[NVVM] Update intrinsic definitions to include more attributes

[jdoerfert]

Summary

A lot of NVVM intrinsics can use the default intrinsic attributes (e.g.,
nosync, nofree, ...) as well as speculatable. The latter is important
if we want to recompute intrinsics results instead of communicating them
via memory.

I did use default attributes for almost all readnone attributes but
speculatable only where I had reasonable confidence they cannot
experience UB. That said, someone should double check.

TODO: There seem to be various intrinsics marked Commutative which

should not, e.g., fma and div.

[jdoerfert]

@JonChesterfield @arsenm @jhuber6 Something like this should be done for AMDGPU as well in case it wasn't yet.

llvm/include/llvm/IR/IntrinsicsNVVM.td

657

These commutative annotations seem wrong to me. Same for fma.

[tra]

LGTM in general, modulo the comments below.

llvm/include/llvm/IR/IntrinsicsNVVM.td

657

Agreed.

817

I'm not sure what Commutative attribute would mean for FMA. I guess the multiplicand args are commutative with each other, but the addend is not. I guess the Commutative should ideally list the operands it applies to.

1088–1089

Should this be speculatable, too? It just bitcasts two ints to a double. There are no side effects.

4272

Some of the special registers provide clock and performance monitoring counter info. Speculating those may not be a good idea.

[jdoerfert]

llvm/include/llvm/IR/IntrinsicsNVVM.td

817

I'm for removing it here and above.

1088–1089

It escaped my search pattern due to the commutative. Will add speculatable.

4272

As far as I can tell, those are the PTXReadNCSRegIntrinsic_r32 ones below. If any of the IntrNoMem ones would do this, we would CSE them and their usefulness would be rather limited. As long as we allow CSE, I think, all we care about is the potential for UB in case we speculate on them. Reading registers hopefully won't cause UB.

[tra]

llvm/include/llvm/IR/IntrinsicsNVVM.td

817

SGTM.

4272

You're right. These are fine.

[jdoerfert]

Addresses comments

[jdoerfert]

Fix typo

[jdoerfert]

Rebase

[jdoerfert]

llvm/include/llvm/IR/IntrinsicsNVVM.td

4284

I think I forgot to make these DefaultAttr too.

llvm/include/llvm/IR/IntrinsicsNVVM.td

class SHFL_INFO<bit sync, string mode, string type, bit return_pred> {
   list<LLVMType> RetTy = !if(return_pred, [OpType, llvm_i1_ty], [OpType]);
   list<LLVMType> ArgsTy = !if(sync,
     [llvm_i32_ty, OpType, llvm_i32_ty, llvm_i32_ty],
     [OpType, llvm_i32_ty, llvm_i32_ty]);
 }
 
 let TargetPrefix = "nvvm" in {
   def int_nvvm_prmt : GCCBuiltin<"__nvvm_prmt">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
         [IntrNoMem, Commutative]>;
 
 //
 // Min Max
 //
 
   def int_nvvm_fmin_f : GCCBuiltin<"__nvvm_fmin_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_fmin_ftz_f : GCCBuiltin<"__nvvm_fmin_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_fmax_f : GCCBuiltin<"__nvvm_fmax_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty]
-        , [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty]
+        , [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_fmax_ftz_f : GCCBuiltin<"__nvvm_fmax_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_fmin_d : GCCBuiltin<"__nvvm_fmin_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_fmax_d : GCCBuiltin<"__nvvm_fmax_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
 //
 // Multiplication
 //
 
