[llvm] [AMDGPU] Add waterfall intrinsics (PR #192409)

[Petar Avramovic via llvm-commits]

================
@@ -0,0 +1,11412 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -global-isel=0 -march=amdgcn -mcpu=fiji -verify-machineinstrs < %s | FileCheck -check-prefixes=PRE-GFX10,VI,VI-SDAG %s
+; RUN: llc -global-isel=1 -march=amdgcn -mcpu=fiji -verify-machineinstrs < %s | FileCheck -check-prefixes=PRE-GFX10,VI,VI-GISEL %s
----------------
petar-avramovic wrote:

can we get `-global-isel=1 -new-reg-bank-select` instead?

hint:
```
  addRulesForIOpcs({amdgcn_waterfall_begin})
      .Any({{UniB32, _, UniB32, BRC}, {{SgprB32}, {IntrId, SgprB32, VgprBRC}}});

  addRulesForIOpcs({amdgcn_waterfall_readfirstlane})
      .Any({{UniBRC}, {{SgprBRC}, {IntrId, SgprB32, VgprBRC}}});

  addRulesForIOpcs({amdgcn_waterfall_end})
      .Any({{DivBRC}, {{VgprBRC}, {IntrId, SgprB32, VgprBRC}}});
```

https://github.com/llvm/llvm-project/pull/192409