[gnome-remote-desktop] rdp: Split up AVC420 encoding
- From: Jonas Ådahl <jadahl src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [gnome-remote-desktop] rdp: Split up AVC420 encoding
- Date: Thu, 3 Mar 2022 14:23:09 +0000 (UTC)
commit 48ec7559cbf59302904e255b73716449dffbbc2b
Author: Pascal Nowack <Pascal Nowack gmx de>
Date: Thu Jan 13 15:14:29 2022 +0100
rdp: Split up AVC420 encoding
Calling nvEncEncodePicture() does not block until the encoding
operation is done.
Locking the bitstream, on the other hand, is a blocking operation.
This allows doing other tasks before calling nvEncLockBitstream() to
lock the bitstream, which will be done in the next commit.
So, split up the AVC420 encoding to create the opportunity for this
optimization.
src/grd-hwaccel-nvidia.c | 92 ++++++++++++++++++++++++++++++-----------
src/grd-hwaccel-nvidia.h | 7 +++-
src/grd-rdp-graphics-pipeline.c | 9 +++-
3 files changed, 81 insertions(+), 27 deletions(-)
---
diff --git a/src/grd-hwaccel-nvidia.c b/src/grd-hwaccel-nvidia.c
index 7ddc8ed1..a2b88f97 100644
--- a/src/grd-hwaccel-nvidia.c
+++ b/src/grd-hwaccel-nvidia.c
@@ -58,6 +58,9 @@ typedef struct _NvEncEncodeSession
uint16_t enc_height;
NV_ENC_OUTPUT_PTR buffer_out;
+
+ NV_ENC_REGISTERED_PTR registered_resource;
+ NV_ENC_INPUT_PTR mapped_resource;
} NvEncEncodeSession;
struct _GrdHwAccelNvidia
@@ -361,6 +364,7 @@ void
grd_hwaccel_nvidia_free_nvenc_session (GrdHwAccelNvidia *hwaccel_nvidia,
uint32_t encode_session_id)
{
+ NV_ENCODE_API_FUNCTION_LIST *nvenc_api = &hwaccel_nvidia->nvenc_api;
NvEncEncodeSession *encode_session;
NV_ENC_PIC_PARAMS pic_params = {0};
@@ -369,13 +373,25 @@ grd_hwaccel_nvidia_free_nvenc_session (GrdHwAccelNvidia *hwaccel_nvidia,
NULL, (gpointer *) &encode_session))
return;
- hwaccel_nvidia->nvenc_api.nvEncDestroyBitstreamBuffer (encode_session->encoder,
- encode_session->buffer_out);
+ if (encode_session->mapped_resource)
+ {
+ nvenc_api->nvEncUnmapInputResource (encode_session->encoder,
+ encode_session->mapped_resource);
+ encode_session->mapped_resource = NULL;
+ }
+ if (encode_session->registered_resource)
+ {
+ nvenc_api->nvEncUnregisterResource (encode_session->encoder,
+ encode_session->registered_resource);
+ encode_session->registered_resource = NULL;
+ }
+
+ nvenc_api->nvEncDestroyBitstreamBuffer (encode_session->encoder,
+ encode_session->buffer_out);
pic_params.encodePicFlags = NV_ENC_PIC_FLAG_EOS;
- hwaccel_nvidia->nvenc_api.nvEncEncodePicture (encode_session->encoder,
- &pic_params);
- hwaccel_nvidia->nvenc_api.nvEncDestroyEncoder (encode_session->encoder);
+ nvenc_api->nvEncEncodePicture (encode_session->encoder, &pic_params);
+ nvenc_api->nvEncDestroyEncoder (encode_session->encoder);
g_free (encode_session);
}
@@ -389,15 +405,12 @@ grd_hwaccel_nvidia_avc420_encode_bgrx_frame (GrdHwAccelNvidia *hwaccel_nvidia,
uint16_t src_height,
uint16_t aligned_width,
uint16_t aligned_height,
- uint8_t **bitstream,
- uint32_t *bitstream_size,
CUstream cuda_stream)
{
NvEncEncodeSession *encode_session;
NV_ENC_REGISTER_RESOURCE register_res = {0};
NV_ENC_MAP_INPUT_RESOURCE map_input_res = {0};
NV_ENC_PIC_PARAMS pic_params = {0};
- NV_ENC_LOCK_BITSTREAM lock_bitstream = {0};
uint16_t src_stride;
unsigned int grid_dim_x, grid_dim_y, grid_dim_z;
unsigned int block_dim_x, block_dim_y, block_dim_z;
@@ -411,6 +424,9 @@ grd_hwaccel_nvidia_avc420_encode_bgrx_frame (GrdHwAccelNvidia *hwaccel_nvidia,
g_assert (encode_session->enc_width == aligned_width);
g_assert (encode_session->enc_height == aligned_height);
+ g_assert (encode_session->mapped_resource == NULL);
+ g_assert (encode_session->registered_resource == NULL);
+
if (!(*main_view_nv12) &&
!grd_hwaccel_nvidia_alloc_mem (hwaccel_nvidia, main_view_nv12,
aligned_width * (aligned_height + aligned_height / 2)))
@@ -500,32 +516,60 @@ grd_hwaccel_nvidia_avc420_encode_bgrx_frame (GrdHwAccelNvidia *hwaccel_nvidia,
return FALSE;
}
+ encode_session->mapped_resource = map_input_res.mappedResource;
