[llvm] [AArch64][MachineCombiner] Reassociate long chains of accumulation instructions into a tree to increase ILP (PR #126060)

[Jonathan Cohen via llvm-commits]

================
@@ -6674,6 +6674,118 @@ static bool getMaddPatterns(MachineInstr &Root,
   }
   return Found;
 }
+
+bool AArch64InstrInfo::isAccumulationOpcode(unsigned Opcode) const {
+  switch (Opcode) {
+  default:
+    break;
+  case AArch64::UABALB_ZZZ_D:
+  case AArch64::UABALB_ZZZ_H:
+  case AArch64::UABALB_ZZZ_S:
+  case AArch64::UABALT_ZZZ_D:
+  case AArch64::UABALT_ZZZ_H:
+  case AArch64::UABALT_ZZZ_S:
+  case AArch64::UABALv16i8_v8i16:
+  case AArch64::UABALv2i32_v2i64:
+  case AArch64::UABALv4i16_v4i32:
+  case AArch64::UABALv4i32_v2i64:
+  case AArch64::UABALv8i16_v4i32:
+  case AArch64::UABALv8i8_v8i16:
+  case AArch64::UABAv16i8:
+  case AArch64::UABAv2i32:
+  case AArch64::UABAv4i16:
+  case AArch64::UABAv4i32:
+  case AArch64::UABAv8i16:
+  case AArch64::UABAv8i8:
+  case AArch64::SABALv16i8_v8i16:
+  case AArch64::SABALv2i32_v2i64:
+  case AArch64::SABALv4i16_v4i32:
+  case AArch64::SABALv4i32_v2i64:
+  case AArch64::SABALv8i16_v4i32:
+  case AArch64::SABALv8i8_v8i16:
+  case AArch64::SABAv16i8:
+  case AArch64::SABAv2i32:
+  case AArch64::SABAv4i16:
+  case AArch64::SABAv4i32:
+  case AArch64::SABAv8i16:
+  case AArch64::SABAv8i8:
+  case AArch64::MLAv8i8:
+    return true;
+  }
+
+  return false;
+}
+
+std::optional<unsigned> AArch64InstrInfo::getAccumulationStartOpcode(
+    unsigned AccumulationOpcode) const {
+  switch (AccumulationOpcode) {
+  default:
+    llvm_unreachable("Unknown accumulator opcode");
----------------
jcohen-apple wrote:

Ah, I see what you mean. My thinking was to get these opcodes before applying the pattern, and if for some reason an opcode is missing to fail silently and not apply the optimization, rather than crashing the compiler. 

https://github.com/llvm/llvm-project/pull/126060