
vk::Buffer buffer = device.GetLogical().CreateBuffer({
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.size = 1ULL << log2,
.usage = VK_BUFFER_USAGE_TRANSFER_SRC_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT |
VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_BUFFER_BIT |
VK_BUFFER_USAGE_INDEX_BUFFER_BIT | VK_BUFFER_USAGE_VERTEX_BUFFER_BIT,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
.queueFamilyIndexCount = 0,
.pQueueFamilyIndices = nullptr,
});
if (device.HasDebuggingToolAttached()) {
++buffer_index;
buffer.SetObjectNameEXT(fmt::format("Staging Buffer {}", buffer_index).c_str());
}
MemoryCommit commit = memory_allocator.Commit(buffer, usage);
const std::span<u8> mapped_span = IsHostVisible(usage) ? commit.Map() : std::span<u8>{};
StagingBuffer& entry = GetCache(usage)[log2].entries.emplace_back(StagingBuffer{
.buffer = std::move(buffer),
.commit = std::move(commit),
.mapped_span = mapped_span,
.tick = scheduler.CurrentTick(),
});
return entry.Ref();
}
StagingBufferPool::StagingBuffersCache& StagingBufferPool::GetCache(MemoryUsage usage) {
switch (usage) {
case MemoryUsage::DeviceLocal:
return device_local_cache;
case MemoryUsage::Upload:
return upload_cache;
case MemoryUsage::Download:
return download_cache;
default:
UNREACHABLE_MSG("Invalid memory usage={}", usage);
return upload_cache;
}
}
void StagingBufferPool::ReleaseCache(MemoryUsage usage) {
ReleaseLevel(GetCache(usage), current_delete_level);
}
void StagingBufferPool::ReleaseLevel(StagingBuffersCache& cache, size_t log2) {
constexpr size_t deletions_per_tick = 16;
auto& staging = cache[log2];
auto& entries = staging.entries;
const size_t old_size = entries.size();
const auto is_deleteable = [this](const StagingBuffer& entry) {
return scheduler.IsFree(entry.tick);
};
const size_t begin_offset = staging.delete_index;
const size_t end_offset = std::min(begin_offset + deletions_per_tick, old_size);
const auto begin = entries.begin() + begin_offset;
const auto end = entries.begin() + end_offset;
entries.erase(std::remove_if(begin, end, is_deleteable), end);
const size_t new_size = entries.size();
staging.delete_index += deletions_per_tick;
if (staging.delete_index >= new_size) {
staging.delete_index = 0;
}
if (staging.iterate_index > new_size) {
staging.iterate_index = 0;
}
}
} // namespace Vulkan