Welcome to ftp.vim.org,
Hosted by ftp.nluug.nl Current directory: /pub/NetBSD/NetBSD-release-9/xsrc/external/mit/MesaLib/dist/src/compiler/nir/ |
Contents of README:New IR, or NIR, is an IR for Mesa intended to sit below GLSL IR and Mesa IR. Its design inherits from the various IRs that Mesa has used in the past, as well as Direct3D assembly, and it includes a few new ideas as well. It is a flat (in terms of using instructions instead of expressions), typeless IR, similar to TGSI and Mesa IR. It also supports SSA (although it doesn't require it). Variables ========= NIR includes support for source-level GLSL variables through a structure mostly copied from GLSL IR. These will be used for linking and conversion from GLSL IR (and later, from an AST), but for the most part, they will be lowered to registers (see below) and loads/stores. Registers ========= Registers are light-weight; they consist of a structure that only contains its size, its index for liveness analysis, and an optional name for debugging. In addition, registers can be local to a function or global to the entire shader; the latter will be used in ARB_shader_subroutine for passing parameters and getting return values from subroutines. Registers can also be an array, in which case they can be accessed indirectly. Each ALU instruction (add, subtract, etc.) works directly with registers or SSA values (see below). SSA ======== Everywhere a register can be loaded/stored, an SSA value can be used instead. The only exception is that arrays/indirect addressing are not supported with SSA; although research has been done on extensions of SSA to arrays before, it's usually for the purpose of parallelization (which we're not interested in), and adds some overhead in the form of adding copies or extra arrays (which is much more expensive than introducing copies between non-array registers). SSA uses point directly to their corresponding definition, which in turn points to the instruction it is part of. This creates an implicit use-def chain and avoids the need for an external structure for each SSA register. Functions ========= Support for function calls is mostly similar to GLSL IR. Each shader contains a list of functions, and each function has a list of overloads. Each overload contains a list of parameters, and may contain an implementation which specifies the variables that correspond to the parameters and return value. Inlining a function, assuming it has a single return point, is as simple as copying its instructions, registers, and local variables into the target function and then inserting copies to and from the new parameters as appropriate. After functions are inlined and any non-subroutine functions are deleted, parameters and return variables will be converted to global variables and then global registers. We don't do this lowering earlier (i.e. the fortranizer idea) for a few reasons: - If we want to do optimizations before link time, we need to have the function signature available during link-time. - If we do any inlining before link time, then we might wind up with the inlined function and the non-inlined function using the same global variables/registers which would preclude optimization. Intrinsics ========= Any operation (other than function calls and textures) which touches a variable or is not referentially transparent is represented by an intrinsic. Intrinsics are similar to the idea of a "builtin function," i.e. a function declaration whose implementation is provided by the backend, except they are more powerful in the following ways: - They can also load and store registers when appropriate, which limits the number of variables needed in later stages of the IR while obviating the need for a separate load/store variable instruction. - Intrinsics can be marked as side-effect free, which permits them to be treated like any other instruction when it comes to optimizations. This allows load intrinsics to be