summaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/r600
Commit message (Collapse)AuthorAgeFilesLines
* vl: replace decode_buffers with auxiliary data fieldChristian König2012-01-091-2/+0
| | | | | | | Based on patches from Maarten Lankhorst <[email protected]> Signed-off-by: Christian König <[email protected]> Acked-by: Maarten Lankhorst <[email protected]>
* r600g: remove obsolete MULTIWRITE commentAlex Deucher2012-01-061-1/+0
| | | | | | fs writes all is implemented in the shader on eg+. Signed-off-by: Alex Deucher <[email protected]>
* u_vbuf: implement another upload codepath which unrolls indicesMarek Olšák2012-01-051-1/+1
| | | | | | | Improves performance from cca 1 fps to 23 fps in Cogs. This new codepath is not always used, instead, there is a heuristic which determines whether to use it. Using translate for uploads is generally slower than what we have had already, it's a win only in a few cases.
* gallium: add flag PIPE_TRANSFER_MAP_PERMANENTLYMarek Olšák2012-01-051-0/+4
| | | | | | | | | | | Please see the diff for further info. This paves the way for moving user buffer uploads out of drivers and should allow to clean up the mess in u_upload_mgr in the meantime. For now only allowed for buffers on r300 and r600. Acked-by: Christian König <[email protected]>
* gallium: remove deprecated PIPE_TRANSFER_DISCARDMarek Olšák2012-01-051-1/+1
| | | | PIPE_TRANSFER_DISCARD_RANGE is defined the same.
* u_upload_mgr: remove the 'flushed' parameterMarek Olšák2012-01-052-7/+4
| | | | | | Not used by anybody. Reviewed-by: Brian Paul <[email protected]>
* r600g: add missing colorswaps for r8 uint/sint.Dave Airlie2012-01-022-0/+4
| | | | | | fixes some warnings in GL3.0 tests. Signed-off-by: Dave Airlie <[email protected]>
* r600g: Manage fences per screen rather than per context.Michel Dänzer2011-12-302-56/+65
| | | | | | | | | | | | | | | | A fence is a screen object and can outlive the context it was created from. The previous code would access freed memory in that case, resulting in various problems. Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=44151 https://bugs.freedesktop.org/show_bug.cgi?id=44007 Probably fixes: https://bugs.freedesktop.org/show_bug.cgi?id=43993 NOTE: This is a candidate for the 7.11 branch. Signed-off-by: Michel Dänzer <[email protected]> Reviewed-by: Mathias Fröhlich <[email protected]>
* r600g: fix a warning that a variable may be uninitializedMarek Olšák2011-12-251-0/+2
|
* gallium: remove PIPE_CAP_MAX_STREAM_OUTPUT_SEPARATE_ATTRIBSMarek Olšák2011-12-251-1/+0
| | | | | | It's the same as PIPE_CAP_MAX_STREAM_OUTPUT_BUFFERS. Reviewed-by: Brian Paul <[email protected]>
* winsys/radeon: move managing GEM domains back to driversMarek Olšák2011-12-244-2/+35
| | | | | | | | This partially reverts commit 363ff844753c46ac9c13866627e096b091ea81f8. It caused severe performance drops in Nexuiz. Reported by Phoronix. Tested by me on r300g and by IRC people on r600g.
* r600g: fix MAX_STREAM_OUTPUT_SEPARATE_ATTRIBS limitMarek Olšák2011-12-171-2/+1
|
* r600g: handle new capsMarek Olšák2011-12-171-0/+2
|
* r600g: implement transform feedbackMarek Olšák2011-12-1718-13/+873
| | | | | | | | r600: DONE. r700: MOSTLY (done but locks up). Evergreen: MOSTLY (done but doesn't work for an unknown reason). The kernel support will come soon.
