Linuxydable/suyu
1
0
Fork 0
forked from suyu/suyu
Code Pull requests Activity
suyu/src/video_core/renderer_vulkan/vk_stream_buffer.h

76 lines
2.4 KiB
C++
Raw Normal View History

vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
// Copyright 2019 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include <optional>
#include <tuple>
#include <vector>
#include "common/common_types.h"
renderer_vulkan: Drop Vulkan-Hpp
2020-03-27 05:33:21 +01:00
#include "video_core/renderer_vulkan/wrapper.h"
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
namespace Vulkan {
class VKDevice;
class VKFence;
class VKFenceWatch;
class VKScheduler;
vk_stream_buffer/vk_buffer_cache: Avoid halting and use generic cache The stream buffer before this commit once it was full (no more bytes to write before looping) waiting for all previous operations to finish. This was a temporary solution and had a noticeable performance penalty in performance (from what a profiler showed). To avoid this mark with fences usages of the stream buffer and once it loops wait for them to be signaled. On average this will never wait. Each fence knows where its usage finishes, resulting in a non-paged stream buffer. On the other side, the buffer cache is reimplemented using the generic buffer cache. It makes use of the staging buffer pool and the new stream buffer.
2020-01-06 21:59:20 +01:00
class VKStreamBuffer final {
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
public:
vk_stream_buffer/vk_buffer_cache: Avoid halting and use generic cache The stream buffer before this commit once it was full (no more bytes to write before looping) waiting for all previous operations to finish. This was a temporary solution and had a noticeable performance penalty in performance (from what a profiler showed). To avoid this mark with fences usages of the stream buffer and once it loops wait for them to be signaled. On average this will never wait. Each fence knows where its usage finishes, resulting in a non-paged stream buffer. On the other side, the buffer cache is reimplemented using the generic buffer cache. It makes use of the staging buffer pool and the new stream buffer.
2020-01-06 21:59:20 +01:00
explicit VKStreamBuffer(const VKDevice& device, VKScheduler& scheduler,
renderer_vulkan: Drop Vulkan-Hpp
2020-03-27 05:33:21 +01:00
VkBufferUsageFlags usage);
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
~VKStreamBuffer();
/**
* Reserves a region of memory from the stream buffer.
* @param size Size to reserve.
* @returns A tuple in the following order: Raw memory pointer (with offset added), buffer
vk_stream_buffer: Remove copy code path
2019-02-26 06:09:43 +01:00
* offset and a boolean that's true when buffer has been invalidated.
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
*/
vk_stream_buffer/vk_buffer_cache: Avoid halting and use generic cache The stream buffer before this commit once it was full (no more bytes to write before looping) waiting for all previous operations to finish. This was a temporary solution and had a noticeable performance penalty in performance (from what a profiler showed). To avoid this mark with fences usages of the stream buffer and once it loops wait for them to be signaled. On average this will never wait. Each fence knows where its usage finishes, resulting in a non-paged stream buffer. On the other side, the buffer cache is reimplemented using the generic buffer cache. It makes use of the staging buffer pool and the new stream buffer.
2020-01-06 21:59:20 +01:00
std::tuple<u8*, u64, bool> Map(u64 size, u64 alignment);
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
/// Ensures that "size" bytes of memory are available to the GPU, potentially recording a copy.
vk_stream_buffer/vk_buffer_cache: Avoid halting and use generic cache The stream buffer before this commit once it was full (no more bytes to write before looping) waiting for all previous operations to finish. This was a temporary solution and had a noticeable performance penalty in performance (from what a profiler showed). To avoid this mark with fences usages of the stream buffer and once it loops wait for them to be signaled. On average this will never wait. Each fence knows where its usage finishes, resulting in a non-paged stream buffer. On the other side, the buffer cache is reimplemented using the generic buffer cache. It makes use of the staging buffer pool and the new stream buffer.
2020-01-06 21:59:20 +01:00
void Unmap(u64 size);
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
renderer_vulkan: Drop Vulkan-Hpp
2020-03-27 05:33:21 +01:00
VkBuffer GetHandle() const {
vk_stream_buffer: Remove copy code path
2019-02-26 06:09:43 +01:00
return *buffer;
}
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
private:
vk_stream_buffer/vk_buffer_cache: Avoid halting and use generic cache The stream buffer before this commit once it was full (no more bytes to write before looping) waiting for all previous operations to finish. This was a temporary solution and had a noticeable performance penalty in performance (from what a profiler showed). To avoid this mark with fences usages of the stream buffer and once it loops wait for them to be signaled. On average this will never wait. Each fence knows where its usage finishes, resulting in a non-paged stream buffer. On the other side, the buffer cache is reimplemented using the generic buffer cache. It makes use of the staging buffer pool and the new stream buffer.