   def int_nvvm_mulhi_i : GCCBuiltin<"__nvvm_mulhi_i">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mulhi_ui : GCCBuiltin<"__nvvm_mulhi_ui">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_mulhi_ll : GCCBuiltin<"__nvvm_mulhi_ll">,
-      Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mulhi_ull : GCCBuiltin<"__nvvm_mulhi_ull">,
-      Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_mul_rn_ftz_f : GCCBuiltin<"__nvvm_mul_rn_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rn_f : GCCBuiltin<"__nvvm_mul_rn_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rz_ftz_f : GCCBuiltin<"__nvvm_mul_rz_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rz_f : GCCBuiltin<"__nvvm_mul_rz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rm_ftz_f : GCCBuiltin<"__nvvm_mul_rm_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rm_f : GCCBuiltin<"__nvvm_mul_rm_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rp_ftz_f : GCCBuiltin<"__nvvm_mul_rp_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rp_f : GCCBuiltin<"__nvvm_mul_rp_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_mul_rn_d : GCCBuiltin<"__nvvm_mul_rn_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rz_d : GCCBuiltin<"__nvvm_mul_rz_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rm_d : GCCBuiltin<"__nvvm_mul_rm_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul_rp_d : GCCBuiltin<"__nvvm_mul_rp_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_mul24_i : GCCBuiltin<"__nvvm_mul24_i">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_mul24_ui : GCCBuiltin<"__nvvm_mul24_ui">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
 //
 // Div
 //
 
   def int_nvvm_div_approx_ftz_f : GCCBuiltin<"__nvvm_div_approx_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_approx_f : GCCBuiltin<"__nvvm_div_approx_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
 
   def int_nvvm_div_rn_ftz_f : GCCBuiltin<"__nvvm_div_rn_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_rn_f : GCCBuiltin<"__nvvm_div_rn_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
 
   def int_nvvm_div_rz_ftz_f : GCCBuiltin<"__nvvm_div_rz_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_rz_f : GCCBuiltin<"__nvvm_div_rz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
 
   def int_nvvm_div_rm_ftz_f : GCCBuiltin<"__nvvm_div_rm_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_rm_f : GCCBuiltin<"__nvvm_div_rm_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
 
   def int_nvvm_div_rp_ftz_f : GCCBuiltin<"__nvvm_div_rp_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_rp_f : GCCBuiltin<"__nvvm_div_rp_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem]>;
 
   def int_nvvm_div_rn_d : GCCBuiltin<"__nvvm_div_rn_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_rz_d : GCCBuiltin<"__nvvm_div_rz_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_rm_d : GCCBuiltin<"__nvvm_div_rm_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem]>;
   def int_nvvm_div_rp_d : GCCBuiltin<"__nvvm_div_rp_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem]>;
 
 //
 // Sad
 //
 
   def int_nvvm_sad_i : GCCBuiltin<"__nvvm_sad_i">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
         [IntrNoMem, Commutative]>;
   def int_nvvm_sad_ui : GCCBuiltin<"__nvvm_sad_ui">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
         [IntrNoMem, Commutative]>;
 
 //
 // Floor  Ceil
 //
 
   def int_nvvm_floor_ftz_f : GCCBuiltin<"__nvvm_floor_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_floor_f : GCCBuiltin<"__nvvm_floor_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_floor_d : GCCBuiltin<"__nvvm_floor_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_ceil_ftz_f : GCCBuiltin<"__nvvm_ceil_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ceil_f : GCCBuiltin<"__nvvm_ceil_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ceil_d : GCCBuiltin<"__nvvm_ceil_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
 //
 // Abs
 //
 
   def int_nvvm_fabs_ftz_f : GCCBuiltin<"__nvvm_fabs_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fabs_f : GCCBuiltin<"__nvvm_fabs_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fabs_d : GCCBuiltin<"__nvvm_fabs_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
 //
 // Round
 //
 
   def int_nvvm_round_ftz_f : GCCBuiltin<"__nvvm_round_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_round_f : GCCBuiltin<"__nvvm_round_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_round_d : GCCBuiltin<"__nvvm_round_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
 //
 // Trunc
 //
 
   def int_nvvm_trunc_ftz_f : GCCBuiltin<"__nvvm_trunc_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_trunc_f : GCCBuiltin<"__nvvm_trunc_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_trunc_d : GCCBuiltin<"__nvvm_trunc_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
 //
 // Saturate
 //
 
   def int_nvvm_saturate_ftz_f : GCCBuiltin<"__nvvm_saturate_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_saturate_f : GCCBuiltin<"__nvvm_saturate_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_saturate_d : GCCBuiltin<"__nvvm_saturate_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
 //
 // Exp2  Log2
 //
 
