summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Merge branch 'mesa'Eric Anholt2010-06-2411-6/+3270
|\ | | | | | | | | | | This brings in the ir_to_mesa.cpp code I've been developing to codegen to the Mesa IR. It does not actually generate a complete Mesa fragment/vertex program yet.
| * ir_to_mesa: Handle a limited subset of matrix multiplication.Eric Anholt2010-06-241-116/+137
| | | | | | | | | | glsl-mvp.vert now generates believable code, and mesa mode fails only 5 tests that master doesn't. I must have left out some asserts...
| * ir_to_mesa: Handle constant matrices.Eric Anholt2010-06-241-3/+3
| | | | | | | | There's not much to it since we're not actually storing constant data yet.
| * ir_to_mesa: Fix copy-and-wasted second argument to compare expresssion ops.Eric Anholt2010-06-241-9/+10
| | | | | | | | Fixes CorrectParse2.vert assertion due to uninitialized values.
| * ir_to_mesa: Don't allocate temps for swizzles.Eric Anholt2010-06-241-3/+2
| | | | | | | | We do them in place by actually, you know, swizzling.
| * ir_to_mesa: Set up storage for uniform vars.Eric Anholt2010-06-241-17/+44
| |
| * ir_to_mesa: Move the classes into the file now that we don't have the burg.Eric Anholt2010-06-243-159/+115
| | | | | | | | | | | | At 1kloc, it doesn't look like I'll want to split the ir_to_mesa file up even once it's feature-complete. Move definitions closer to usage, and prevent rebuilding the world when changing the definitions.
| * ir_to_mesa: Remove old monoburg structure.Eric Anholt2010-06-241-19/+0
| |
| * ir_to_mesa: Restrict dst writemasks like we did in the monoburg setup.Eric Anholt2010-06-241-1/+6
| |
| * ir_to_mesa: Fix copy-and-wasted DIV instruction sequence.Eric Anholt2010-06-241-1/+1
| |
| * ir_to_mesa: Remove the BURG code.Eric Anholt2010-06-244-623/+258
| | | | | | | | | | | | | | | | | | | | | | The promise of the BURG was to recognize multi-instruction sequences and emit reduced sequences for them. It would have worked well for recognizing MUL+ADD -> MAD and possibly even MIN(MAX(val, 0), 1) -> MOV_SAT with some grammar changes. However, that potential benefit in making those optimizations easy is outweighed by the fragility of monoburg, the amount of (incorrect, as I wrote it) code for using it, and the burden it was going to cause for handling operations on aggregate types.
| * ir_to_mesa: Fix mapping of FS texcoord inputs and color output.Eric Anholt2010-06-241-3/+16
| |
| * ir_to_mesa: Try to fix up the dereference handling for the visitor rework.Eric Anholt2010-06-241-24/+15
| | | | | | | | | | One of the gstreamer shaders I play with now compiles, but input mappings are wrong.
| * ir_to_mesa: Implement min and max expressions.Eric Anholt2010-06-242-0/+11
| | | | | | | | fixes glsl-orangebook-ch06-bump.frag.
| * ir_to_mesa: Don't assert over assignments with a constant-true condition.Eric Anholt2010-06-241-1/+7
| |
| * ir_to_mesa: Add support for trunc/ceil/floor.Eric Anholt2010-06-242-3/+22
| |
| * ir_to_mesa: Implement neg expression.Eric Anholt2010-06-242-1/+7
| |
| * ir_to_mesa: Add sin/cos.Eric Anholt2010-06-242-0/+22
| |
| * ir_to_mesa: Start trying to support struct storage.Eric Anholt2010-06-241-15/+75
| |
| * ir_to_mesa: Fix up array indexing.Eric Anholt2010-06-241-7/+2
| | | | | | | | | | The grammar for array_reference_vec4_vec4 was set up wrong, so we weren't generating instructions if necessary for the array index.
| * ir_to_mesa: Remove stale comment about monoburg.Eric Anholt2010-06-241-4/+0
| |
| * ir_to_mesa: Add support for variable indexing of temporary arrays.Eric Anholt2010-06-243-7/+38
| | | | | | | | Fixes loop-01.vert, loop-02.vert.
| * ir_to_mesa: Clean up some handling of builtins and arrays.Eric Anholt2010-06-241-16/+41
| | | | | | | | | | Constant-index dereferences of arrays should work now. One test is regressed, but it should have been failing before this commit, too.
| * ir_to_mesa: Add support for loops.Eric Anholt2010-06-241-17/+70
| | | | | | | | Fixes CorrectParse1 and the glsl2 loop tests that don't use arrays.