+ encode_session->registered_resource = register_res.registeredResource;
+
+ return TRUE;
+}
+
+gboolean
+grd_hwaccel_nvidia_avc420_retrieve_bitstream (GrdHwAccelNvidia *hwaccel_nvidia,
+ uint32_t encode_session_id,
+ uint8_t **bitstream,
+ uint32_t *bitstream_size)
+{
+ NV_ENCODE_API_FUNCTION_LIST *nvenc_api = &hwaccel_nvidia->nvenc_api;
+ NvEncEncodeSession *encode_session;
+ NV_ENC_LOCK_BITSTREAM lock_bitstream = {0};
+ gboolean success = FALSE;
+
+ if (!g_hash_table_lookup_extended (hwaccel_nvidia->encode_sessions,
+ GUINT_TO_POINTER (encode_session_id),
+ NULL, (gpointer *) &encode_session))
+ g_assert_not_reached ();
+
+ g_assert (encode_session->mapped_resource != NULL);
+ g_assert (encode_session->registered_resource != NULL);
+
lock_bitstream.version = NV_ENC_LOCK_BITSTREAM_VER;
lock_bitstream.outputBitstream = encode_session->buffer_out;
- if (hwaccel_nvidia->nvenc_api.nvEncLockBitstream (
- encode_session->encoder, &lock_bitstream) != NV_ENC_SUCCESS)
+ if (nvenc_api->nvEncLockBitstream (encode_session->encoder,
+ &lock_bitstream) != NV_ENC_SUCCESS)
{
g_warning ("[HWAccel.NVENC] Failed to lock bitstream");
- hwaccel_nvidia->nvenc_api.nvEncUnmapInputResource (encode_session->encoder,
- map_input_res.mappedResource);
- hwaccel_nvidia->nvenc_api.nvEncUnregisterResource (encode_session->encoder,
- register_res.registeredResource);
- return FALSE;
+ goto out;
}
- *bitstream_size = lock_bitstream.bitstreamSizeInBytes;
- *bitstream = g_memdup2 (lock_bitstream.bitstreamBufferPtr, *bitstream_size);
+ if (bitstream_size)
+ *bitstream_size = lock_bitstream.bitstreamSizeInBytes;
+ if (bitstream)
+ *bitstream = g_memdup2 (lock_bitstream.bitstreamBufferPtr, *bitstream_size);
- hwaccel_nvidia->nvenc_api.nvEncUnlockBitstream (encode_session->encoder,
- lock_bitstream.outputBitstream);
+ nvenc_api->nvEncUnlockBitstream (encode_session->encoder,
+ lock_bitstream.outputBitstream);
- hwaccel_nvidia->nvenc_api.nvEncUnmapInputResource (encode_session->encoder,
- map_input_res.mappedResource);
- hwaccel_nvidia->nvenc_api.nvEncUnregisterResource (encode_session->encoder,
- register_res.registeredResource);
+ success = TRUE;
- return TRUE;
+out:
+ nvenc_api->nvEncUnmapInputResource (encode_session->encoder,
+ encode_session->mapped_resource);
+ nvenc_api->nvEncUnregisterResource (encode_session->encoder,
+ encode_session->registered_resource);
+ encode_session->mapped_resource = NULL;
+ encode_session->registered_resource = NULL;
+
+ return success;
}
static gboolean
diff --git a/src/grd-hwaccel-nvidia.h b/src/grd-hwaccel-nvidia.h
index 915a2e33..f8a54c59 100644
--- a/src/grd-hwaccel-nvidia.h
+++ b/src/grd-hwaccel-nvidia.h
@@ -93,8 +93,11 @@ gboolean grd_hwaccel_nvidia_avc420_encode_bgrx_frame (GrdHwAccelNvidia *hwaccel
uint16_t src_height,
uint16_t aligned_width,
uint16_t aligned_height,
- uint8_t **bitstream,
- uint32_t *bitstream_size,
CUstream cuda_stream);
+gboolean grd_hwaccel_nvidia_avc420_retrieve_bitstream (GrdHwAccelNvidia *hwaccel_nvidia,
+ uint32_t encode_session_id,
+ uint8_t **bitstream,
+ uint32_t *bitstream_size);
+
#endif /* GRD_HWACCEL_NVIDIA_H */
diff --git a/src/grd-rdp-graphics-pipeline.c b/src/grd-rdp-graphics-pipeline.c
index 01963e5e..088b92e4 100644
--- a/src/grd-rdp-graphics-pipeline.c
+++ b/src/grd-rdp-graphics-pipeline.c
@@ -422,13 +422,20 @@ refresh_gfx_surface_avc420 (GrdRdpGraphicsPipeline *graphics_pipeline,
&rdp_surface->avc.main_view,
surface_width, surface_height,
aligned_width, aligned_height,
- &avc420.data, &avc420.length,
rdp_surface->cuda_stream))
{
g_warning ("[RDP.RDPGFX] Failed to encode YUV420 frame");
return FALSE;
}
+ if (!grd_hwaccel_nvidia_avc420_retrieve_bitstream (graphics_pipeline->hwaccel_nvidia,
+ hwaccel_context->encode_session_id,
+ &avc420.data, &avc420.length))
+ {
+ g_warning ("[RDP.RDPGFX] Failed to retrieve AVC420 bitstream");
+ return FALSE;
+ }
+
GetSystemTime (&system_time);
cmd_start.timestamp = system_time.wHour << 22 |
system_time.wMinute << 16 |
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]