Commit graph

70 commits

Author SHA1 Message Date
Valeri
0394e4bb8e
vic: Allow surface to be higher than frame
Touhou Genso Wanderer Lotus Labyrinth R decodes 1920x1080 videos into 1920x1088 surface.
Only allow mismatch for height, since larger width would result in increasingly offset rows and somewhat defeat entire purpose of this check.
2021-10-09 20:22:09 +03:00
ameerj
403fc86c11 vic: Avoid memory corruption when multiple streams with different dimensions are decoded
This is a work around to avoid buffer overflow errors until multi channel/multi stream decoding is supported.
2021-10-08 01:22:38 -04:00
ameerj
5aae61775f vic: Refactor frame writing methods 2021-10-07 14:56:44 -04:00
ameerj
899fdb9c44 vic: Implement RGBX frame format 2021-10-07 11:06:57 -04:00
Morph
ae028ddf22 codec: Add missing <string_view> include 2021-09-11 17:19:14 -04:00
Fernando S
be4e192903
Merge pull request #6846 from ameerj/nvdec-gpu-decode
nvdec: Add GPU video decoding for all capable drivers and platforms
2021-09-11 23:11:32 +02:00
ameerj
eb2624ed65 vp9_types: Minor refactor of VP9 info structs. 2021-08-25 21:42:43 -04:00
ameerj
3de38c9a70 vp9_types: Remove unused Vp9PictureInfo members 2021-08-25 21:29:22 -04:00
ameerj
b384129c63 h264: Lower max_num_ref_frames
GPU decoding seems to be more picky when it comes to the maximum number of reference frames.
2021-08-16 14:40:53 -04:00
ameerj
cd016d3cb5 configure_graphics: Add GPU nvdec decoding as an option
Some system configurations may see visual regressions or lower performance using GPU decoding compared to CPU decoding. This setting provides the option for users to specify their decoding preference.

Co-Authored-By: yzct12345 <87620833+yzct12345@users.noreply.github.com>
2021-08-16 14:40:53 -04:00
ameerj
a832aa699f codec: Improve libav memory alloc and cleanup 2021-08-16 14:40:53 -04:00
ameerj
bc3efb79cc codec: Fallback to CPU decoding if no compatible GPU format is found 2021-08-16 14:40:53 -04:00
bunnei
0509fe3377
Merge pull request #6838 from ameerj/sws-align
vic: Specify sws_scale height stride.
2021-08-12 11:28:33 -07:00
ameerj
356e10898f codec: Replace deprecated av_init_packet usage 2021-08-12 01:28:01 -04:00
ameerj
659039ca6d nvdec: Implement GPU accelerated decoding for all platforms
Supplements the VAAPI intel gpu decoder by implementing the D3D11VA decoder for Windows, and CUVID/VDPAU for Nvidia and AMD on drivers linux respectively.
2021-08-12 01:28:01 -04:00
ameerj
a779cede7c vic: Specify sws_scale height stride.
Silences a sws_scale runtime warning about unaligned strides.
2021-08-09 23:24:16 -04:00
ameerj
fa22695705 vp9: Ensure the first frame is complete
Silences a runtime error due to the first frame missing the frame data, and being set to hidden despite being a key-frame.
2021-08-08 13:49:00 -04:00
ameerj
928b64d2ce nvdec: Better logging for unimplemented codecs 2021-08-07 01:08:33 -04:00
bunnei
f183668a87
Merge pull request #6799 from ameerj/vp9-fixes
nvdec: Fix VP9 reference frame refreshes
2021-08-06 17:46:46 -07:00
ameerj
e3688f0627 vp9: Cleanup unused variables
With reference frames refreshes fix, we no longer need to buffer two frames in advance.
We can also remove other unused or otherwise unneeded variables.
2021-08-06 20:08:11 -04:00
ameerj
a3f80a97a3 vp9: Fix reference frame refreshes
This resolves the artifacting when decoding VP9 streams.
2021-08-06 20:08:08 -04:00
yzct12345
2868d4ba84
nvdec: Implement VA-API hardware video acceleration (#6713)
* nvdec: VA-API

