Commit graph

41 commits

Author SHA1 Message Date
ReinUsesLisp 6481d91e4a gl_buffer_cache: Copy to buffers created as STREAM_READ before downloading
After marking buffers as resident, Nvidia's driver seems to take a
slow path. To workaround this issue, copy to a STREAM_READ buffer and
then call GetNamedBufferSubData on it.

This is a temporary solution until we have asynchronous flushing.
2020-06-26 16:58:40 -03:00
ReinUsesLisp 32a2dcd415 buffer_cache: Use buffer methods instead of cache virtual methods 2020-06-24 02:36:14 -03:00
ReinUsesLisp 32485917ba gl_buffer_cache: Mark buffers as resident
Make stream buffer and cached buffers as resident and query their
address. This allows us to use GPU addresses for several proprietary
Nvidia extensions.
2020-06-24 02:36:14 -03:00
Lioncash 4514b80b3e buffer_cache: Eliminate local variable shadowing
We can just make use of the instance in the scope above this one.
2020-06-19 21:55:02 -04:00
Rodrigo Locatti 0bd9bc7201
Merge pull request #4066 from ReinUsesLisp/shared-ptr-buf
buffer_cache: Avoid passing references of shared pointers and misc style changes
2020-06-15 22:29:32 -03:00
ReinUsesLisp 6508cdd003 buffer_cache: Avoid passing references of shared pointers and misc style changes
Instead of using as template argument a shared pointer, use the
underlying type and manage shared pointers explicitly. This can make
removing shared pointers from the cache more easy.

While we are at it, make some misc style changes and general
improvements (like insert_or_assign instead of operator[] + operator=).
2020-06-09 18:30:49 -03:00
ReinUsesLisp 6e122f0b2c buffer_cache: Return stream buffer invalidation in Map instead of Unmap
We have to invalidate whatever cache is being used before uploading the
data, hence it makes more sense to return this on Map instead of Unmap.
2020-06-08 20:22:31 -03:00
ReinUsesLisp 3b2dee88e6 buffer_cache: Avoid copying twice on certain cases
Avoid copying to a staging buffer on non-granular memory addresses.
Add a callable argument to StreamBufferUpload to be able to copy to the
staging buffer directly from ReadBlockUnsafe.
2020-05-27 23:05:50 -03:00
ReinUsesLisp ebaace294f buffer_cache: Remove unused boost headers 2020-05-21 16:44:00 -03:00
ReinUsesLisp a2dcc642c1 map_interval: Add interval allocator and drop hack
Drop the std::list hack to allocate memory indefinitely.

Instead use a custom allocator that keeps references valid until
destruction. This allocates fixed chunks of memory and puts pointers in
a free list. When an allocation is no longer used put it back to the
free list, this doesn't heap allocate because std::vector doesn't change
the capacity. If the free list is empty, allocate a new chunk.
2020-05-21 16:44:00 -03:00
ReinUsesLisp 19d4f28001 buffer_cache: Use boost::container::small_vector for maps in range
Most overlaps in the buffer cache only contain one mapped address.
We can avoid close to all heap allocations once the buffer cache is
warmed up by using a small_vector with a stack size of one.
2020-05-21 16:44:00 -03:00
ReinUsesLisp 891236124c buffer_cache: Use boost::intrusive::set for caching
Instead of using boost::icl::interval_map for caching, use
boost::intrusive::set. interval_map is intended as a container where the
keys can overlap with one another; we don't need this for caching
buffers and a std::set-like data structure that allows us to search with
lower_bound is enough.
2020-05-21 16:44:00 -03:00
ReinUsesLisp 3b0baf746e buffer_cache: Remove shared pointers
Removing shared pointers is a first step to be able to use intrusive
objects and keep allocations close to one another in memory.
2020-05-21 16:02:54 -03:00
ReinUsesLisp 599274e3f0 buffer_cache: Minor style changes
Minor style changes. Mostly done so I avoid editing it while doing other
changes.
2020-05-21 16:02:20 -03:00
bunnei 2aff0b4733
Merge pull request #3808 from ReinUsesLisp/wait-for-idle
{maxwell_3d,buffer_cache}: Implement memory barriers using 3D registers
2020-05-03 02:43:18 -04:00
Lioncash 6c53edd4d3 vulkan: Remove unnecessary includes
Reduces some header churn and reduces rebuilds when some header
internals change.

While we're at it we can also resolve a missing include in buffer_cache.
2020-04-28 21:54:46 -04:00
ReinUsesLisp fe931ac976 {maxwell_3d,buffer_cache}: Implement memory barriers using 3D registers
Drop MemoryBarrier from the buffer cache and use Maxwell3D's register
WaitForIdle.

To implement this on OpenGL we just call glMemoryBarrier with the
necessary bits.

