[Libva] [LIBVA_INTEL_DRIVER][PATCH v2 0/8] Add the 10bit VPP conversion for P010
Zhao Yakui
yakui.zhao at intel.com
Fri Nov 11 01:56:02 UTC 2016
From: "Zhao, Yakui" <yakui.zhao at intel.com>
This is the patch set that tries to optimize the 10bit VPP conversion on
P010 surface. Currently it can support the 10bit VPP conversion on P010
surface.But when the size is changed, it needs the three-step conversion, which
leverage the VEBOX. (P010->NV12->scaled NV12->P010).
V1->V2: Follow Haihao's comments to update the patch set.
1. Use the is_16bpp field
2. Add the shader(conv_p010.g9b) into dist package
3. Remove the unnecessary function initialization in gen75_vpp_fmt_cvt
as it is already initalized.
Zhao Yakui (8):
Fix the incorrect sampler_state offset for INTERFACE_DESCRIPTOR_DATA
on BDW+
Add the 10-bit flag for MEDIA_BLOCK_RW operation on P010 surface
Add one API to initialize MEDIA_OBJECT_WALKER parameter for video
post-processing
Initialize internal context based on Render ring earlier
Initialize one 10bit-scaling gpe_context for Gen9
VPP select the appropriate fourcc for deferred surface based on
expected format
Execute the 10-bit scaling for P010 surface on Gen9+
Align coordinate/width of DST rect for 10-bit VPP conversion
src/gen75_picture_process.c | 73 ++-
src/gen75_vpp_vebox.c | 2 +
src/gen75_vpp_vebox.h | 1 -
src/gen8_post_processing.c | 5 +
src/gen8_post_processing.h | 9 +
src/gen9_post_processing.c | 487 +++++++++++++++
src/i965_drv_video.c | 2 +
src/i965_encoder.c | 2 +
src/i965_gpe_utils.c | 96 ++-
src/i965_gpe_utils.h | 19 +-
src/i965_post_processing.h | 10 +
src/i965_render.h | 1 -
src/shaders/post_processing/gen9/Makefile.am | 4 +
src/shaders/post_processing/gen9/conv_p010.g9b | 783 +++++++++++++++++++++++++
14 files changed, 1474 insertions(+), 20 deletions(-)
create mode 100644 src/shaders/post_processing/gen9/conv_p010.g9b
--
1.9.0
More information about the Libva
mailing list