Commit graph

98 commits

Author SHA1 Message Date
Seungha Yang dce45586e8 nvvp9sldec: Increase DPB size to cover render delay
This should've included in the previous MR
https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/987
already, but missed.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2313>
2022-04-27 18:35:07 +00:00
Thibault Saunier 3c959ff53e cudaconvert: Enhance debugging when setting video info fails
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2300>
2022-04-26 19:04:06 +00:00
Seungha Yang 205a23f456 nvcodec: Stop building for non-{Windows,Linux} target
The other platforms, specifically macOS have not been supported already
because this plugin loads so or dll. Moreover, NVIDIA dropped
support for macOS as of CUDA 11.0. See also
https://developer.nvidia.com/nvidia-cuda-toolkit-11_0_0-developer-tools-mac-hosts

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2283>
2022-04-25 15:50:52 +00:00
Corentin Damman df64280874 cuda-converter: fix nvrtc compilation on non-English locale systems
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2268>
2022-04-22 18:57:22 +00:00
Thibault Saunier dcd2ccad59 cuda: Fix typo in gst_cuda_buffery_copy_type_to_string
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1472>
2022-04-19 16:39:43 +00:00
Thibault Saunier c033f8fad2 cuda: Factor out a public GstCUDA library
So applications and elements implemented outside GStreamer can reuse
our infrastructure

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1472>
2022-04-19 16:39:43 +00:00
Thibault Saunier 3549d59725 nvenc: Give more details about errors
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1472>
2022-04-19 16:39:43 +00:00
Corentin Damman c68c40a6bc cudautils: fix critical typo in gst_cuda_buffer_fallback_copy
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2194>
2022-04-16 02:16:22 +00:00
Seungha Yang a0d8133d60 nvencoder: Update property names and default values
Update properties of newly written D3D11/CUDA mode encoders
to be consistent with Intel QSV plugin. And rename "zerolatency"
to "zero-reorder-delay" because previous "zerolatency" may
misleading

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2140>
2022-04-09 14:00:35 +00:00
Seungha Yang 5ed8e416fe nvcodec: Fix typos
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2107>
2022-04-03 22:12:33 +09:00
Seungha Yang 9b5fe81d01 nvencoder: Fix operator overloading
GUID equality check was inverted

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2107>
2022-04-03 22:09:19 +09:00
Seungha Yang 60735deded nvcodec: Add new Direct3D11/CUDA mode encoder implementation
Adding new encoder elements nvd3d11{h264,h265}enc for Direct3D11
input support and re-written nvcuda{h264,h265}enc elements.
Newly writeen elements have some differences compared with old
nv{h264,h265}enc including non-backward compatible changes.

* RGBA is not a supported input format any more:
  New elements will support only YUV formats to avoid implicit conversion
  done by hardware. Ideally it should be done by upstream element
  in order to have more control on it. Moreover, RGBA support can cause
  redundant RGBA -> YUV conversion if multiple encoders are
  used for the same RGBA input
* Subsampled planar format support is dropped:
  I420 and YV12 format are not supported formats for Direct3D11.
  Although it's supported in CUDA mode, it's not a hardware friendly
  memory layout and it will waste GPU memory since UV planes
  will have large padding due to the memory layout requirement of NVENC.
* GL support is dropped: Similar to the RGBA case,
  GL support in encoder would be suboptimal if GL input is
  used by multiple encoders, because each encoder will copy GL memory
  into CUDA memory.
  Upstream cudaupload element can be used for GL <-> CUDA
  interop instead.
* No more pre-allocation of encoder input surfaces. New implementation
  will use input CUDA memory without copy (zero-copy) or
  will copy into a NVENC's input buffer struct in case of
  system memory input.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1997>
2022-03-23 13:48:53 +00:00
Seungha Yang c08ce58753 nvcodec: Move CUDA <-> GL, D3D11, NVMM copy function to utils
This method can be used in other elements as well.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1945>
2022-03-15 21:51:50 +00:00
Seungha Yang b3df58add1 nvh265sldec: Add support for delayed output
Delay 4 frames in case of non-live to improve throughput

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1937>
2022-03-15 21:08:06 +00:00
Seungha Yang e270e2967f nvenc: Fix deadlock because of too strict buffer pool size
The pool size might need to be larger than encoding surface pool size.
Also, because we always copy input frame into internal CUDA memory,
there's no reason to restrict max size of buffer pool.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1939>
2022-03-12 16:44:44 +00:00
Seungha Yang aa476452fb codecs: Rename picture clear functions
Our convention for clear method is gst_clear_foo_bar().

