Zhao Yakui [Fri, 1 Mar 2013 02:38:13 +0000 (10:38 +0800)]
Set the max thread num for PS thread on BDW
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Fri, 1 Mar 2013 02:38:13 +0000 (10:38 +0800)]
Update the MI_BATCH_BUFFER_START for BDW
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Fri, 1 Mar 2013 02:38:13 +0000 (10:38 +0800)]
Fix the VPP error during porting patch from master to staging
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Xiang, Haihao [Thu, 28 Feb 2013 04:49:55 +0000 (12:49 +0800)]
Setup VP8 decoding pipeline
Update the pipeline state later.
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 27 Feb 2013 05:36:19 +0000 (13:36 +0800)]
New macros for Gen8
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 27 Feb 2013 07:56:24 +0000 (15:56 +0800)]
Add support for VAProbabilityBufferType
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 28 Feb 2013 04:43:51 +0000 (12:43 +0800)]
Temporarily remove assert() to make vainfo happy
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Thu, 28 Feb 2013 04:40:52 +0000 (12:40 +0800)]
Advertise VP8 decoding on Gen8
The pipeline isn't implemented yet.
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 27 Feb 2013 04:53:07 +0000 (12:53 +0800)]
Surface fourcc format on Gen8
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 27 Feb 2013 05:25:37 +0000 (13:25 +0800)]
Configuration for VP8 decoding/encoding
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 23 Jan 2013 06:27:20 +0000 (14:27 +0800)]
Enlarge deblocking filter row store on BDW
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 23 Jan 2013 06:24:41 +0000 (14:24 +0800)]
No workaround for JPEG decoding on BDW
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Tue, 22 Jan 2013 02:53:26 +0000 (10:53 +0800)]
Use the BDW surface/sampler state and memory address allocation for rendering
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
[Haihao: directly use object intead of id]
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Tue, 22 Jan 2013 02:53:26 +0000 (10:53 +0800)]
Update the BDW surface/sampler state and media command for encoding/decoding
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Tue, 22 Jan 2013 02:53:26 +0000 (10:53 +0800)]
Avoid the duplicated macro-definition of surface size
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Tue, 22 Jan 2013 02:53:26 +0000 (10:53 +0800)]
Use the updated structure/command for VPP on BDW
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
[Haihao: directly use object instead of id]
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Mon, 7 Jan 2013 05:18:47 +0000 (13:18 +0800)]
Add the BDW idrt/surface/sampler state definition
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Mon, 7 Jan 2013 05:18:47 +0000 (13:18 +0800)]
Add the separated media encoding/decoding files for BDW
As a lot of changes about the media are added between Haswell and BDW, the
separated media encoding/decoding files are added for BDW. This
is to avoid complex backward logic for Haswell.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
[Haihao: directly use object instead of id]
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Mon, 7 Jan 2013 05:18:47 +0000 (13:18 +0800)]
Add the initial support for GEN8
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
[Haihao: fix conflict when rebasing]
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Wed, 26 Dec 2012 07:07:19 +0000 (15:07 +0800)]
Add the PCI ids for BDW
This is from the kernel driver.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Xiang, Haihao [Mon, 30 Dec 2013 05:15:40 +0000 (13:15 +0800)]
Render: Adjust the default value for contrast/saturation
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=73016
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Tested-by: Mark Lee <mark@markelee.com>
Xiang, Haihao [Mon, 30 Dec 2013 05:13:02 +0000 (13:13 +0800)]
Bump version for development
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 16 Dec 2013 08:05:43 +0000 (16:05 +0800)]
Intel driver 1.2.2
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Wed, 11 Dec 2013 02:03:54 +0000 (10:03 +0800)]
1.2.2.pre2
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 10 Dec 2013 02:19:09 +0000 (10:19 +0800)]
VPP: Use the right top/bottom field flag used for DI
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=72518
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 9 Sep 2013 01:10:51 +0000 (09:10 +0800)]
H.264: Support Constrained Baseline profile instead of Baseline profile
GENx doesn't support FMO/ASO, so remove the support
of Baseline profile for conformance testing. In addition, add the support
for Constrained Baseline profile.