* gallium: interface changes necessary to implement transform feedback (v5)Marek Olšák2011-12-151-1/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Namely: - EXT_transform_feedback - ARB_transform_feedback2 - ARB_transform_feedback_instanced The old interface was not useful for OpenGL and had to be reworked. This interface was originally designed for OpenGL, but additional changes have been made in order to make st/d3d1x support easier. The most notable change is the stream-out info must be linked with a vertex or geometry shader and cannot be set independently. This is due to limitations of existing hardware (special shader instructions must be used to write into stream-out buffers), and it's also how OpenGL works (stream outputs must be specified prior to linking shaders). Other than that, each stream output buffer has a "view" into it that internally maintains the number of bytes which have been written into it. (one buffer can be bound in several different transform feedback objects in OpenGL, so we must be able to have several views around) The set_stream_output_targets function contains a parameter saying whether new data should be appended or not. Also, the view can optionally be used to provide the vertex count for draw_vbo. Note that the count is supposed to be stored in device memory and the CPU never gets to know its value. OpenGL way | Gallium way ------------------------------------ BeginTF = set_so_targets(append_bitmask = 0) PauseTF = set_so_targets(num_targets = 0) ResumeTF = set_so_targets(append_bitmask = ~0) EndTF = set_so_targets(num_targets = 0) DrawTF = use pipe_draw_info::count_from_stream_output v2: * removed the reset_stream_output_targets function * added a parameter append_bitmask to set_stream_output_targets, each bit specifies whether new data should be appended to each buffer or not. v3: * added PIPE_CAP_STREAM_OUTPUT_PAUSE_RESUME for ARB_tfb2, note that the draw-auto subset is always required (for d3d10), only the pause/resume functionality is limited if the CAP is not advertised v4: * update gallium/docs v5: * compactified struct pipe_stream_output_info, updated dump/trace
* r600g: check shaders presence in r600_draw_vboVadim Girlin2011-12-061-0/+3
| | | | | | | | | | This patch should prevent the crashes when some shaders are absent, see https://bugs.freedesktop.org/show_bug.cgi?id=43341 Note this is a candidate for the stable branch. Signed-off-by: Vadim Girlin <[email protected]> Signed-off-by: Alex Deucher <[email protected]>
* r600g: fix error path and use util_slab_freeJerome Glisse2011-12-051-1/+1
| | | | Signed-off-by: Jerome Glisse <[email protected]>
* r600g: add framebuffer support for 2/10/10/10 integerDave Airlie2011-11-272-0/+4
|
* r600g: add int support for 2/10/10/10 format.Dave Airlie2011-11-271-0/+3
| | | | | | integer wasn't set properly for the non-uniform types. Signed-off-by: Dave Airlie <[email protected]>
* android: pass -std=c99 by defaultChia-I Wu2011-11-261-2/+0
| | | | | Several modules expect a C99 compiler already. It is also the default for Makefile build.
* r600g: handle all remaining CAPsMarek Olšák2011-11-221-0/+6
|
* r600g: remove default case statements in get_paramMarek Olšák2011-11-221-9/+3
| | | | This will throw a compile warning if there's an unhandled CAP.
* u_vbuf_mgr: rename to u_vbufMarek Olšák2011-11-223-4/+4
|
* gallium: separate out floating-point CAPs into its own enumMarek Olšák2011-11-221-7/+8
| | | | | | | | | | | | | | | The motivation behind this is to add some self-documentation in the code about how each CAP can be used. The idea is: - enum pipe_cap is only valid in get_param - enum pipe_capf is only valid in get_paramf Which CAPs are floating-point have been determined based on how everybody except svga implemented the functions. svga have been modified to match all the other drivers. Besides that, the floating-point CAPs are now prefixed with PIPE_CAPF_.
* gallium: remove PIPE_CAP_GLSL and enable GLSL unconditionallyMarek Olšák2011-11-221-1/+0
| | | | | Only i965g does not enable GLSL, but that driver has been unmaintained and bitrotting for quite a while anyway.
* r600g: handle PIPE_SHADER_CAP_OUTPUT_READVadim Girlin2011-11-211-0/+2
| | | | Signed-off-by: Vadim Girlin <[email protected]>
* r600g: use u_vbuf_mgr to set/get the index bufferMarek Olšák2011-11-202-14/+5
|
* r600g: set MIN/MAX_VTX_INDX to 0 and ~0, respectivelyMarek Olšák2011-11-201-10/+4
| | | | | | | The CS checker doesn't check the regs and the state-tracker-provided values are not to be trusted. This also removes the hack for non-zero index bias.
