From 02e9773bc8526f64e4d79e3d9ac11f49882c022f Mon Sep 17 00:00:00 2001 From: Neil Roberts Date: Thu, 23 Apr 2015 16:56:53 -0700 Subject: i965/fs: Strip trailing constant zeroes in sample messages If a send message is emitted with a message length that is less than required for the message then the remaining parameters default to zero. We can take advantage of this to save a register when a shader passes constant zeroes as the final coordinates to the sample function. I think this might be useful for GLES applications that are using 2D textures to simulate 1D textures. On Skylake it will be useful for shaders that do texelFetch(tex,something,0) which I think is fairly common. This helps more on Skylake because in that case the order of the instruction operands are u,v,lod,r which is good for 2D textures whereas before they were u,lod,v,r which is only good for 1D textures. On Haswell: total instructions in shared programs: 8535730 -> 8533261 (-0.03%) instructions in affected programs: 236968 -> 234499 (-1.04%) helped: 1174 On Skylake: total instructions in shared programs: 10345646 -> 10341237 (-0.04%) instructions in affected programs: 293011 -> 288602 (-1.50%) helped: 1218 Reviewed-by: Matt Turner v2: Applied suggestions by Kenneth Graunke: - Only apply on Gen5+ - Apply to all texture opcodes, not just TEX and TXF. Moved the optimisation into the loop as suggested by Matt Turner. Fix the array index when there is a header. --- src/mesa/drivers/dri/i965/brw_fs.cpp | 49 ++++++++++++++++++++++++++++++++++++ src/mesa/drivers/dri/i965/brw_fs.h | 1 + 2 files changed, 50 insertions(+) diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp index a6483230e52..ed86e0b291c 100644 --- a/src/mesa/drivers/dri/i965/brw_fs.cpp +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp @@ -2535,6 +2535,54 @@ fs_visitor::opt_algebraic() return progress; } +/** + * Optimize sample messages that have constant zero values for the trailing + * texture coordinates. We can just reduce the message length for these + * instructions instead of reserving a register for it. Trailing parameters + * that aren't sent default to zero anyway. This will cause the dead code + * eliminator to remove the MOV instruction that would otherwise be emitted to + * set up the zero value. + */ +bool +fs_visitor::opt_zero_samples() +{ + /* Gen4 infers the texturing opcode based on the message length so we can't + * change it. + */ + if (devinfo->gen < 5) + return false; + + bool progress = false; + + foreach_block_and_inst(block, fs_inst, inst, cfg) { + if (!inst->is_tex()) + continue; + + fs_inst *load_payload = (fs_inst *) inst->prev; + + if (load_payload->is_head_sentinel() || + load_payload->opcode != SHADER_OPCODE_LOAD_PAYLOAD) + continue; + + /* We don't want to remove the message header. Removing all of the + * parameters is avoided because it seems to cause a GPU hang but I + * can't find any documentation indicating that this is expected. + */ + while (inst->mlen > inst->header_present + dispatch_width / 8 && + load_payload->src[(inst->mlen - inst->header_present) / + (dispatch_width / 8) + + inst->header_present - 1].is_zero()) { + inst->mlen -= dispatch_width / 8; + progress = true; + } + } + + if (progress) + invalidate_live_intervals(); + + return progress; +} + /** * Optimize sample messages which are followed by the final RT write. * @@ -3818,6 +3866,7 @@ fs_visitor::optimize() OPT(opt_register_renaming); OPT(opt_redundant_discard_jumps); OPT(opt_saturate_propagation); + OPT(opt_zero_samples); OPT(register_coalesce); OPT(compute_to_mrf); diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h index 4e17d444747..6200deb7218 100644 --- a/src/mesa/drivers/dri/i965/brw_fs.h +++ b/src/mesa/drivers/dri/i965/brw_fs.h @@ -324,6 +324,7 @@ public: bool opt_peephole_predicated_break(); bool opt_saturate_propagation(); bool opt_cmod_propagation(); + bool opt_zero_samples(); void emit_bool_to_cond_code(ir_rvalue *condition); void emit_bool_to_cond_code_of_reg(ir_expression *expr, fs_reg op[3]); void emit_if_gen6(ir_if *ir); -- cgit v1.2.3