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Mon, 2 Dec 2013 07:56:20 +0000 (15:56 +0800)]
Upload the constant buffer on ILK/CTG so that subpicture can work
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
2a61b0d17072bdc5b58608e3dfa2c4f9f80dcec4)
Zhao Yakui [Fri, 22 Nov 2013 05:39:34 +0000 (13:39 +0800)]
Support the smpte240m color standard for conversion from YUV to RGB
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
34627c96f331f7a344270c3d51b634f5f166073e)
Zhao Yakui [Fri, 22 Nov 2013 05:39:34 +0000 (13:39 +0800)]
Support the BT709 color standard for conversion from YUV to RGB
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
4c43ff9234a0a6f18744078d2e743cfa0cf8f34c)
Zhao Yakui [Fri, 22 Nov 2013 05:39:34 +0000 (13:39 +0800)]
Constant buffer passes YUV2RGB CSC matrix instead of hardcoded matrix
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
0aa6ccc405726e5521452ee437be6b3cc6fabdee)
Zhao Yakui [Fri, 22 Nov 2013 05:39:34 +0000 (13:39 +0800)]
Increase the size of constant buffer for PS thread to pass more info
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
a6372850042e2a9201f4893193f77c9a55a6598e)
Zhao Yakui [Fri, 22 Nov 2013 05:39:34 +0000 (13:39 +0800)]
Fix the error in render shader on Ivy/Haswell
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
7af528d17924cf2ec855c0d0b1550b6c3d095682)
Xiang, Haihao [Thu, 14 Nov 2013 07:48:59 +0000 (15:48 +0800)]
Enlarge the size of array misc_param
And check the type before storing misc parameters
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 26 Nov 2013 01:21:11 +0000 (09:21 +0800)]
dec/mpeg2: ignore slices which aren't in raster scan order on SNB
Sometimes codec layer incorrectly fills slice parameters due to
the corrupted video
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=71276
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 26 Nov 2013 02:43:45 +0000 (10:43 +0800)]
SNB doesn't support MPEG-2 encoding
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=72016
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Fri, 6 Sep 2013 08:43:17 +0000 (16:43 +0800)]
Workaround for SNB
Backporting from d0184b5 in xf86-video-intel
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
e622ecedf169bccddc8910b45d92dbec7675441e)
Xiang, Haihao [Wed, 13 Nov 2013 06:18:20 +0000 (14:18 +0800)]
VPP: remove some assert()
Instead check the input parameters and return corresponding error status if failed
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
40fa7d9ede00e804f15df4b7b805c7345a925e17)
Xiang, Haihao [Wed, 13 Nov 2013 05:19:16 +0000 (13:19 +0800)]
Correct the usage of width/height in struct object_surface
Add comments for width/height, orig_width/orig_height as well
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
f886f24eaaacba9544fa5f6405b7382c686f3a1f)
Xiang, Haihao [Wed, 13 Nov 2013 02:27:59 +0000 (10:27 +0800)]
VPP: use the target widht/height to calculate the horizontal/vertical step on IVB
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
ba020ed2c7730980649d9131e41e9677f603c52a)
Xiang, Haihao [Thu, 26 Sep 2013 03:02:09 +0000 (11:02 +0800)]
render: add support for brightness/contrast/hue/saturation
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
04ecb6e79f4382d96eb5d4b51733049d420f592a)
Xiang, Haihao [Wed, 18 Sep 2013 07:12:55 +0000 (15:12 +0800)]
VPP: change the default values for Saturation and Contrast
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
ce0c8f0019e8545d0db529b0f28338be4b8adc15)
Zhao Yakui [Mon, 11 Nov 2013 06:51:50 +0000 (14:51 +0800)]
Remove the unused variable to avoid the warning
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
0d37a309bd99f6bded4df922d0ece22bf3bb1757)
Zhao Yakui [Fri, 8 Nov 2013 07:36:36 +0000 (15:36 +0800)]
Use GPU to construct MFX command buffer for H264 encoding on Haswell
This is to optimze the performance of h264 encoding. The GPU can
accelerate the construction of MFX command buffer for H264 encoding.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
87bc38d4300212dea51b5635f184aa1ae37fa71c)
Zhao Yakui [Fri, 8 Nov 2013 07:36:32 +0000 (15:36 +0800)]
Encoding reuses aux_batchbuffer instead of allocating another new buffer
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
052ce2930cd4661b7ce62902e6553eec0e2db9f1)
Zhao Yakui [Fri, 8 Nov 2013 07:36:28 +0000 (15:36 +0800)]
Calculate the required space of batch buffer to avoid buffer overflow in encoding
The required size is based on the number of macroblocks and slice parameter.