* r600g: don't change the order of writes in merge_inst_groupVadim Girlin2011-11-171-1/+14
| | | | | | | | | | | | | | Merge may produce incorrect order of operations for r600-eg: x: inst1 R0.x, ... ; //from current group ... t: inst0 R0.x, ... ; //from previous group, same destination Result of inst1 will be lost. So compare destinations and don't allow this. Signed-off-by: Vadim Girlin <[email protected]>
* r600g: fix the representation of control-flow instructionsMarek Olšák2011-11-159-426/+536
| | | | | | | | | | | | | | | | | | | | | | | | | | We need something that looks like a compiler and not like some hacker put some functions together. /rant This is a band-aid for these two problems: - The R600 and EG control-flow instructions appear in switch statements next to each other, causing conflicts when adding new instructions. - The ALU control-flow instructions are bitshifted by 3 (from CF_INST 26:29 to CF_INST 23:29, as is defined by r600 ISA) even for EG, where CF_INST is 22:29. To fix this mess, the 'inst' field is bitshifted to the left either by 22, 23, or 26 (directly in the definitions), such that it can be just or'd when making bytecode without any shifting. All switch statements have been divided into two, one for R600 and the other for EG. Of course, there is a better way to do this, but that is left for future work. Tested on RV730 and REDWOOD with no regressions. v2: minor cleanup as per Alex's comment. Reviewed-by: Alex Deucher <[email protected]>
* r600g: mask write for fake VS export componentsVadim Girlin2011-11-141-4/+4
|
* r600g: fix op3 & write in merge_inst_groupsVadim Girlin2011-11-141-1/+1
|
* r600g: set max max tex/vtx instructions count to 16 for caymanAlex Deucher2011-11-141-3/+1
| | | | | | Cayman is 16 as well. Signed-off-by: Alex Deucher <[email protected]>
* r600g: set max tex/vtx instructions count to 16 for evergreenVadim Girlin2011-11-141-1/+1
| | | | | | | According to evergreen-isa doc 16 is max value for evergreen. More than 16 doesn't work for me. Signed-off-by: Vadim Girlin <[email protected]>
* r600g: properly handle cayman in is_alu_vec_unit_inst()Alex Deucher2011-11-131-7/+15
| | | | Signed-off-by: Alex Deucher <[email protected]>
* r600g: fix cb offset for flushed_depth_textureVadim Girlin2011-11-131-1/+1
| | | | Signed-off-by: Vadim Girlin <[email protected]>
* r600g: fix stencil buffer ref counting on evergreenVadim Girlin2011-11-131-1/+4
| | | | Signed-off-by: Vadim Girlin <[email protected]>
* r600g: lazy load for AR registerVadim Girlin2011-11-133-46/+61
| | | | | | Emit MOVA* instruction only when AR is used. Signed-off-by: Vadim Girlin <[email protected]>
* r600g: include INTERP_[XY|ZW] in is_alu_vec_unit_instVadim Girlin2011-11-131-1/+3
| | | | | | This will disallow moving them to the trans slot in merge_inst_groups Signed-off-by: Vadim Girlin <[email protected]>
* r600g: don't set the query result to 0 after a call to get_query_resultMarek Olšák2011-11-111-2/+0
| | | | get_query_result doesn't reset the result. Only begin_query does.
* r600g: the type of OCCLUSION_PREDICATE result should be booleanMarek Olšák2011-11-113-11/+17
|
* r600g: implement PIPE_QUERY_OCCLUSION_PREDICATEMarek Olšák2011-11-101-5/+21
|
* r600g: remove useless texture barrierMarek Olšák2011-11-101-2/+0
|
* r600g: flush the whole context when CS is full, not just hw_contextMarek Olšák2011-11-103-1/+6
| | | | Because we disable render condition in r600_flush, but not in r600_context_flush.
* r600g: turn some CS overflow checks into assertionsMarek Olšák2011-11-102-12/+2
| | | | They're not really exhaustive and not so useful either.
* r600g: put the rest of CS overflow checks in r600_need_cs_spaceMarek Olšák2011-11-103-18/+9
|
* r600g: flush caches regardless of render conditionMarek Olšák2011-11-101-5/+5
| | | | What if somebody enables render condition just before we flush...
* r600g: disable render condition at the end of CS, re-enable at the beginningMarek Olšák2011-11-102-0/+19
|
* r600g: don't suspend queries if they have already been suspended by blitterMarek Olšák2011-11-101-2/+11
| | | | And add some assertions.
* r600g: properly reserve CS space for queries_suspendMarek Olšák2011-11-103-21/+18
|
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-23 01:46:49 +0000