* Verify formatting

* Forgot a semicolon for Windows

* Clarify comment about AV_PIX_FMT_NV12

* Fix assert log spam from missing negation

* vic: Remove forgotten debug code

* Address lioncash's review

* Mention VA-API is Intel/AMD

* Address v1993's review

* Hopefully fix CMakeLists style this time

* vic: Improve cache locality

* vic: Fix off-by-one error

* codec: Async

* codec: Forgot the GetValue()

* nvdec: Address ameerj's review

* codec: Fallback to CPU without VA-API support

* cmake: Address lat9nq's review

* cmake: Make VA-API optional

* vaapi: Multiple GPU

* Apply suggestions from code review

Co-authored-by: Ameer J <52414509+ameerj@users.noreply.github.com>

* nvdec: Address ameerj's review

* codec: Use anonymous instead of static

* nvdec: Remove enum and fix memory leak

* nvdec: Address ameerj's review

* codec: Remove preparation for threading

Co-authored-by: Ameer J <52414509+ameerj@users.noreply.github.com>
2021-08-03 23:43:11 -04:00
Fernando S
da4ca4f2f9
Merge pull request #6525 from ameerj/nvdec-fixes
nvdec: Fix Submit Ioctl data source, vic frame dimension computations
2021-07-15 15:17:50 +02:00
ameerj
b7fa264749 vic: Fix dimension compuation of YUV frames
Fixes out of bound memory crashes in Mario Golf
2021-07-15 00:51:50 -04:00
bunnei
bf50345d4c
Merge pull request #6537 from Morph1984/warnings
general: Enforce multiple warnings in MSVC
2021-07-05 17:09:23 -07:00
Kelebek1
208a04dcff Slightly refactor NVDEC and codecs for readability and safety 2021-07-01 06:22:05 +01:00
Morph
22d7b89c15 video_core: Remove #pragma warning directives for external headers 2021-06-28 14:21:40 -04:00
ReinUsesLisp
05bd50a1cf codec,vic: Disable warnings in ffmpeg headers 2021-06-26 03:29:31 -03:00
lat9nq
a60653dcd3 vp9: Avoid memcpy with null pointers
Avoid sending null pointer to memcpy as reported by Undefined Behaviour
Sanitizer. Replaces the std::memcpy calls in SpliceVectors with
std::copy calls. Opting to replace all the memcpy's with copy's.

Co-authored-by: LC <mathew1800@gmail.com>
2021-04-05 00:44:38 -04:00
ameerj
b675c44e49 rebase, fix name shadowing, more const 2021-02-13 13:07:56 -05:00
ameerj
09722cb4a7 streamline cdma_pusher/command_classes 2021-02-13 13:07:56 -05:00
ameerj
77564f987c streamline cdma_pusher/command_classes 2021-02-13 13:07:53 -05:00
ameerj
ac265a72ce nvdec cleanup 2021-02-13 13:07:31 -05:00
ReinUsesLisp
82c2601555 video_core: Reimplement the buffer cache
Reimplement the buffer cache using cached bindings and page level
granularity for modification tracking. This also drops the usage of
shared pointers and virtual functions from the cache.

- Bindings are cached, allowing to skip work when the game changes few
  bits between draws.
- OpenGL Assembly shaders no longer copy when a region has been modified
  from the GPU to emulate constant buffers, instead GL_EXT_memory_object
  is used to alias sub-buffers within the same allocation.
- OpenGL Assembly shaders stream constant buffer data using
  glProgramBufferParametersIuivNV, from NV_parameter_buffer_object. In
  theory this should save one hash table resolve inside the driver
  compared to glBufferSubData.
- A new OpenGL stream buffer is implemented based on fences for drivers
  that are not Nvidia's proprietary, due to their low performance on
  partial glBufferSubData calls synchronized with 3D rendering (that
  some games use a lot).
- Most optimizations are shared between APIs now, allowing Vulkan to
  cache more bindings than before, skipping unnecesarry work.