Vulkan lacks this synchronization primitive, so we set an event and
immediately wait for it. This is not a pretty solution, but it's what
Vulkan can do without submitting the current command buffer to the queue
(which ends up being more expensive on the CPU).
2020-04-28 02:18:12 -03:00
Fernando Sahmkow 4e37f1b113 Address Feedback. 2020-04-22 11:36:27 -04:00
Fernando Sahmkow f616dc0b59 Address Feedback. 2020-04-22 11:36:24 -04:00
ReinUsesLisp b752faf2d3 vk_fence_manager: Initial implementation 2020-04-22 11:36:19 -04:00
Fernando Sahmkow b7bc3c2549 FenceManager: Manage syncpoints and rename fences to semaphores. 2020-04-22 11:36:16 -04:00
Fernando Sahmkow 96bb961a64 BufferCache: Refactor async managing. 2020-04-22 11:36:15 -04:00
Fernando Sahmkow b10db7e4a5 FenceManager: Implement async buffer cache flushes on High settings 2020-04-22 11:36:15 -04:00
Fernando Sahmkow 165ae823f5 ThreadManager: Sync async reads on accurate gpu. 2020-04-22 11:36:12 -04:00
Fernando Sahmkow 8b1eb44b3e BufferCache: Implement OnCPUWrite and SyncGuestHost 2020-04-22 11:36:07 -04:00
ReinUsesLisp 090fd3fefa buffer_cache: Return handles instead of pointer to handles
The original idea of returning pointers is that handles can be moved.
The problem is that the implementation didn't take that in mind and made
everything harder to work with. This commit drops pointer to handles and
returns the handles themselves. While it is still true that handles can
be invalidated, this way we get an old handle instead of a dangling
pointer.

This problem can be solved in the future with sparse buffers.
2020-04-16 02:33:34 -03:00
Fernando Sahmkow 1c18dc6577 Memory: Correct GCC errors. 2020-04-08 18:09:16 -04:00
Fernando Sahmkow e00d992848 GPUMemoryManager: Improve safety of memory reads. 2020-04-08 12:08:06 -04:00
Fernando Sahmkow 7fcd0fee6d Buffer Cache: Use vAddr instead of physical memory. 2020-04-06 09:23:06 -04:00
ReinUsesLisp b35449c85d buffer_cache: Delay buffer destructions
Delay buffer destruction some extra frames to avoid destroying buffers
that are still being used from older frames. This happens on Nvidia's
driver with mailbox.
2020-01-29 17:53:11 -03:00
ReinUsesLisp 73aaf365e7
buffer_cache: Remove brace initialized for objects with default constructor 2019-11-20 16:00:40 -03:00
Morph 0e8a3bf3e5 buffer_cache: Add missing includes (#3079)
`boost::make_iterator_range` is available when `boost/range/iterator_range.hpp` is included.
Also include `boost/icl/interval_map.hpp` and `boost/icl/interval_set.hpp`.
2019-11-07 06:25:53 +00:00
ReinUsesLisp 76ca2a5f82
gl_rasterizer: Upload constant buffers with glNamedBufferSubData
Nvidia's OpenGL driver maps gl(Named)BufferSubData with some requirements
to a fast. This path has an extra memcpy but updates the buffer without
orphaning or waiting for previous calls. It can be seen as a better
model for "push constants" that can upload a whole UBO instead of 256
bytes.

This path has some requirements established here:
http://on-demand.gputechconf.com/gtc/2014/presentations/S4379-opengl-44-scene-rendering-techniques.pdf#page=24

Instead of using the stream buffer, this commits moves constant buffers
uploads to calls of glNamedBufferSubData and from my testing it brings a
performance improvement. This is disabled when the vendor is not Nvidia
since it brings performance regressions.
2019-11-02 05:05:34 -03:00
Rodrigo Locatti 4d4f9cc104 video_core: Silent miscellaneous warnings (#2820)
* texture_cache/surface_params: Remove unused local variable

* rasterizer_interface: Add missing documentation commentary

* maxwell_dma: Remove unused rasterizer reference

* video_core/gpu: Sort member declaration order to silent -Wreorder warning

* fermi_2d: Remove unused MemoryManager reference

* video_core: Silent unused variable warnings

* buffer_cache: Silent -Wreorder warnings

* kepler_memory: Remove unused MemoryManager reference

* gl_texture_cache: Add missing override

* buffer_cache: Add missing include

* shader/decode: Remove unused variables
2019-08-30 14:08:00 -04:00
Fernando Sahmkow 83ec2091c1 Buffer Cache: Adress Feedback. 2019-08-21 12:14:27 -04:00
Fernando Sahmkow 6ce2c85047 Buffer_Cache: Implement flushing. 2019-08-21 12:14:26 -04:00
Fernando Sahmkow de8ff8a1c6 Buffer_Cache: Implement barriers. 2019-08-21 12:14:25 -04:00
Fernando Sahmkow 286f4c446a Buffer_Cache: Optimize and track written areas. 2019-08-21 12:14:25 -04:00
Fernando Sahmkow 5f4b746a1e BufferCache: Rework mapping caching. 2019-08-21 12:14:24 -04:00
Fernando Sahmkow 86d8563314 Buffer_Cache: Fixes and optimizations. 2019-08-21 12:14:23 -04:00
Fernando Sahmkow 862bec001b Video_Core: Implement a new Buffer Cache 2019-08-21 12:14:22 -04:00