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1897>
2022-03-11 20:20:17 +00:00
Seungha Yang 423111480c nvh265sldec: Always fill SPS/PPS related parameters
Address compare was not a valid approach since it works
only if SPS/PPS id are changed. Otherwise it will always point to
the same address of member variables of h265parser.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1931>
2022-03-11 16:14:14 +00:00
Corentin Damman 1fb3e35708 cudamemorycopy: add D3D11 resource support
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1807>
2022-03-10 18:08:10 +00:00
Corentin Damman 1568db2c3e cudacontext: find associated DXGI Adapter LUID
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1807>
2022-03-10 18:08:10 +00:00
Corentin Damman 895f11401d cudautils: add support of D3D11 resource as Cuda graphics resource type
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1807>
2022-03-10 18:08:10 +00:00
Corentin Damman f76ecf1e63 cudaloader: add D3D11 API
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1807>
2022-03-10 18:08:10 +00:00
Seungha Yang 5d298b98da nvh264dec,nvh265dec: Fix broken key-unit trick and reverse playback
On GstVideoDecoder::{drain,flush}, we send null packet with
CUVID_PKT_ENDOFSTREAM flag to drain out decoder. Which will
reset CUVID parser as well.
To continue decoding after the drain, the next input buffer
should include sequence headers otherwise CUVID parser will
not report any decodeable frame.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1911>
2022-03-10 12:40:05 +00:00
Seungha Yang 496b77e6aa cudamemorycopy: Fix GL resource leak
Clear GL resources on stop()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1896>
2022-03-08 18:02:25 +00:00
Seungha Yang 4aa516f305 cudamemorycopy: Remove texture-target caps field
It's GL specific field, and we can remove it unconditionally

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1865>
2022-03-07 19:05:29 +00:00
Seungha Yang e5132a8508 cudaupload,cudadownload: Add support for dGPU NVMM
Implement NVMM <-> CUDA, GL, SYSTEM memory conversion. Jetson is
not supported yet. Note that NVMM <-> GL interop on Jetson platform
is supported by GstGL

Some example pipelines are:
- Convert NVMM to GstGL memory
  nvv4l2decoder ! "video/x-raw(memory:NVMM)" ! cudadownload ! "video/x-raw(memory:GLMemory)" ! glimagesink

- Upload system memory to NVMM and encode
  video/x-raw,format=NV12 ! cudaupload ! "video/x-raw(memory:NVMM)" ! nvv4l2h264enc

- Convert NVMM to GstCUDA memory and encode
  nvvideoconvert ! "video/x-raw(memory:NVMM)" ! cudaupload ! "video/x-raw(memory:CUDAMemory)" ! nvh264enc

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1865>
2022-03-07 19:05:29 +00:00
Seungha Yang a63d1cf4c4 cudamemorycopy: Fix build when gl is disabled
Fixes: https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/issues/1707
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1868>
2022-03-07 10:22:21 +00:00
Seungha Yang 30ddb0cd97 nvcodec: Add missing null check in context sharing code
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang c31bf2db4d nvcodec: Refactor cudaupload/download and add support for GL memory
* Implement new baseclass GstCudaMemoryCopy to remove duplicated
  cudaupload/download code
* Add support for CUDA <-> GL memory conversion via cudaupload/download

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang 111b2c3f53 nvcodec: Refactor GstCudaMemory abstraction
* Hide GstCudaMemory member variables
* Make GstCudaAllocator object GstCudaContext independent
* Set offset/stride of memory correctly via video meta
* Drop GST_BUFFER_POOL_OPTION_VIDEO_ALIGNMENT support.
  This implementation actually does not support custom alignment
  because we allocate device memory via cuMemAllocPitch
  of which alignment is almost uncontrollable

