summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2007-03-20Add a state-checking stage to the render pipeline for attribute size changes.index-swtnl-0.1Keith Whitwell14-76/+162
Use this to correctly set the vertex format for texture coordinates.
2007-03-19make sure bindprogram notifications are made when texenvprogram activeKeith Whitwell1-11/+16
2007-03-19make the link between state emit and batch packet size strongerKeith Whitwell3-48/+57
2007-03-19Begin to deal with full batch buffers.Keith Whitwell28-357/+558
Move the state differencer out of the regular state mechanism. Split state handling into two steps - first, update software state and calculate what we think hardware state should be. This is the old state mechanism. The new, second phase which is delayed until the very point at which triangles or other primitives are emitted to the hardware is to perform the state difference step and emit the state to hardware. This is done with the information about the required batchbuffer space for the primitive emit, so that there is a guarentee that the state emit and primitive can both fit in the current batch buffer. If the batchbuffer wraps, the full state is emitted to cope with potential lost contexts. Also, split the vertex-buffer handling out of intel_idx_render.c and make it a standalone facility. Move the old intel_tris.c/intel_render.c code to use this facility to allocate vb space for their primitives and have them issue indirect primitives rather than inline ones. Reinstate the render stage in the pipeline above the indexed render stage, as it should be slightly more efficient when it is available.
2007-03-19Fix minor glitches - packet name, unused variableKeith Whitwell3-10/+4
2007-03-19add DEBUG_ALWAYS_SYNC optionKeith Whitwell3-5/+16
2007-03-19clear state offsets, sizes after cache clearKeith Whitwell1-1/+5
2007-03-19improve program disassembly formatting, use _mesa_printfKeith Whitwell2-38/+45
2007-03-18disassemble pixel shader packetsKeith Whitwell1-9/+18
2007-03-18fix packet declarationKeith Whitwell1-2/+2
2007-03-18fix packet size calculationKeith Whitwell1-1/+1
2007-03-18add a couple of blit commandsKeith Whitwell1-1/+8
2007-03-18fix program and constant packet size calculationKeith Whitwell1-2/+2
2007-03-18fix modes4/modes5 mixupKeith Whitwell1-2/+2
2007-03-18fix inline primtiive emit, remove debugKeith Whitwell1-27/+14
2007-03-18fixes for indirect stateKeith Whitwell2-13/+9
2007-03-18add defines for debuggerKeith Whitwell1-2/+12
2007-03-18remove unnecessary zero padKeith Whitwell1-3/+1
2007-03-18remove unused fileKeith Whitwell1-58/+0
2007-03-18Add a basic human-readable batchbuffer dump facility.Keith Whitwell3-19/+165
2007-03-18fix typo in subrect_disable packetKeith Whitwell1-1/+1
2007-03-18Fix several packet emit glitches.Keith Whitwell2-11/+12
Texturing works, but updates aren't being handled correctly.
2007-03-18Add a final state differencing step to minimise emitted state.Keith Whitwell12-519/+643
Rearrange packets so that there is just a single static indirect state. Move stipple to dynamic state. Move all actualy state emit code to the differencer.
2007-03-16Get program parameter tracking and fog working.Keith Whitwell13-248/+101
Fog is implemented with the shader-instruction-appending code in core mesa.
2007-03-16Remove const from program _Current pointers.Keith Whitwell1-2/+2
We do want to modify these objects in the drivers, especially when they are really pointing to a driver defined struct derived from the base program structures.
2007-03-16Turn off debugKeith Whitwell1-1/+1
2007-03-16Fix warningKeith Whitwell1-0/+1
2007-03-16Free clipspace vertices in tnl context destroyKeith Whitwell1-0/+2
2007-03-16Get indirect state emit workingKeith Whitwell2-17/+12
2007-03-16Reorganize state to match indirect state types.Keith Whitwell26-1309/+2441
Indirect state emit support added, but disabled. Trivial tri.c works.
2007-03-15Merge branch 'index-swtnl-0.1-origin' into index-swtnl-0.1Keith Whitwell14-539/+478
Conflicts: src/mesa/drivers/dri/i915tex/i915_context.h src/mesa/drivers/dri/i915tex/i915_state.c src/mesa/drivers/dri/i915tex/i915_vtbl.c src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
2007-03-14Turn off lots of debug. Enable dynamic indirect state.Keith Whitwell3-69/+41
2007-03-14Fix off-by-one in LOAD_IMMEDIATE packet sizeKeith Whitwell1-1/+1
2007-03-14Emit sampler and map indirect state.Keith Whitwell1-64/+93
2007-03-14Add intel_batchbuffer_set_reloc - add relocation at arbitary offsetKeith Whitwell2-6/+32
2007-03-14Experiments with indirect state.Keith Whitwell7-161/+406
Split the batchbuffer into 3 segments for the three different types of state (immediate, dynamic-indirect, other-indirect). This is primarily so that relocations continue to work without too many changes. Most stuff broken, but trivial/tri.c works. Dynamic indirect state not working.
2007-03-14Missing filesKeith Whitwell13-0/+3201
2007-03-14Hack to avoid lockup on 2nd & subsequent frames.Keith Whitwell1-0/+6
2007-03-13Fixes, and make state look more like old driver for diffing.Keith Whitwell5-25/+42
2007-03-13Assign AttribPtr[POS] to fix depth output.Keith Whitwell1-1/+6
2007-03-13More fixes.Keith Whitwell2-3/+4
Need to emit some packets at least once, even if their state isn't active.
2007-03-13Fix basic blend and texture state.Keith Whitwell6-52/+53
2007-03-13trivial/tri.c worksKeith Whitwell10-86/+80
2007-03-13Major state rework to match i965 driver. Builds.Keith Whitwell44-5391/+1449
2007-03-08Beginnings of i915 state rework. Won't compile.Keith Whitwell24-2330/+2761
2007-02-16Fix thinko trying to remap buffer objects immediately after use.Keith Whitwell1-6/+9
Would like a way to keep a buffer object mapped in the situation where it is being stuffed with vertices incrementally.
2007-02-16Increase vb->Size to allow space for clipped vertices in t_vb_index.cKeith Whitwell1-1/+1
2007-02-16Enable debugging on debug builds...Keith Whitwell1-2/+2
2007-02-16Fix a couple more clipping bugs.Keith Whitwell2-2/+5
2007-02-16Update VB->Count after clipping.Keith Whitwell1-2/+6
Don't keep overwriting vertices generated in clipping. We need to keep them around until the primitives are emitted.
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-03 18:50:55 +0000