[llvm] [X86][NewPM] Port X86LowerAMXType to NewPM (PR #165084)

[Matt Arsenault via llvm-commits]

================
@@ -1432,8 +1434,58 @@ bool X86LowerAMXCast::transformAllAMXCast() {
   return Change;
 }
 
+bool lowerAmxType(Function &F, const TargetMachine *TM,
+                  TargetLibraryInfo *TLI) {
+  // Performance optimization: most code doesn't use AMX, so return early if
+  // there are no instructions that produce AMX values. This is sufficient, as
+  // AMX arguments and constants are not allowed -- so any producer of an AMX
+  // value must be an instruction.
+  // TODO: find a cheaper way for this, without looking at all instructions.
+  if (!containsAMXCode(F))
+    return false;
+
+  bool C = false;
+  ShapeCalculator SC(TM);
+  X86LowerAMXCast LAC(F, &SC);
+  C |= LAC.combineAMXcast(TLI);
+  // There might be remaining AMXcast after combineAMXcast and they should be
+  // handled elegantly.
+  C |= LAC.transformAllAMXCast();
+
+  X86LowerAMXType LAT(F, &SC);
+  C |= LAT.visit();
+
+  // Prepare for fast register allocation at O0.
+  // Todo: May better check the volatile model of AMX code, not just
+  // by checking Attribute::OptimizeNone and CodeGenOptLevel::None.
+  if (TM->getOptLevel() == CodeGenOptLevel::None) {
+    // If Front End not use O0 but the Mid/Back end use O0, (e.g.
+    // "Clang -O2 -S -emit-llvm t.c" + "llc t.ll") we should make
+    // sure the amx data is volatile, that is nessary for AMX fast
----------------
arsenm wrote:

```suggestion
    // sure the amx data is volatile, that is necessary for AMX fast
```

https://github.com/llvm/llvm-project/pull/165084