Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | g3dvl: remove unused vs output from create_ref_vert_shader | Christian König | 2011-07-25 | 1 | -4/+1 |
| | | | | | The position of the quad vertex is calculated in calc_position, so we don't need the output here any more. | ||||
* | Revert "g3dvl: Preserve previously rendered components for MC output." | Younes Manton | 2011-07-21 | 1 | -4/+4 |
| | | | | | | This reverts commit b56daf71d2f63d044d4c53ab49c6f87e02991a28. The bug is actually in softpipe's blend and writemask interaction. | ||||
* | g3dvl: remove unused vertex shader inputs | Marek Olšák | 2011-07-20 | 2 | -4/+4 |
| | | | | See also comments in the code. | ||||
* | g3dvl: Preserve previously rendered components for MC output. | Younes Manton | 2011-07-20 | 1 | -4/+4 |
| | | | | Fixes xvmc-softpipe MC entrypoint, amongst others. | ||||
* | g3dvl: Init/clean pipe fully when a shader-based decoder isn't used. | Younes Manton | 2011-07-20 | 2 | -0/+26 |
| | | | | Fixes VDPAU CSC-only mode. | ||||
* | g3dvl: no need for flushing inside the compositor any more | Christian König | 2011-07-15 | 2 | -6/+2 |
| | | | | Move that also inside the state tracker where needed. | ||||
* | g3dvl: correctly distinct dst area and clip area in the compositor | Christian König | 2011-07-15 | 2 | -10/+19 |
| | | | | Otherwise xine won't scale correctly. | ||||
* | g3dvl: change picture parameter of decode_bitstream to general version | Christian König | 2011-07-15 | 1 | -3/+5 |
| | | | | Using pipe_mpeg12_picture_desc was unintentional here. | ||||
* | g3dvl: Remove non-constant expression array initializers. | Vinson Lee | 2011-07-13 | 2 | -7/+31 |
| | | | | The array initializer must be a constant expression in MSVC. | ||||
* | g3dvl: Remove designated initializers. | Vinson Lee | 2011-07-13 | 1 | -4/+4 |
| | | | | MSVC does not support designated initializers. | ||||
* | g3dvl: s/inline/INLINE/ | Vinson Lee | 2011-07-13 | 3 | -36/+40 |
| | | | | The inline keyword is not available in MSVC C. | ||||
* | [g3dvl] don't upload all quant buffer layers at once | Christian König | 2011-07-13 | 3 | -21/+14 |
| | | | | | | | There seems to be a bug in r600g when uploading more than one layer of a 3D resource at once with a hardware blit. So just do them one at a time to workaround this. | ||||
* | [g3dvl] implement workaround for missing blender clamp control | Christian König | 2011-07-13 | 2 | -7/+35 |
| | | | | It's about 20% slower, but should at least work with every hardware. | ||||
* | [g3dvl] keep a pointer in idct buffer to idct object | Christian König | 2011-07-13 | 3 | -20/+26 |
| | | | | So we always know to which idct object a buffer belongs | ||||
* | [g3dvl] add some more PIPE_VIDEO_CAPs | Christian König | 2011-07-12 | 4 | -0/+34 |
| | |||||
* | [g3dvl] fix a whole bunch of memory leaks | Christian König | 2011-07-11 | 3 | -4/+11 |
| | |||||
* | [g3dvl] and finally remove pipe_video_context | Christian König | 2011-07-08 | 6 | -121/+92 |
| | |||||
* | [g3dvl] move video buffer creation out of video context | Christian König | 2011-07-08 | 4 | -79/+91 |
| | |||||
* | [g3dvl] move compositor creation and handling directly into the state trackers | Christian König | 2011-07-08 | 3 | -83/+130 |
| | |||||
* | [g3dvl] remove sampler view handling from video context | Christian König | 2011-07-08 | 1 | -86/+0 |
| | |||||
* | [g3dvl] remove create_surface from video context | Christian König | 2011-07-08 | 1 | -13/+0 |
| | |||||
* | [g3dvl] make pipe_context mandatory for creation pipe_video_context | Christian König | 2011-07-08 | 1 | -2/+0 |
