[llvm] [NVPTX] Constant-folding for f2i, d2ui, f2ll etc. (PR #118965)

[Lewis Crawford via llvm-commits]

================
@@ -2300,6 +2360,211 @@ static Constant *ConstantFoldScalarCall1(StringRef Name,
       return ConstantFP::get(Ty->getContext(), U);
     }
 
+    // NVVM float/double to signed/unsigned int32/int64 conversions:
+    switch (IntrinsicID) {
+    // f2i
+    case Intrinsic::nvvm_f2i_rm:
+    case Intrinsic::nvvm_f2i_rn:
+    case Intrinsic::nvvm_f2i_rp:
+    case Intrinsic::nvvm_f2i_rz:
+    case Intrinsic::nvvm_f2i_rm_ftz:
+    case Intrinsic::nvvm_f2i_rn_ftz:
+    case Intrinsic::nvvm_f2i_rp_ftz:
+    case Intrinsic::nvvm_f2i_rz_ftz:
+    // f2ui
+    case Intrinsic::nvvm_f2ui_rm:
+    case Intrinsic::nvvm_f2ui_rn:
+    case Intrinsic::nvvm_f2ui_rp:
+    case Intrinsic::nvvm_f2ui_rz:
+    case Intrinsic::nvvm_f2ui_rm_ftz:
+    case Intrinsic::nvvm_f2ui_rn_ftz:
+    case Intrinsic::nvvm_f2ui_rp_ftz:
+    case Intrinsic::nvvm_f2ui_rz_ftz:
+    // d2i
+    case Intrinsic::nvvm_d2i_rm:
+    case Intrinsic::nvvm_d2i_rn:
+    case Intrinsic::nvvm_d2i_rp:
+    case Intrinsic::nvvm_d2i_rz:
+    // d2ui
+    case Intrinsic::nvvm_d2ui_rm:
+    case Intrinsic::nvvm_d2ui_rn:
+    case Intrinsic::nvvm_d2ui_rp:
+    case Intrinsic::nvvm_d2ui_rz:
+    // f2ll
+    case Intrinsic::nvvm_f2ll_rm:
+    case Intrinsic::nvvm_f2ll_rn:
+    case Intrinsic::nvvm_f2ll_rp:
+    case Intrinsic::nvvm_f2ll_rz:
+    case Intrinsic::nvvm_f2ll_rm_ftz:
+    case Intrinsic::nvvm_f2ll_rn_ftz:
+    case Intrinsic::nvvm_f2ll_rp_ftz:
+    case Intrinsic::nvvm_f2ll_rz_ftz:
+    // f2ull
+    case Intrinsic::nvvm_f2ull_rm:
+    case Intrinsic::nvvm_f2ull_rn:
+    case Intrinsic::nvvm_f2ull_rp:
+    case Intrinsic::nvvm_f2ull_rz:
+    case Intrinsic::nvvm_f2ull_rm_ftz:
+    case Intrinsic::nvvm_f2ull_rn_ftz:
+    case Intrinsic::nvvm_f2ull_rp_ftz:
+    case Intrinsic::nvvm_f2ull_rz_ftz:
+    // d2ll
+    case Intrinsic::nvvm_d2ll_rm:
+    case Intrinsic::nvvm_d2ll_rn:
+    case Intrinsic::nvvm_d2ll_rp:
+    case Intrinsic::nvvm_d2ll_rz:
+    // d2ull
+    case Intrinsic::nvvm_d2ull_rm:
+    case Intrinsic::nvvm_d2ull_rn:
+    case Intrinsic::nvvm_d2ull_rp:
+    case Intrinsic::nvvm_d2ull_rz: {
+      // In float-to-integer conversion, NaN inputs are converted to 0.
+      if (U.isNaN())
+        return ConstantInt::get(Ty, 0);
+
+      APFloat::roundingMode RMode = APFloat::roundingMode::Invalid;
+      switch (IntrinsicID) {
----------------
LewisCrawford wrote:

This individual patch contains 12 intrinsics for each rounding mode, all of which are related to float-to-int conversion.

However, in total, there are 44 intrinsics for each of the 4 rounding modes, meaning a helper function like `getNVVMIntrinsicRoundingMode` would end up with 176 largely unrelated intrinsics in its case statement.

Similarly, checking for the FTZ modifier would require up to 144 unrelated intrinsics in a switch statement.

I think that some ugliness around this code is unavoidable due to the sheer number of different intrinsics required in all the switch statements. However, having huge 176-element switch statements with unrelated intrinsics may be the lesser of two evils here. I'll rewrite it with helper functions now.

https://github.com/llvm/llvm-project/pull/118965