gstreamer/subprojects/gst-plugins-bad/sys/nvcodec/plugin.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

283 lines
8.9 KiB
C
Raw Normal View History

/*
* Copyright (C) 2017 Ericsson AB. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer
* in the documentation and/or other materials provided with the
* distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "gstnvdec.h"
#include "gstnvenc.h"
#include "gstnvav1dec.h"
#include "gstnvh264dec.h"
#include "gstnvh265dec.h"
#include "gstnvvp8dec.h"
#include "gstnvvp9dec.h"
#include "gstnvdecoder.h"
#include "gstcudamemorycopy.h"
#include "gstcudafilter.h"
#include <gst/cuda/gstcudamemory.h>
#ifdef HAVE_NVCODEC_NVMM
#include "gstcudanvmm.h"
#endif
#ifdef G_OS_WIN32
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>
2022-03-15 19:58:16 +00:00
#include <gst/d3d11/gstd3d11.h>
#endif
#include "gstnvh264encoder.h"
#include "gstnvh265encoder.h"
#include "gstcudaipcsink.h"
#include "gstcudaipcsrc.h"
#include "gstnvcodecutils.h"
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>
2022-03-15 19:58:16 +00:00
GST_DEBUG_CATEGORY (gst_nvcodec_debug);
GST_DEBUG_CATEGORY (gst_nvdec_debug);
GST_DEBUG_CATEGORY (gst_nvenc_debug);
GST_DEBUG_CATEGORY (gst_nv_decoder_debug);
#ifdef HAVE_NVCODEC_NVMM
GST_DEBUG_CATEGORY (gst_cuda_nvmm_debug);
#endif
#define GST_CAT_DEFAULT gst_nvcodec_debug
static void
plugin_deinit (gpointer data)
{
gst_cuda_ipc_client_deinit ();
}
static gboolean
plugin_init (GstPlugin * plugin)
{
CUresult cuda_ret;
const char *err_name = NULL, *err_desc = NULL;
gint dev_count = 0;
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>
2022-03-15 19:58:16 +00:00
guint i;
gboolean nvdec_available = TRUE;
gboolean nvenc_available = TRUE;
/* hardcoded minimum supported version */
guint api_major_ver = 8;
guint api_minor_ver = 1;
GList *h264_enc_cdata = NULL;
GList *h265_enc_cdata = NULL;
GST_DEBUG_CATEGORY_INIT (gst_nvcodec_debug, "nvcodec", 0, "nvcodec");
GST_DEBUG_CATEGORY_INIT (gst_nvdec_debug, "nvdec", 0, "nvdec");
GST_DEBUG_CATEGORY_INIT (gst_nvenc_debug, "nvenc", 0, "nvenc");
GST_DEBUG_CATEGORY_INIT (gst_nv_decoder_debug, "nvdecoder", 0, "nvdecoder");
#ifdef HAVE_NVCODEC_NVMM
GST_DEBUG_CATEGORY_INIT (gst_cuda_nvmm_debug, "cudanvmm", 0, "cudanvmm");
#endif
if (!gst_cuda_load_library ()) {
GST_WARNING ("Failed to load cuda library");
return TRUE;
}
/* get available API version from nvenc and it will be passed to
* nvdec */
if (!gst_nvenc_load_library (&api_major_ver, &api_minor_ver)) {
GST_WARNING ("Failed to load nvenc library");
nvenc_available = FALSE;
}
if (!gst_cuvid_load_library (api_major_ver, api_minor_ver)) {
GST_WARNING ("Failed to load nvdec library version %u.%u", api_major_ver,
api_minor_ver);
nvdec_available = FALSE;
}
if (!nvdec_available && !nvenc_available)
return TRUE;
cuda_ret = CuInit (0);
if (cuda_ret != CUDA_SUCCESS) {
CuGetErrorName (cuda_ret, &err_name);
CuGetErrorString (cuda_ret, &err_desc);
GST_ERROR ("Failed to init cuda, cuInit ret: 0x%x: %s: %s",
(int) cuda_ret, err_name, err_desc);
return TRUE;
}
cuda_ret = CuDeviceGetCount (&dev_count);
if (cuda_ret != CUDA_SUCCESS || !dev_count) {
CuGetErrorName (cuda_ret, &err_name);
CuGetErrorString (cuda_ret, &err_desc);
GST_ERROR ("No available device, cuDeviceGetCount ret: 0x%x: %s %s",
(int) cuda_ret, err_name, err_desc);
return TRUE;
}
for (i = 0; i < dev_count; i++) {
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>
2022-03-15 19:58:16 +00:00
GstCudaContext *context = gst_cuda_context_new (i);
CUcontext cuda_ctx;
gint64 adapter_luid = 0;
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>
2022-03-15 19:58:16 +00:00
if (!context) {
GST_WARNING ("Failed to create context for device %d", i);
continue;
}
#ifdef G_OS_WIN32
g_object_get (context, "dxgi-adapter-luid", &adapter_luid, NULL);
#endif
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>
2022-03-15 19:58:16 +00:00
cuda_ctx = gst_cuda_context_get_handle (context);
if (nvdec_available) {
gint j;
for (j = 0; j < cudaVideoCodec_NumCodecs; j++) {
