[PATCH] D129072: [RISCV] Fix the scale of getIntMatCost and also need adjust for getIntImmCost

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/lib/Target/RISCV/MCTargetDesc/RISCVMatInt.cpp:395
   }
-  return std::max(1, Cost);
+  return Cost;
 }
----------------
Can `Cost` ever be 0?


================
Comment at: llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp:38
   return RISCVMatInt::getIntMatCost(Imm, DL.getTypeSizeInBits(Ty),
-                                    getST()->getFeatureBits());
+                                    getST()->getFeatureBits()) / 100;
 }
----------------
This is now an assumption that the cost is >= 100 or it will produce 0. This is why I don't really like this patch. The compressed cost with its 75 and 100 only make sense for a relative comparison not an absolute cost.

Perhaps the real problem is that getIntMatCost is the external interface we use for both cases. Maybe we need a getIntMatSize for the one case that cares about compressed size.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D129072/new/

https://reviews.llvm.org/D129072