Then it can avoid that too large buffer is allocated or possible overflow.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
8acdfd023e50af37a5642e2517683c34accd78b0)
Zhao Yakui [Tue, 15 Oct 2013 03:04:00 +0000 (11:04 +0800)]
Return the error instead of assert in vaEndPicture
This is to fix the crash issue caused by the incorrect parameter.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
1cee858036a87837deddc87586701ed869f96261)
Zhao Yakui [Tue, 15 Oct 2013 03:01:04 +0000 (11:01 +0800)]
Add more strict check to fix crash issue caused by invalid parameter
This is to fix the crash issue in https://bugs.freedesktop.org/show_bug.cgi?id=70397
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
80d665eb670fd700d03f9a2486e452947177a058)
Zhao Yakui [Mon, 14 Oct 2013 01:56:30 +0000 (09:56 +0800)]
Fix one error of VME shader for MPEG2 encoding
Otherwise the MPEG2 encoding will use the incorrect prediction result for the
macroblocks in the first row if the MVP is used.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
44889dc0f3054cce226d5c09d431022fdffe3aac)
Zhao Yakui [Mon, 14 Oct 2013 01:56:26 +0000 (09:56 +0800)]
Fix one error of VME shader for H264 encoding
Otherwise the h264 encoding will use the incorrect prediction result for the
macroblocks in the first row if the MVP is used.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhao Yakui [Mon, 14 Oct 2013 01:56:21 +0000 (09:56 +0800)]
Fix an incorrect makefile rule for VME shader on Ivybridge
Otherwise when the corresponding source file is modified, the binary shader
is not updated.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Zhong Li [Wed, 25 Sep 2013 07:56:58 +0000 (15:56 +0800)]
VPP: add vebox motion compensation support on HSW
Signed-off-by: Zhong Li <zhong.li@intel.com>
(cherry picked from commit
2c7c7c4d20014342538a80bfd1525f9bef5ea971)
Conflicts:
src/i965_drv_video.c
Xiang, Haihao [Mon, 30 Sep 2013 07:06:57 +0000 (15:06 +0800)]
Fix the Reference Index Mapping Table L0/L1
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
3ffbe0297e98a88db18ae90ba5c1f8c429183baf)
Xiang, Haihao [Mon, 30 Sep 2013 05:12:17 +0000 (13:12 +0800)]
Follow the input Picture/Slice parameters to generate slice header/data
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
f5a694e64d0163178c28dc25d9a3e7b9b1b5d162)
Xiang, Haihao [Sun, 29 Sep 2013 08:25:02 +0000 (16:25 +0800)]
Pass the reference frame index in List0/1 into the PAK command
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
68380a7f141bedcc0f6fbbbcee2f5e42b6ade0e0)
Xiang, Haihao [Mon, 30 Sep 2013 07:17:11 +0000 (15:17 +0800)]
Select a reference frame from the reference list0/1
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
3a51e5271773a637ef63ca397285ebdf326daba2)
Xiang, Haihao [Sun, 29 Sep 2013 06:29:16 +0000 (14:29 +0800)]
Track the used reference surface
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
97e1b531d85bd7b7d3bc1d3e5a7c3355af87a204)
Xiang, Haihao [Sun, 29 Sep 2013 06:53:12 +0000 (14:53 +0800)]
Clean up for setting up reference surface state
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
939b6bef6d8f8ecfee589cf70fde51f7a34175a1)
Xiang, Haihao [Sun, 29 Sep 2013 05:16:29 +0000 (13:16 +0800)]
Fix the reference for list1
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
a45edbef143808ee925ef7708c516e6df21fa36b)
Xiang, Haihao [Sun, 29 Sep 2013 05:11:10 +0000 (13:11 +0800)]
Check the reference surface id against VA_INVALID_SURFACE
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
0c2def319f52bdb222e5480d81feea486cbf3e11)
Xiang, Haihao [Sun, 29 Sep 2013 06:44:35 +0000 (14:44 +0800)]
Indent the code of encoding
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
6ad68f55e9ba49af541a5e4d86a305bbd0f22d63)
Xiang, Haihao [Tue, 24 Sep 2013 05:25:37 +0000 (13:25 +0800)]
fix the broken package generated by make dist
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 24 Sep 2013 04:52:37 +0000 (12:52 +0800)]
fix make dist
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 24 Sep 2013 01:13:22 +0000 (09:13 +0800)]
1.2.2.pre1 for development
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 23 Sep 2013 07:36:41 +0000 (15:36 +0800)]
Intel driver 1.2.1
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Sun, 22 Sep 2013 05:03:42 +0000 (13:03 +0800)]
VPP: fix the check for DI filter flags on SNB/IVB
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
ea3c725a7f9fa66e7ef05e1c08af30c28f8c1aa0)
Zhao Halley [Thu, 12 Sep 2013 00:09:23 +0000 (08:09 +0800)]
