1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
|
// Copyright 2019 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#include <cstring>
#include <memory>
#include <optional>
#include <tuple>
#include "common/alignment.h"
#include "common/assert.h"
#include "core/memory.h"
#include "video_core/renderer_vulkan/declarations.h"
#include "video_core/renderer_vulkan/vk_buffer_cache.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_stream_buffer.h"
namespace Vulkan {
CachedBufferEntry::CachedBufferEntry(VAddr cpu_addr, std::size_t size, u64 offset,
std::size_t alignment, u8* host_ptr)
: RasterizerCacheObject{host_ptr}, cpu_addr{cpu_addr}, size{size}, offset{offset},
alignment{alignment} {}
VKBufferCache::VKBufferCache(Tegra::MemoryManager& tegra_memory_manager,
VideoCore::RasterizerInterface& rasterizer, const VKDevice& device,
VKMemoryManager& memory_manager, VKScheduler& scheduler, u64 size)
: RasterizerCache{rasterizer}, tegra_memory_manager{tegra_memory_manager} {
const auto usage = vk::BufferUsageFlagBits::eVertexBuffer |
vk::BufferUsageFlagBits::eIndexBuffer |
vk::BufferUsageFlagBits::eUniformBuffer;
const auto access = vk::AccessFlagBits::eVertexAttributeRead | vk::AccessFlagBits::eIndexRead |
vk::AccessFlagBits::eUniformRead;
stream_buffer =
std::make_unique<VKStreamBuffer>(device, memory_manager, scheduler, size, usage, access,
vk::PipelineStageFlagBits::eAllCommands);
buffer_handle = stream_buffer->GetBuffer();
}
VKBufferCache::~VKBufferCache() = default;
u64 VKBufferCache::UploadMemory(GPUVAddr gpu_addr, std::size_t size, u64 alignment, bool cache) {
const auto cpu_addr{tegra_memory_manager.GpuToCpuAddress(gpu_addr)};
ASSERT_MSG(cpu_addr, "Invalid GPU address");
// Cache management is a big overhead, so only cache entries with a given size.
// TODO: Figure out which size is the best for given games.
cache &= size >= 2048;
const auto& host_ptr{Memory::GetPointer(*cpu_addr)};
if (cache) {
auto entry = TryGet(host_ptr);
if (entry) {
if (entry->GetSize() >= size && entry->GetAlignment() == alignment) {
return entry->GetOffset();
}
Unregister(entry);
}
}
AlignBuffer(alignment);
const u64 uploaded_offset = buffer_offset;
if (!host_ptr) {
return uploaded_offset;
}
std::memcpy(buffer_ptr, host_ptr, size);
buffer_ptr += size;
buffer_offset += size;
if (cache) {
auto entry = std::make_shared<CachedBufferEntry>(*cpu_addr, size, uploaded_offset,
alignment, host_ptr);
Register(entry);
}
return uploaded_offset;
}
u64 VKBufferCache::UploadHostMemory(const u8* raw_pointer, std::size_t size, u64 alignment) {
AlignBuffer(alignment);
std::memcpy(buffer_ptr, raw_pointer, size);
const u64 uploaded_offset = buffer_offset;
buffer_ptr += size;
buffer_offset += size;
return uploaded_offset;
}
std::tuple<u8*, u64> VKBufferCache::ReserveMemory(std::size_t size, u64 alignment) {
AlignBuffer(alignment);
u8* const uploaded_ptr = buffer_ptr;
const u64 uploaded_offset = buffer_offset;
buffer_ptr += size;
buffer_offset += size;
return {uploaded_ptr, uploaded_offset};
}
void VKBufferCache::Reserve(std::size_t max_size) {
bool invalidate;
std::tie(buffer_ptr, buffer_offset_base, invalidate) = stream_buffer->Reserve(max_size);
buffer_offset = buffer_offset_base;
if (invalidate) {
InvalidateAll();
}
}
VKExecutionContext VKBufferCache::Send(VKExecutionContext exctx) {
return stream_buffer->Send(exctx, buffer_offset - buffer_offset_base);
}
void VKBufferCache::AlignBuffer(std::size_t alignment) {
// Align the offset, not the mapped pointer
const u64 offset_aligned = Common::AlignUp(buffer_offset, alignment);
buffer_ptr += offset_aligned - buffer_offset;
buffer_offset = offset_aligned;
}
} // namespace Vulkan
|