| * Make loop jump mode public so I can switch on it.Eric Anholt2010-06-241-3/+5
| |
| * ir_to_mesa: Add logic_or and logic_and to get CorrectFunction1.vert working.Eric Anholt2010-06-241-0/+13
| |
| * ir_to_mesa: add logic_xor to get CorrectParse2.vert working.Eric Anholt2010-06-241-0/+1
| |
| * ir_to_mesa: add logic_not and f2b to get CorrectParse2.frag working.Eric Anholt2010-06-242-0/+32
| |
| * ir_to_mesa: Add support for ir_if.Eric Anholt2010-06-243-30/+125
| |
| * ir_to_mesa: Add support for comparison operations.Eric Anholt2010-06-243-0/+36
| |
| * ir_to_mesa: Introduce shorthand for common Mesa IR emit patterns.Eric Anholt2010-06-243-65/+59
| |
| * ir_to_mesa: Add ir_unop_f2i -> OPCODE_TRUNC.Eric Anholt2010-06-242-1/+12
| |
| * ir_to_mesa: Add codegen for rsq expression operation.Eric Anholt2010-06-242-0/+11
| |
| * ir_to_mesa: Add exp/log expression operations.Eric Anholt2010-06-242-0/+44
| |
| * ir_to_mesa: Add (almost) the rest of the builtin varyings.Eric Anholt2010-06-242-8/+58
| |
| * ir_to_mesa: Support gl_Position output.Eric Anholt2010-06-241-0/+3
| |
| * ir_to_mesa: Support gl_FragData[] output.Eric Anholt2010-06-241-2/+2
| |
| * ir_to_mesa: Support gl_FragData[] output.Eric Anholt2010-06-241-6/+12
| |
| * ir_to_mesa: Start doing some int support.Eric Anholt2010-06-241-1/+8
| |
| * ir_to_mesa: Fix bugs in swizzle handling for scalar operations.Eric Anholt2010-06-241-6/+7
| | | | | | | | | | Looking at a vec2 / float codegen, the writemasks on the RCPs were wrong and the swizzle on the multiply by the RCP results was wrong.
| * ir_to_mesa: Fix copy'n'paste bug where divide multiplied left by 1/left.Eric Anholt2010-06-241-1/+1
| | | | | | | | Multiply left by 1/right, please.
| * ir_to_mesa: Emit more reduced writemasks for ops on small types.Eric Anholt2010-06-242-4/+14
| | | | | | | | | | This should help prevent Mesa from having to be smart to give channel-wise drivers better information.
| * ir_to_mesa: Handle swizzles on LHS of assignment (writemasks).Eric Anholt2010-06-243-31/+68
| |
| * ir_to_mesa: Produce multiple scalar ops when required to produce vec4s.Eric Anholt2010-06-243-9/+59
| | | | | | | | Fixes the code emitted in a test shader for vec2 texcoord / vec2 tex_size.
| * ir_to_mesa: Get temps allocated at the right times.Eric Anholt2010-06-242-31/+20
| | | | | | | | | | | | | | | | | | | | | | The alloced_vec4/vec4 distinction was an experiment to expose the cost of temps to the codegen. But the problem is that the temporary production rule gets called after the emit rule that was using the temp. We could have the args to emit_op be pointers to where the temp would get allocated later, but that seems overly hard while just trying to bring this thing up. Besides, the temps used in expressions bear only the vaguest relation to how many temps will be used after register allocation.
| * ir_to_mesa: Make the first temp index we use 1 to show off bugs.Eric Anholt2010-06-242-6/+8
| | | | | | | | Regs aren't allocated at the right times yet, so we see TEMP[0] a lot.
| * ir_to_mesa: Fix up the assign rule to use left and right correctly.Eric Anholt2010-06-241-2/+2
| | | | | | | | The destination of assign is in left, not in the node itself.
| * ir_to_mesa: Do my best to explain how the codegen rules work.Eric Anholt2010-06-241-0/+52
| |
| * ir_to_mesa: Print out the ir along with the Mesa IR.Eric Anholt2010-06-243-14/+40
| | | | | | | | | | | | | | | | Ideally this would be hooked up by ir_print_visitor dumping into a string that we could include as prog_instruction->Comment when in debug mode, and not try keeping ir_instruction trees around after conversion to Mesa. The ir_print_visitor isn't set up to do that for us today.
| * ir_to_mesa: Fix up src reg swizzling.Eric Anholt2010-06-241-3/+4
| |
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-20 22:04:38 +0000