support 422H/422V/411P/444P format for wayland video output
(cherry picked from commit
7a02190eda3abc245ba918f6078a121373f50703)
Xiang, Haihao [Mon, 23 Sep 2013 01:51:39 +0000 (09:51 +0800)]
VPP: also remove an assert on forward temporal reference on HSW
Do the same thing of commit 8a0a61c
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=69669
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
f6685c309d94fb7679c9772703c8790cb71cdd73)
Xiang, Haihao [Fri, 13 Sep 2013 07:47:50 +0000 (15:47 +0800)]
VPP: Remove an assert on forward temporal reference on IVB
Add a warning and return VA_STATUS_ERROR_INVALID_PARAMETER instead
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
8a0a61ce47e086ac156602c3daaf70f20e4ca222)
Xiang, Haihao [Fri, 13 Sep 2013 05:21:09 +0000 (13:21 +0800)]
VPP: MADI on IVB
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
219452451ede9d52d8940be55cd38feda50baea7)
Conflicts:
src/i965_drv_video.c
Xiang, Haihao [Fri, 13 Sep 2013 02:12:53 +0000 (10:12 +0800)]
VPP: track the frame sequence for DI on IVB
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
c008b4222d593594330fbb087204ff53722f9765)
Xiang, Haihao [Wed, 11 Sep 2013 03:05:03 +0000 (11:05 +0800)]
VPP: move the BO for STMM into sub-contexts
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
6ce827df0a9d6cb841308813ec54a019e3c63d37)
Xiang, Haihao [Wed, 11 Sep 2013 02:33:25 +0000 (10:33 +0800)]
VPP: Create separate sub-context for each processing
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
4faf6bf47f8e4e2fe587e3bb6a004340edd59c4c)
Conflicts:
src/i965_post_processing.c
src/i965_post_processing.h
Xiang, Haihao [Wed, 11 Sep 2013 01:04:52 +0000 (09:04 +0800)]
VPP: the similar fix to Bob DI on SNB
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
c55cc476b1f75ddf0504a8b85ac80c0168585c5c)
Xiang, Haihao [Wed, 11 Sep 2013 00:43:17 +0000 (08:43 +0800)]
VPP: Update DEINTERLACE_SAMPLER_STATE on IVB
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
b05729d0feea5ceb4217997f26442d5a8d94fa48)
Xiang, Haihao [Tue, 10 Sep 2013 08:45:30 +0000 (16:45 +0800)]
VPP: VAProcFilterParameterBufferDeinterlacing::flags overrides VAProcPipelineParameterBuffer::filter_flags
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
(cherry picked from commit
5334ca07c5ad1a2edeb279784bde3d03cbb10c49)
Xiang, Haihao [Mon, 9 Sep 2013 07:16:56 +0000 (15:16 +0800)]
Update NEWS
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Gwenole Beauchesne [Fri, 26 Jul 2013 11:08:27 +0000 (13:08 +0200)]
allow vaDeriveImage() to work for UYVY formats
Hi,
I would push the following as obvious around next week.
Add support for UYVY format to vaDeriveImage(). Also remove dead code
along the way, i.e. packed YUV 4:2:2 formats have a single plane.
Regards,
Gwenole.
Gwenole Beauchesne [Thu, 25 Jul 2013 08:52:31 +0000 (10:52 +0200)]
vpp: fix output filter count from QueryVideoProcFilters().
When vaQueryVideoProcFilters() returns VA_STATUS_ERROR_MAX_NUM_EXCEEDED,
i.e. when the caller allocated too few entries for the filters argument,
then the num_filters argument shall be adjusted to correct number of
entries that need to be re-allocated.
https://bugs.freedesktop.org/show_bug.cgi?id=67292
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Gwenole Beauchesne [Fri, 26 Jul 2013 08:23:39 +0000 (10:23 +0200)]
vpp: add basic processing support for packed YUV to packed YUV.
Add support for video processing from packed YUV 4:2:2 formats (YUY2, UYVY)
to packed YUV 4:2:2 formats. In particular, add support for HW accelerated
vaPutImage() for packed YUV formats.
https://bugs.freedesktop.org/show_bug.cgi?id=67338
Signed-off-by: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
[Haihao: the shader for IVB+ has been added in 0eb2288]
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 9 Sep 2013 06:44:53 +0000 (14:44 +0800)]
VPP: Packed 4:2:2 to packed 4:2:2 on IVB+
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 9 Sep 2013 06:03:17 +0000 (14:03 +0800)]
VPP: cleanup code
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 9 Sep 2013 05:18:43 +0000 (13:18 +0800)]
Return the status of image processing
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Fri, 23 Aug 2013 05:39:07 +0000 (13:39 +0800)]
More conversions between two images with different pixel formsts
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Tue, 20 Aug 2013 14:02:09 +0000 (22:02 +0800)]
Convert RGBx/RGBA into non-NV12 format
It is the combined conversions of RGBx/RGBA->NV12 and NV12->non-NV12.