| | |||||
* | [g3dvl] rename is_format_supported to is_video_format_supported and move it ↵ | Christian König | 2011-07-08 | 3 | -31/+35 |
| | | | | into screen object | ||||
* | [g3dvl] rename get_param to get_video_param and move into screen object | Christian König | 2011-07-07 | 2 | -23/+17 |
| | |||||
* | [g3dvl] rework video buffer format handling | Christian König | 2011-06-07 | 3 | -30/+53 |
| | |||||
* | [g3dvl] move dummy quantification into xvmc state tracker | Christian König | 2011-06-07 | 1 | -15/+3 |
| | |||||
* | [g3dvl] split quant matrix out of picture info | Christian König | 2011-06-05 | 1 | -7/+14 |
| | |||||
* | [g3dvl] rename map/unmap to begin/end frame | Christian König | 2011-06-05 | 1 | -4/+4 |
| | | | | | mapping and unmapping of buffers is just an implementation detail. begining and ending an frame is much more descriptive | ||||
* | [g3dvl] remove unused and dublicate fields from picture structure | Christian König | 2011-06-05 | 1 | -3/+3 |
| | |||||
* | [g3dvl] use a vertex element instead of the instance id | Christian König | 2011-06-05 | 1 | -2/+1 |
| | | | | Enable this change permanently | ||||
* | [g3dvl] use a vertex element instead of the instance id | Christian König | 2011-06-04 | 5 | -21/+85 |
| | |||||
* | [g3dvl] respect maximum instruction for idct render targets | Christian König | 2011-06-01 | 1 | -3/+8 |
| | |||||
* | [g3dvl] dynamical adjust blocks per line | Christian König | 2011-06-01 | 1 | -5/+4 |
| | |||||
* | [g3dvl] move quantification into shaders | Christian König | 2011-05-29 | 4 | -112/+168 |
| | |||||
* | [g3dvl] implement mismatch control inside idct shaders | Christian König | 2011-05-29 | 4 | -46/+198 |
| | |||||
* | [g3dvl] fix resource type for idct | Christian König | 2011-05-29 | 1 | -12/+11 |
| | | | | Only the intermediate resource are 3D textures | ||||
* | [g3dvl] rework resource format handling | Christian König | 2011-05-22 | 1 | -131/+88 |
| | |||||
* | [g3dvl] fix missing swizzle on instanceid | Christian König | 2011-05-18 | 1 | -1/+1 |
| | |||||
* | [g3dvl] add missing ureg_END | Christian König | 2011-05-16 | 1 | -0/+2 |
| | |||||
* | [g3dvl] store the idct matrix as hex values | Christian König | 2011-05-16 | 1 | -10/+17 |
| | | | | Give the dct matrix a higher precision. | ||||
* | [g3dvl] enable gl_rasterization_rules for idct | Christian König | 2011-05-15 | 1 | -11/+6 |
| | |||||
* | [g3dvl] enable gl_rasterization_rules for zscan | Christian König | 2011-05-15 | 1 | -6/+5 |
| | |||||
* | [g3dvl] store mvpos seperately from x,y cord | Christian König | 2011-05-15 | 1 | -23/+27 |
| | |||||
* | [g3dvl] pre apply zscan to quant matrix | Christian König | 2011-05-15 | 1 | -68/+51 |
| | |||||
* | [g3dvl] no need for seperate intermediate buffers any more | Christian König | 2011-05-15 | 2 | -102/+110 |
| | | | | This should reduce the video memory footprint drastically. | ||||
* | [g3dvl] move alignment of buffers to mb size into context | Christian König | 2011-05-15 | 2 | -7/+4 |
| | |||||
* | [g3dvl] fix fi_frame mb increment | Christian König | 2011-05-14 | 1 | -2/+19 |
| | |||||
* | [g3dvl] move dct_type detection out of get_macroblock_modes | Christian König | 2011-05-08 | 1 | -29/+23 |
| | |||||
* | [g3dvl] fix field selection of mb without mc | Christian König | 2011-05-08 | 1 | -18/+23 |
| | |||||
* | [g3dvl] memcpy the DCT buffer instead of writing it directly | Christian König | 2011-05-07 | 1 | -2/+6 |
| | | | | It looks like texture buffers are not cached so this seems to be alot faster |