   def int_nvvm_ex2_approx_ftz_f : GCCBuiltin<"__nvvm_ex2_approx_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_ex2_approx_f : GCCBuiltin<"__nvvm_ex2_approx_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_ex2_approx_d : GCCBuiltin<"__nvvm_ex2_approx_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
 
   def int_nvvm_lg2_approx_ftz_f : GCCBuiltin<"__nvvm_lg2_approx_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_lg2_approx_f : GCCBuiltin<"__nvvm_lg2_approx_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_lg2_approx_d : GCCBuiltin<"__nvvm_lg2_approx_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
 
 //
 // Sin  Cos
 //
 
   def int_nvvm_sin_approx_ftz_f : GCCBuiltin<"__nvvm_sin_approx_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sin_approx_f : GCCBuiltin<"__nvvm_sin_approx_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
 
   def int_nvvm_cos_approx_ftz_f : GCCBuiltin<"__nvvm_cos_approx_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_cos_approx_f : GCCBuiltin<"__nvvm_cos_approx_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
 
 //
 // Fma
 //
 
   def int_nvvm_fma_rn_ftz_f : GCCBuiltin<"__nvvm_fma_rn_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rn_f : GCCBuiltin<"__nvvm_fma_rn_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rz_ftz_f : GCCBuiltin<"__nvvm_fma_rz_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rz_f : GCCBuiltin<"__nvvm_fma_rz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rm_ftz_f : GCCBuiltin<"__nvvm_fma_rm_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rm_f : GCCBuiltin<"__nvvm_fma_rm_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rp_ftz_f : GCCBuiltin<"__nvvm_fma_rp_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rp_f : GCCBuiltin<"__nvvm_fma_rp_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_fma_rn_d : GCCBuiltin<"__nvvm_fma_rn_d">,
-      Intrinsic<[llvm_double_ty],
+      DefaultAttrsIntrinsic<[llvm_double_ty],
         [llvm_double_ty, llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rz_d : GCCBuiltin<"__nvvm_fma_rz_d">,
-      Intrinsic<[llvm_double_ty],
+      DefaultAttrsIntrinsic<[llvm_double_ty],
         [llvm_double_ty, llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rm_d : GCCBuiltin<"__nvvm_fma_rm_d">,
-      Intrinsic<[llvm_double_ty],
+      DefaultAttrsIntrinsic<[llvm_double_ty],
         [llvm_double_ty, llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+        [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_fma_rp_d : GCCBuiltin<"__nvvm_fma_rp_d">,
-      Intrinsic<[llvm_double_ty],
+      DefaultAttrsIntrinsic<[llvm_double_ty],
         [llvm_double_ty, llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+        [IntrNoMem, IntrSpeculatable]>;
 
 //
 // Rcp
 //
 
   def int_nvvm_rcp_rn_ftz_f : GCCBuiltin<"__nvvm_rcp_rn_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rn_f : GCCBuiltin<"__nvvm_rcp_rn_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rz_ftz_f : GCCBuiltin<"__nvvm_rcp_rz_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rz_f : GCCBuiltin<"__nvvm_rcp_rz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rm_ftz_f : GCCBuiltin<"__nvvm_rcp_rm_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rm_f : GCCBuiltin<"__nvvm_rcp_rm_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rp_ftz_f : GCCBuiltin<"__nvvm_rcp_rp_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rp_f : GCCBuiltin<"__nvvm_rcp_rp_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
 
   def int_nvvm_rcp_rn_d : GCCBuiltin<"__nvvm_rcp_rn_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rz_d : GCCBuiltin<"__nvvm_rcp_rz_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rm_d : GCCBuiltin<"__nvvm_rcp_rm_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
   def int_nvvm_rcp_rp_d : GCCBuiltin<"__nvvm_rcp_rp_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
 
   def int_nvvm_rcp_approx_ftz_d : GCCBuiltin<"__nvvm_rcp_approx_ftz_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
 