It would be better to implement RGBx/RGBA->YUV444 and YUV444->non-444 later
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Xiang, Haihao [Mon, 2 Sep 2013 08:23:36 +0000 (16:23 +0800)]
Support B frame for reference frame
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Remove the unnecessary shader binary for MPEG2 encoding on Haswell/Ivb
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
42bb613e72d235bcbe141c906dec9431e4c29661)
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Optimize quantization rounding precision for MPEG2 encoding on Ivy
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
05ea96da3e0b09648bfdeb35967f6ab9bb3b23e4)
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Optimize the VME shader for MPEG2 encoding on Ivb
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
15db142076321e9523db0c8be8e6bae5e1b64c8a)
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Optimize MPEG2 encoding on Ivb
This is backported from Haswell. The scoreboard/walker/cost-center
is applied.
Signed-off-by: Zhao yakui <yakui.zhao@intel.com>
(cherry picked from commit
cc7452d14f7faa1d5b5fa8c16db3e3cb3fd4f53d)
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Rewrite the VME shader for MPEG2 encoding on Ivy
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
6200c9a7779c1309f5a85b7c62aec1b9796793c6)
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Remove the dead code in file of gen7_vme.c
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
d15582b54486847811f3feab46ddd51181561776)
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Pass the constant buffer info for MPEG2 encoding correctly on Ivb
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
7690091889eac91dcab53e0318f9810c25071e18)
Zhao Yakui [Mon, 12 Aug 2013 07:13:24 +0000 (15:13 +0800)]
Optimize quantization rounding precision for MPEG2 encoding on haswell
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
24d8bf31c8aeb326bc8b33c1ac9700ec1d169666)
Zhao Yakui [Mon, 12 Aug 2013 07:13:23 +0000 (15:13 +0800)]
Optimize the VME shader for MPEG2 on Haswell
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
19e93152f0e10f94ecaf3ddecf95c1dc7b97dfed)
Zhao Yakui [Mon, 12 Aug 2013 07:13:23 +0000 (15:13 +0800)]
Configure the cost-center of MPEG2 VME shader on haswell
This is derived from the neighbour macroblock based on MPEG2 spec.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
a51860aba2d2713a9c8c817c430ffa93abd5145a)
Zhao Yakui [Mon, 12 Aug 2013 07:13:23 +0000 (15:13 +0800)]
Use the scoreboard/walker to assure MB dependency for MPEG2 encoding
If MVP is added for MPEG2 encoding, it must be assured that the left macroblock
should be already finished before processing the current macroblock.
And this needs the scoreboard/walker mechanism to assure MB dependency.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
5d2163d02d395fc0a794d834979a06c287bf9ba5)
Zhao Yakui [Mon, 12 Aug 2013 07:13:23 +0000 (15:13 +0800)]
configure the dynamic VME MV/mode cost for MPEG2 encoding on Haswell
Currently it uses the constant VME MV/mode cost when executing the mode/motion
vector prediction for MPEG2 encoding on Haswell, which causes that the unoptimized
mode/MV is used for MPEG2 encoding.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
6fec7e353704dc23c9675966467caffa95f792fe)
Zhao Yakui [Mon, 12 Aug 2013 07:13:23 +0000 (15:13 +0800)]
Enable the Intra-prediction for MPEG2 P-B frame
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
c074d4d61ad931d044fbe0836a45c49768090b4b)
Zhao Yakui [Mon, 12 Aug 2013 07:13:23 +0000 (15:13 +0800)]
Restrict the MV search range based on MPEG2 encoding LEVEL
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
c05073b1f8764271ccf4fe1aa037f881dedd3818)
Zhao Yakui [Mon, 12 Aug 2013 07:13:23 +0000 (15:13 +0800)]
Rewrite inter-frame shader for MPEG2 encoding on HSW to follow MPEG2 spec
Now the MPEG2/H264 uses the same mode/motion vector prediction shader. But
the MV search region of mpeg2 is different with that on H264, which causes
that the wrong mode/motion vector prediction is used for MPEG2.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
(cherry picked from commit
6842f08aa375b5942cee4b9d06421609c212895a)