[llvm] [RISCV][CostModel] Add cost for fabs/fsqrt of type bf16/f16 (PR #118608)

[Luke Lau via llvm-commits]

================
@@ -1035,21 +1035,40 @@ RISCVTTIImpl::getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
     }
     break;
   }
-  case Intrinsic::fabs:
+  case Intrinsic::fabs: {
+    auto LT = getTypeLegalizationCost(RetTy);
+    if (ST->hasVInstructions() && LT.second.isVector()) {
+      // lui a0, 8
+      // addi a0, a0, -1
+      // vsetvli a1, zero, e16, m1, ta, ma
+      // vand.vx v8, v8, a0
+      // f16 with zvfhmin and bf16 with zvfhbmin
+      if (LT.second.getVectorElementType() == MVT::bf16 ||
+          (LT.second.getVectorElementType() == MVT::f16 &&
+           !ST->hasVInstructionsF16()))
+        return LT.first * getRISCVInstructionCost(RISCV::VAND_VX, LT.second,
+                                                  CostKind) +
+               2;
+      else
+        return LT.first *
+               getRISCVInstructionCost(RISCV::VFSGNJX_VV, LT.second, CostKind);
+    }
+    break;
+  }
   case Intrinsic::sqrt: {
     auto LT = getTypeLegalizationCost(RetTy);
-    // TODO: add f16/bf16, bf16 with zvfbfmin && f16 with zvfhmin
     if (ST->hasVInstructions() && LT.second.isVector()) {
-      unsigned Op;
-      switch (ICA.getID()) {
-      case Intrinsic::fabs:
-        Op = RISCV::VFSGNJX_VV;
-        break;
-      case Intrinsic::sqrt:
-        Op = RISCV::VFSQRT_V;
-        break;
-      }
-      return LT.first * getRISCVInstructionCost(Op, LT.second, CostKind);
+      SmallVector<unsigned, 3> Opcodes;
+      // f16 with zvfhmin and bf16 with zvfbfmin
+      if (LT.second.getVectorElementType() == MVT::bf16)
+        Opcodes = {RISCV::VFWCVTBF16_F_F_V, RISCV::VFSQRT_V,
+                   RISCV::VFNCVTBF16_F_F_W};
+      else if (LT.second.getVectorElementType() == MVT::f16 &&
+               !ST->hasVInstructionsF16())
+        Opcodes = {RISCV::VFWCVT_F_F_V, RISCV::VFSQRT_V, RISCV::VFNCVT_F_F_W};
+      else
+        Opcodes = {RISCV::VFSQRT_V};
+      return LT.first * getRISCVInstructionCost(Opcodes, LT.second, CostKind);
----------------
lukel97 wrote:

I think both VFWCVT_F_F_V and VFNCVT_F_F_W should use the widened LMUL type, since presumably we should cost things based on the largest EMUL in the instruction, and that seems to match the spacemit-x60? https://camel-cdr.github.io/rvv-bench-results/bpi_f3/index.html

The easiest thing is probably just to update LT.second's element type to f32 to double the LMUL.

I'm not sure it's worthwhile trying to be fancy and handle it in getRISCVInstructionCost

https://github.com/llvm/llvm-project/pull/118608