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang ad0e7fca14 nvcodec: Refactor basetransform subclasses
* cudaupload/download
  - Specify only formats actually we can deal with
    nvcodec elements, not all video formats
  - Supports CUDA output for download and input for upload in order
    to make passthrough possible, like other upload/download elements.
* cudabasetransform
  - Reset conversion element if upstream CUDA memory
    holds different CUDA context and the element can accept it.
    This is the same behavior as corresponding d3d11 filter elements.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang 89bbcf0061 cudabufferpool: Move GstCudaContext object to public member
... so that plugins can understand assosicated cuda context with the
pool.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang c8a1c953d8 cudacontext: Require explicit gpu id instead of auto (-1)
Sync up with GstD3D11Device implementation. The auto stuff should
be handled in context sharing step, not device creation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang 1b9274307a nvcodec: Indent update
... to prevent gnu indent version dependent diffs.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang cec7ac2a1b cudacontext: Remove unnecessary cuInit() call
It's already called at plugin init time

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>
2022-03-06 15:39:04 +00:00
Seungha Yang 28eeba3e18 nvh265sldec: Fix for decoding 12bits stream
We've been exposing main-444-12 profile as a supported profile
in its sinkpad template but not actaully. Adding code to
covert 12 and 16 bits as well.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1612>
2022-02-07 18:27:48 +00:00
Seungha Yang 80bbc71ec0 nvdecoder: Fix for HEVC 4:4:4 format decoding
Map chroma_format_idc == 3 (which means 4:4:4 subsampling) correctly,
also pass coded bitdepth for decoder initialization instead of
inferring it from output format since they can be different.

Fixes: https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/949
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1612>
2022-02-07 18:27:48 +00:00
Seungha Yang 6434d69f8c codecs: Stop claiming constness for refcounted object
It's almost pointless and makes little sense as subclass might
want to modify refcount of the object or so. And all subclasses
are already casting them to non-const version as well.
In a general sense, we need to avoid passing refcounted object
with const qualifier.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1238>
2022-02-06 17:17:22 +00:00
Seungha Yang 34b54ca665 nvdecoder: Fix for display resolution setup
Display resolution should be cropped rect, not coded resolution.
Otherwise decoded output from NVDEC might be wrong.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1617>
2022-02-01 17:58:06 +00:00
Seungha Yang 2c834b5fc7 nvvp9sldec: Drop frames on non-keyframe format change
NVDEC doesn't seem to be able to handle the case

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1407>
2021-12-02 19:19:59 +00:00
Thibault Saunier d0c5778669 doc: Update nvdec documentation
Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/870

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1309>
2021-11-05 13:26:33 +00:00
Mathieu Duponchelle 2f8030d98b nvh264enc: add constrained-baseline to the caps profiles
In practice, when baseline is requested from the encoder it
produces constrained baseline, and it is already reflected
in the profile-iop flags.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1213>
2021-10-21 09:06:08 +00:00
Seungha Yang e322745763 codecs: h264decoder: Use GstFlowReturn everywhere
boolean return value is not sufficient for representing the reason
of error in most cases. For instance, any errors around new_sequence()
would mean negotiation error, not just *ERROR*.
And some subclasses will allocate buffer/memory/surface on new_picture()
but it could be failed because of expected error, likely flushing

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1019>
2021-10-04 20:56:46 +00:00
Seungha Yang 5b405d1585 codecs: h265decoder: Use GstFlowReturn everywhere
boolean return value is not sufficient for representing the reason
of error in most cases. For instance, any errors around new_sequence()
would mean negotiation error, not just *ERROR*.
And some subclasses will allocate buffer/memory/surface on new_picture()
but it could be failed because of expected error, likely flushing

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1019>
2021-10-04 20:56:46 +00:00
Seungha Yang da7f4e0a69 nvcodec: nvh264sldec: Add support for interlaced stream
Implement missing interlaced stream support

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1026>
2021-10-04 08:04:00 +00:00
Seungha Yang ce8965b75b nvcodec: nvh264sldec: Consider additional render delay DPB pictures
At least additional 4 pictures are required

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1026>
2021-10-04 08:04:00 +00:00
Seungha Yang 613d380b97 nvcodec: nvvp9sldec: Fix for VP9 profile2 decoding
Fix for output video format to be selected correctly

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1026>
2021-10-04 08:04:00 +00:00
Seungha Yang 834d6f4031 nvcodec: nv{vp8,vp9}sldec: Implement get_preferred_output_delay()
Equivalent to that of nvh264sldec. Use render delay in case of non-live
pipeline for the better throughput performance.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/987>
2021-10-02 12:11:48 +00:00
Thibault Saunier 019971a3c7 Move files from gst-plugins-bad into the "subprojects/gst-plugins-bad/" subdir 2021-09-24 16:14:36 -03:00