2020-01-06 21:59:20 +01:00
struct Watch final {
VKFenceWatch fence;
u64 upper_bound{};
};
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
/// Creates Vulkan buffer handles committing the required the required memory.
renderer_vulkan: Drop Vulkan-Hpp
2020-03-27 05:33:21 +01:00
void CreateBuffers(VkBufferUsageFlags usage);
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
/// Increases the amount of watches available.
vk_stream_buffer/vk_buffer_cache: Avoid halting and use generic cache The stream buffer before this commit once it was full (no more bytes to write before looping) waiting for all previous operations to finish. This was a temporary solution and had a noticeable performance penalty in performance (from what a profiler showed). To avoid this mark with fences usages of the stream buffer and once it loops wait for them to be signaled. On average this will never wait. Each fence knows where its usage finishes, resulting in a non-paged stream buffer. On the other side, the buffer cache is reimplemented using the generic buffer cache. It makes use of the staging buffer pool and the new stream buffer.
2020-01-06 21:59:20 +01:00
void ReserveWatches(std::vector<Watch>& watches, std::size_t grow_size);
void WaitPendingOperations(u64 requested_upper_bound);
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
renderer_vulkan: Drop Vulkan-Hpp
2020-03-27 05:33:21 +01:00
const VKDevice& device; ///< Vulkan device manager.
VKScheduler& scheduler; ///< Command scheduler.
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
vk_stream_buffer: Fix out of memory on boot on recent Nvidia drivers Nvidia recently introduced a new memory type for data streaming (awesome!), but yuzu was assuming that all heaps had enough memory for the assumed stream buffer size (256 MiB). This worked fine on AMD but Nvidia's new memory heap was smaller than 256 MiB. This commit changes this assumption and allocates a bit less than the size of the preferred heap, with a maximum of 256 MiB (to avoid allocating all system memory on integrated devices). - Fixes a crash on NVIDIA 450.82.0.0
2020-04-17 10:13:35 +02:00
vk::Buffer buffer; ///< Mapped buffer.
vk::DeviceMemory memory; ///< Memory allocation.
u64 stream_buffer_size{}; ///< Stream buffer size.
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
u64 offset{}; ///< Buffer iterator.
u64 mapped_size{}; ///< Size reserved for the current copy.
vk_stream_buffer/vk_buffer_cache: Avoid halting and use generic cache The stream buffer before this commit once it was full (no more bytes to write before looping) waiting for all previous operations to finish. This was a temporary solution and had a noticeable performance penalty in performance (from what a profiler showed). To avoid this mark with fences usages of the stream buffer and once it loops wait for them to be signaled. On average this will never wait. Each fence knows where its usage finishes, resulting in a non-paged stream buffer. On the other side, the buffer cache is reimplemented using the generic buffer cache. It makes use of the staging buffer pool and the new stream buffer.
2020-01-06 21:59:20 +01:00
std::vector<Watch> current_watches; ///< Watches recorded in the current iteration.
std::size_t current_watch_cursor{}; ///< Count of watches, reset on invalidation.
std::optional<std::size_t> invalidation_mark; ///< Number of watches used in the previous cycle.
std::vector<Watch> previous_watches; ///< Watches used in the previous iteration.
std::size_t wait_cursor{}; ///< Last watch being waited for completion.
u64 wait_bound{}; ///< Highest offset being watched for completion.
vk_stream_buffer: Implement a stream buffer This manages two kinds of streaming buffers: one for unified memory models and one for dedicated GPUs. The first one skips the copy from the staging buffer to the real buffer, since it creates an unified buffer. This implementation waits for all fences to finish their operation before "invalidating". This is suboptimal since it should allocate another buffer or start searching from the beginning. There is room for improvement here. This could also handle AMD's "pinned" memory (a heap with 256 MiB) that seems to be designed for buffer streaming.
2019-02-24 08:22:33 +01:00
};
} // namespace Vulkan
Copy permalink