GstCaps *sink_template = NULL;
GstCaps *src_template = NULL;
cudaVideoCodec codec = (cudaVideoCodec) j;
gboolean register_cuviddec = FALSE;
if (gst_nv_decoder_check_device_caps (cuda_ctx,
codec, &sink_template, &src_template)) {
const gchar *codec_name = gst_cuda_video_codec_to_string (codec);
GST_INFO ("CUDA video codec %s, sink template %" GST_PTR_FORMAT
"src template %" GST_PTR_FORMAT, codec_name,
sink_template, src_template);
switch (codec) {
case cudaVideoCodec_H264:
/* higher than avdec_h264 */
gst_nv_h264_dec_register (plugin, i, adapter_luid,
GST_RANK_PRIMARY + 1, sink_template, src_template);
break;
case cudaVideoCodec_HEVC:
/* higher than avdec_h265 */
gst_nv_h265_dec_register (plugin, i, adapter_luid,
GST_RANK_PRIMARY + 1, sink_template, src_template);
break;
case cudaVideoCodec_VP8:
gst_nv_vp8_dec_register (plugin, i, adapter_luid,
GST_RANK_PRIMARY, sink_template, src_template);
break;
case cudaVideoCodec_VP9:
gst_nv_vp9_dec_register (plugin, i, adapter_luid,
GST_RANK_PRIMARY, sink_template, src_template);
break;
case cudaVideoCodec_AV1:
/* rust dav1ddec has "primary" rank */
gst_nv_av1_dec_register (plugin, i, adapter_luid,
GST_RANK_PRIMARY + 1, sink_template, src_template);
break;
default:
register_cuviddec = TRUE;
break;
}
if (register_cuviddec) {
gst_nvdec_plugin_init (plugin,
i, codec, codec_name, sink_template, src_template);
}
gst_caps_unref (sink_template);
gst_caps_unref (src_template);
}
}
}
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>
2022-03-15 19:58:16 +00:00
if (nvenc_available) {
GstNvEncoderClassData *cdata;
#ifdef G_OS_WIN32
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>
2022-03-15 19:58:16 +00:00
if (g_win32_check_windows_version (6, 0, 0, G_WIN32_OS_ANY)) {
GstD3D11Device *d3d11_device;
d3d11_device = gst_d3d11_device_new_for_adapter_luid (adapter_luid,
D3D11_CREATE_DEVICE_BGRA_SUPPORT);
if (!d3d11_device) {
GST_WARNING ("Failed to d3d11 create device");
} else {
cdata = gst_nv_h264_encoder_register_d3d11 (plugin,
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>
2022-03-15 19:58:16 +00:00
d3d11_device, GST_RANK_NONE);
if (cdata)
h264_enc_cdata = g_list_append (h264_enc_cdata, cdata);
cdata = gst_nv_h265_encoder_register_d3d11 (plugin,
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>
2022-03-15 19:58:16 +00:00
d3d11_device, GST_RANK_NONE);
if (cdata)
h265_enc_cdata = g_list_append (h265_enc_cdata, cdata);
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>
2022-03-15 19:58:16 +00:00
gst_object_unref (d3d11_device);
}
}
#endif
cdata =
gst_nv_h264_encoder_register_cuda (plugin, context, GST_RANK_NONE);
if (cdata)
h264_enc_cdata = g_list_append (h264_enc_cdata, cdata);
cdata =
gst_nv_h265_encoder_register_cuda (plugin, context, GST_RANK_NONE);
if (cdata)
h265_enc_cdata = g_list_append (h265_enc_cdata, cdata);
gst_nvenc_plugin_init (plugin, i, cuda_ctx);
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>
2022-03-15 19:58:16 +00:00
}
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>
2022-03-15 19:58:16 +00:00
gst_object_unref (context);
}
if (h264_enc_cdata) {
gst_nv_h264_encoder_register_auto_select (plugin, h264_enc_cdata,
GST_RANK_NONE);
}
if (h265_enc_cdata) {
gst_nv_h265_encoder_register_auto_select (plugin, h265_enc_cdata,
GST_RANK_NONE);
}
gst_cuda_memory_copy_register (plugin, GST_RANK_NONE);
gst_cuda_filter_plugin_init (plugin);
gst_element_register (plugin,
"cudaipcsink", GST_RANK_NONE, GST_TYPE_CUDA_IPC_SINK);
gst_element_register (plugin,
"cudaipcsrc", GST_RANK_NONE, GST_TYPE_CUDA_IPC_SRC);
gst_cuda_memory_init_once ();
#ifdef HAVE_NVCODEC_NVMM
if (gst_cuda_nvmm_init_once ()) {
GST_INFO ("Enable NVMM support");
}
#endif
g_object_set_data_full (G_OBJECT (plugin),
"plugin-nvcodec-shutdown", (gpointer) "shutdown-data",
(GDestroyNotify) plugin_deinit);
return TRUE;
}
GST_PLUGIN_DEFINE (GST_VERSION_MAJOR, GST_VERSION_MINOR, nvcodec,
"GStreamer NVCODEC plugin", plugin_init, VERSION, "LGPL",
GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)