represented as intrinsics while still being optimized away by dead code elimination, common subexpression elimination, etc. Intrinsics are used for: - Atomic operations - Memory barriers - Subroutine calls - Geometry shader emitVertex and endPrimitive - Loading and storing variables (before lowering) - Loading and storing uniforms, shader inputs and outputs, etc (after lowering) - Copying variables (cases where in GLSL the destination is a structure or array) - The kitchen sink - ... Textures ========= Unfortunately, there are far too many texture operations to represent each one of them with an intrinsic, so there's a special texture instruction similar to the GLSL IR one. The biggest difference is that, while the texture instruction has a sampler dereference field used just like in GLSL IR, this gets lowered to a texture unit index (with a possible indirect offset) while the type information of the original sampler is kept around for backends. Also, all the non-constant sources are stored in a single array to make it easier for optimization passes to iterate over all the sources. Control Flow ========= Like in GLSL IR, control flow consists of a tree of "control flow nodes", which include if statements and loops, and jump instructions (break, continue, and return). Unlike GLSL IR, though, the leaves of the tree aren't statements but basic blocks. Each basic block also keeps track of its successors and predecessors, and function implementations keep track of the beginning basic block (the first basic block of the function) and the ending basic block (a fake basic block that every return statement points to). Together, these elements make up the control flow graph, in this case a redundant piece of information on top of the control flow tree that will be used by almost all the optimizations. There are helper functions to add and remove control flow nodes that also update the control flow graph, and so usually it doesn't need to be touched by passes that modify control flow nodes. |
Name Last modified Size
Parent Directory - CVS/ 19-Mar-2020 23:41 - tests/ 19-Mar-2020 23:39 - README 10-Mar-2019 04:42 5.7K meson.build 10-Mar-2019 04:42 7.4K nir.c 10-Mar-2019 04:42 53K nir.h 10-Mar-2019 04:42 93K nir_algebraic.py 10-Mar-2019 04:42 24K nir_builder.h 10-Mar-2019 04:42 25K nir_builder_opcodes.h 10-Mar-2019 04:42 45K nir_builder_opcodes_h.py 10-Mar-2019 04:42 2.6K nir_builtin_builder.c 10-Mar-2019 04:42 2.5K nir_builtin_builder.h 10-Mar-2019 04:42 2.7K nir_clone.c 10-Mar-2019 04:42 23K nir_constant_expressions.c 10-Mar-2019 04:42 317K nir_constant_expressions.h 10-Mar-2019 04:42 1.5K nir_constant_expressions.py 10-Mar-2019 04:42 12K nir_control_flow.c 10-Mar-2019 04:42 22K nir_control_flow.h 10-Mar-2019 04:42 6.4K nir_control_flow_private.h 10-Mar-2019 04:42 1.5K nir_deref.c 10-Mar-2019 04:42 16K nir_deref.h 10-Mar-2019 04:42 2.6K nir_dominance.c 10-Mar-2019 04:42 7.7K nir_format_convert.h 10-Mar-2019 04:42 13K nir_from_ssa.c 10-Mar-2019 04:42 33K nir_gather_info.c 10-Mar-2019 04:42 13K nir_gather_xfb_info.c 01-Jun-2019 09:41 6.4K nir_gs_count_vertices.c 10-Mar-2019 04:42 3.0K nir_inline_functions.c 10-Mar-2019 04:42 5.1K nir_instr_set.c 10-Mar-2019 04:42 18K nir_instr_set.h 10-Mar-2019 04:42 2.4K nir_intrinsics.c 10-Mar-2019 04:42 48K nir_intrinsics.h 10-Mar-2019 04:42 9.3K nir_intrinsics.py 10-Mar-2019 04:42 26K nir_intrinsics_c.py 10-Mar-2019 04:42 2.4K nir_intrinsics_h.py 10-Mar-2019 04:42 2.0K nir_linking_helpers.c 10-Mar-2019 04:42 20K nir_liveness.c 10-Mar-2019 04:42 9.5K nir_loop_analyze.c 10-Mar-2019 04:42 26K nir_loop_analyze.h 10-Mar-2019 04:42 3.4K nir_lower_alpha_test.c 10-Mar-2019 04:42 4.3K nir_lower_alu.c 10-Mar-2019 04:42 8.9K nir_lower_alu_to_scalar.c 10-Mar-2019 04:42 9.6K nir_lower_atomics_to_ssbo.c 10-Mar-2019 04:42 8.7K nir_lower_bit_size.c 10-Mar-2019 04:42 4.1K nir_lower_bitmap.c 10-Mar-2019 04:42 4.8K nir_lower_clamp_color_outputs.c 10-Mar-2019 04:42 3.8K nir_lower_clip.c 