aco: improve phi affinities with p_split_vector

Totals from 5860 (4.59% of 127638) affected shaders:
VGPRs: 460212 -> 460216 (+0.00%)
CodeSize: 65554356 -> 65464816 (-0.14%)
Instrs: 12655972 -> 12633578 (-0.18%)
Copies: 1309994 -> 1292163 (-1.36%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4990>

diff --git a/src/amd/compiler/aco_register_allocation.cpp b/src/amd/compiler/aco_register_allocation.cpp
index 897a5bc50c709b065b91e3a65ce9ce3859dfcf71..a5c8ba784f4e10f14e19349ce2eb4e11cbcdb408 100644 (file)
--- a/src/amd/compiler/aco_register_allocation.cpp
+++ b/src/amd/compiler/aco_register_allocation.cpp
@@ -64,6 +64,7 @@ struct ra_ctx {
    std::unordered_map<unsigned, phi_info> phi_map;
    std::unordered_map<unsigned, unsigned> affinities;
    std::unordered_map<unsigned, Instruction*> vectors;
+   std::unordered_map<unsigned, Instruction*> split_vectors;
    aco_ptr<Instruction> pseudo_dummy;
    unsigned max_used_sgpr = 0;
    unsigned max_used_vgpr = 0;
@@ -919,6 +920,21 @@ PhysReg get_reg(ra_ctx& ctx,
                 std::vector<std::pair<Operand, Definition>>& parallelcopies,
                 aco_ptr<Instruction>& instr)
 {
+   auto split_vec = ctx.split_vectors.find(temp.id());
+   if (split_vec != ctx.split_vectors.end()) {
+      unsigned offset = 0;
+      for (Definition def : split_vec->second->definitions) {
+         auto affinity_it = ctx.affinities.find(def.tempId());
+         if (affinity_it != ctx.affinities.end() && ctx.assignments[affinity_it->second].assigned) {
+            PhysReg reg = ctx.assignments[affinity_it->second].reg;
+            reg.reg_b -= offset;
+            if (get_reg_specified(ctx, reg_file, temp.regClass(), parallelcopies, instr, reg))
+               return reg;
+         }
+         offset += def.bytes();
+      }
+   }
+
    if (ctx.affinities.find(temp.id()) != ctx.affinities.end() &&
        ctx.assignments[ctx.affinities[temp.id()]].assigned) {
       PhysReg reg = ctx.assignments[ctx.affinities[temp.id()]].reg;
@@ -1435,6 +1451,9 @@ void register_allocation(Program *program, std::vector<TempSet>& live_out_per_bl
                }
             }
 
+            if (instr->opcode == aco_opcode::p_split_vector && instr->operands[0].isFirstKillBeforeDef())
+               ctx.split_vectors[instr->operands[0].tempId()] = instr.get();
+
             /* add operands to live variables */
             for (const Operand& op : instr->operands) {
                if (op.isTemp())