This commit adds the necessary infrastructure to use Vulkan object from
OpenGL. Overall, it improves performance and fixes some bugs present on
the old cache. There are still some edge cases hit by some games that
harm performance on some vendors, this are planned to be fixed in later
commits.
2021-02-13 02:17:22 -03:00
Lioncash
8620de6b20 common/bit_util: Replace CLZ/CTZ operations with standardized ones
Makes for less code that we need to maintain.
2021-01-15 02:15:32 -05:00
Ameer J
16392a23cc remove inaccurate reference
Co-authored-by: LC <mathew1800@gmail.com>
2021-01-07 14:33:45 -05:00
ameerj
06cef3355e fix for nvdec disabled, cleanup host1x 2021-01-07 14:33:45 -05:00
ameerj
2c27127d04 nvdec syncpt incorporation
laying the groundwork for async gpu, although this does not fully implement async nvdec operations
2021-01-07 14:33:45 -05:00
Morph
a745d87971 general: Fix various spelling errors 2021-01-02 10:23:41 -05:00
ReinUsesLisp
9764c13d6d video_core: Rewrite the texture cache
The current texture cache has several points that hurt maintainability
and performance. It's easy to break unrelated parts of the cache
when doing minor changes. The cache can easily forget valuable
information about the cached textures by CPU writes or simply by its
normal usage.The current texture cache has several points that hurt
maintainability and performance. It's easy to break unrelated parts
of the cache when doing minor changes. The cache can easily forget
valuable information about the cached textures by CPU writes or simply
by its normal usage.

This commit aims to address those issues.
2020-12-30 03:38:50 -03:00
Lioncash
4c5f5c9bf3 video_core: Remove unnecessary enum class casting in logging messages
fmt now automatically prints the numeric value of an enum class member
by default, so we don't need to use casts any more.

Reduces the line noise a bit.
2020-12-07 00:41:50 -05:00
Lioncash
414a87a4f4 video_core: Resolve more variable shadowing scenarios pt.2
Migrates the video core code closer to enabling variable shadowing
warnings as errors.

This primarily sorts out shadowing occurrences within the Vulkan code.
2020-12-05 06:39:35 -05:00
bunnei
e6a896c4bd
Merge pull request #5124 from lioncash/video-shadow
video_core: Resolve more variable shadowing scenarios
2020-12-05 00:48:08 -08:00
Lioncash
94af77aa7c codec: Remove deprecated usage of AVCodecContext::refcounted_frames
This was only necessary for use with the
avcodec_decode_video2/avcoded_decode_audio4 APIs which are also
deprecated.

Given we use avcodec_send_packet/avcodec_receive_frame, this isn't
necessary, this is even indicated directly within the FFmpeg API changes
document here on 2017-09-26:

https://github.com/FFmpeg/FFmpeg/blob/master/doc/APIchanges#L410

This prevents our code from breaking whenever we update to a newer
version of FFmpeg in the future if they ever decide to fully remove this
API member.
2020-12-04 16:23:13 -05:00
Lioncash
677a8b208d video_core: Resolve more variable shadowing scenarios
Resolves variable shadowing scenarios up to the end of the OpenGL code
to make it nicer to review. The rest will be resolved in a following
commit.
2020-12-04 16:19:09 -05:00
Lioncash
cf9767c608 vp9/vic: Resolve pessimizing moves
Removes the usage of moves that don't result in behavior different from
a copy, or otherwise would prevent copy elision from occurring.
2020-12-03 12:33:07 -05:00
bunnei
9abb23cd27
Merge pull request #5002 from ameerj/nvdec-frameskip
nvdec: Queue and display all decoded frames, cleanup decoders
2020-12-02 15:55:15 -08:00
comex
4681e1ea9e codec: Fix pragma GCC diagnostic pop missing corresponding push 2020-11-26 16:35:42 -05:00
ameerj
979b602738 Limit queue size to 10 frames
Workaround for ZLA, which seems to decode and queue twice as many frames as it displays.
2020-11-26 14:04:06 -05:00
ameerj
c9e3abe206 Address PR feedback
remove some redundant moves, make deleter match naming guidelines.

Co-Authored-By: LC <712067+lioncash@users.noreply.github.com>
2020-11-26 00:18:26 -05:00