10-Mar-2019 04:42 10K nir_lower_clip_cull_distance_arrays.c 10-Mar-2019 04:42 6.7K nir_lower_constant_initializers.c 10-Mar-2019 04:42 4.7K nir_lower_double_ops.c 10-Mar-2019 04:42 20K nir_lower_drawpixels.c 10-Mar-2019 04:42 9.0K nir_lower_global_vars_to_local.c 10-Mar-2019 04:42 3.5K nir_lower_gs_intrinsics.c 10-Mar-2019 04:42 7.0K nir_lower_idiv.c 10-Mar-2019 04:42 4.1K nir_lower_indirect_derefs.c 10-Mar-2019 04:42 7.4K nir_lower_int64.c 10-Mar-2019 04:42 9.7K nir_lower_io.c 10-Mar-2019 04:42 18K nir_lower_io_arrays_to_elements.c 10-Mar-2019 04:42 15K nir_lower_io_to_scalar.c 10-Mar-2019 04:42 13K nir_lower_io_to_temporaries.c 10-Mar-2019 04:42 6.6K nir_lower_load_const_to_scalar.c 10-Mar-2019 04:42 3.7K nir_lower_locals_to_regs.c 10-Mar-2019 04:42 8.8K nir_lower_packing.c 10-Mar-2019 04:42 5.1K nir_lower_passthrough_edgeflags.c 10-Mar-2019 04:42 2.0K nir_lower_patch_vertices.c 10-Mar-2019 04:42 3.9K nir_lower_phis_to_scalar.c 10-Mar-2019 04:42 11K nir_lower_regs_to_ssa.c 10-Mar-2019 04:42 9.2K nir_lower_returns.c 10-Mar-2019 04:42 9.5K nir_lower_subgroups.c 10-Mar-2019 04:42 19K nir_lower_system_values.c 10-Mar-2019 04:42 8.4K nir_lower_tex.c 10-Mar-2019 04:42 29K nir_lower_to_source_mods.c 10-Mar-2019 04:42 6.4K nir_lower_two_sided_color.c 10-Mar-2019 04:42 5.9K nir_lower_var_copies.c 10-Mar-2019 04:42 5.4K nir_lower_vars_to_ssa.c 10-Mar-2019 04:42 25K nir_lower_vec_to_movs.c 10-Mar-2019 04:42 10K nir_lower_wpos_center.c 10-Mar-2019 04:42 4.4K nir_lower_wpos_ytransform.c 10-Mar-2019 04:42 14K nir_metadata.c 10-Mar-2019 04:42 3.0K nir_move_load_const.c 10-Mar-2019 04:42 4.2K nir_move_vec_src_uses_to_dest.c 10-Mar-2019 04:42 6.6K nir_normalize_cubemap_coords.c 10-Mar-2019 04:42 3.8K nir_opcodes.c 10-Mar-2019 04:42 58K nir_opcodes.h 10-Mar-2019 04:42 5.6K nir_opcodes.py 10-Mar-2019 04:42 28K nir_opcodes_c.py 10-Mar-2019 04:42 5.1K nir_opcodes_h.py 10-Mar-2019 04:42 1.6K nir_opt_algebraic.c 10-Mar-2019 04:42 883K nir_opt_algebraic.py 10-Mar-2019 04:42 45K nir_opt_conditional_discard.c 10-Mar-2019 04:42 4.2K nir_opt_constant_folding.c 10-Mar-2019 04:42 6.7K nir_opt_copy_prop_vars.c 01-Jun-2019 09:41 30K nir_opt_copy_propagate.c 10-Mar-2019 04:42 9.3K nir_opt_cse.c 10-Mar-2019 04:42 2.7K nir_opt_dce.c 10-Mar-2019 04:42 4.5K nir_opt_dead_cf.c 10-Mar-2019 04:42 11K nir_opt_dead_write_vars.c 10-Mar-2019 04:42 7.3K nir_opt_find_array_copies.c 10-Mar-2019 04:42 13K nir_opt_gcm.c 10-Mar-2019 04:42 17K nir_opt_global_to_local.c 10-Mar-2019 04:42 2.9K nir_opt_if.c 10-Mar-2019 04:42 20K nir_opt_intrinsics.c 10-Mar-2019 04:42 2.8K nir_opt_large_constants.c 10-Mar-2019 04:42 11K nir_opt_loop_unroll.c 10-Mar-2019 04:42 25K nir_opt_move_comparisons.c 10-Mar-2019 04:42 5.6K nir_opt_move_load_ubo.c 10-Mar-2019 04:42 3.5K nir_opt_peephole_select.c 10-Mar-2019 04:42 7.7K nir_opt_remove_phis.c 10-Mar-2019 04:42 5.4K nir_opt_shrink_load.c 10-Mar-2019 04:42 2.2K nir_opt_trivial_continues.c 10-Mar-2019 04:42 4.8K nir_opt_undef.c 10-Mar-2019 04:42 5.1K nir_phi_builder.c 10-Mar-2019 04:42 11K nir_phi_builder.h 10-Mar-2019 04:42 4.7K nir_print.c 10-Mar-2019 04:42 35K nir_propagate_invariant.c 10-Mar-2019 04:42 5.4K nir_remove_dead_variables.c 10-Mar-2019 04:42 6.4K nir_repair_ssa.c 10-Mar-2019 04:42 4.5K nir_search.c 10-Mar-2019 04:42 18K nir_search.h 10-Mar-2019 04:42 4.2K nir_search_helpers.h 10-Mar-2019 04:42 4.7K nir_serialize.c 10-Mar-2019 04:42 34K nir_serialize.h 10-Mar-2019 04:42 1.5K nir_split_per_member_structs.c 10-Mar-2019 04:42 6.8K nir_split_var_copies.c 10-Mar-2019 04:42 5.0K nir_split_vars.c 10-Mar-2019 04:42 54K nir_sweep.c 10-Mar-2019 04:42 4.9K nir_to_lcssa.c 10-Mar-2019 04:42 6.5K nir_validate.c 10-Mar-2019 04:42 40K nir_vla.h 10-Mar-2019 04:42 2.1K nir_worklist.c 10-Mar-2019 04:42 3.4K nir_worklist.h 10-Mar-2019 04:42 4.7K nir_xfb_info.h 10-Mar-2019 04:42 1.9K
NLUUG - Open Systems. Open Standards
Become a member
and get discounts on conferences and more, see the NLUUG website!