 //
 // Sqrt
 //
 
   def int_nvvm_sqrt_f : GCCBuiltin<"__nvvm_sqrt_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rn_ftz_f : GCCBuiltin<"__nvvm_sqrt_rn_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rn_f : GCCBuiltin<"__nvvm_sqrt_rn_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rz_ftz_f : GCCBuiltin<"__nvvm_sqrt_rz_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rz_f : GCCBuiltin<"__nvvm_sqrt_rz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rm_ftz_f : GCCBuiltin<"__nvvm_sqrt_rm_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rm_f : GCCBuiltin<"__nvvm_sqrt_rm_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rp_ftz_f : GCCBuiltin<"__nvvm_sqrt_rp_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rp_f : GCCBuiltin<"__nvvm_sqrt_rp_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_approx_ftz_f : GCCBuiltin<"__nvvm_sqrt_approx_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_approx_f : GCCBuiltin<"__nvvm_sqrt_approx_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
 
   def int_nvvm_sqrt_rn_d : GCCBuiltin<"__nvvm_sqrt_rn_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rz_d : GCCBuiltin<"__nvvm_sqrt_rz_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rm_d : GCCBuiltin<"__nvvm_sqrt_rm_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
   def int_nvvm_sqrt_rp_d : GCCBuiltin<"__nvvm_sqrt_rp_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
 
 //
 // Rsqrt
 //
 
   def int_nvvm_rsqrt_approx_ftz_f : GCCBuiltin<"__nvvm_rsqrt_approx_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rsqrt_approx_f : GCCBuiltin<"__nvvm_rsqrt_approx_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>;
   def int_nvvm_rsqrt_approx_d : GCCBuiltin<"__nvvm_rsqrt_approx_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>;
 
 //
 // Add
 //
 
   def int_nvvm_add_rn_ftz_f : GCCBuiltin<"__nvvm_add_rn_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rn_f : GCCBuiltin<"__nvvm_add_rn_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rz_ftz_f : GCCBuiltin<"__nvvm_add_rz_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rz_f : GCCBuiltin<"__nvvm_add_rz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rm_ftz_f : GCCBuiltin<"__nvvm_add_rm_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rm_f : GCCBuiltin<"__nvvm_add_rm_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rp_ftz_f : GCCBuiltin<"__nvvm_add_rp_ftz_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rp_f : GCCBuiltin<"__nvvm_add_rp_f">,
-      Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_add_rn_d : GCCBuiltin<"__nvvm_add_rn_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rz_d : GCCBuiltin<"__nvvm_add_rz_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rm_d : GCCBuiltin<"__nvvm_add_rm_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
   def int_nvvm_add_rp_d : GCCBuiltin<"__nvvm_add_rp_d">,
-      Intrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_double_ty, llvm_double_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
 //
 // Convert
 //
 
   def int_nvvm_d2f_rn_ftz : GCCBuiltin<"__nvvm_d2f_rn_ftz">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2f_rn : GCCBuiltin<"__nvvm_d2f_rn">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2f_rz_ftz : GCCBuiltin<"__nvvm_d2f_rz_ftz">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2f_rz : GCCBuiltin<"__nvvm_d2f_rz">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2f_rm_ftz : GCCBuiltin<"__nvvm_d2f_rm_ftz">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2f_rm : GCCBuiltin<"__nvvm_d2f_rm">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2f_rp_ftz : GCCBuiltin<"__nvvm_d2f_rp_ftz">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2f_rp : GCCBuiltin<"__nvvm_d2f_rp">,
-      Intrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_d2i_rn : GCCBuiltin<"__nvvm_d2i_rn">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2i_rz : GCCBuiltin<"__nvvm_d2i_rz">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2i_rm : GCCBuiltin<"__nvvm_d2i_rm">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2i_rp : GCCBuiltin<"__nvvm_d2i_rp">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_d2ui_rn : GCCBuiltin<"__nvvm_d2ui_rn">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ui_rz : GCCBuiltin<"__nvvm_d2ui_rz">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ui_rm : GCCBuiltin<"__nvvm_d2ui_rm">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ui_rp : GCCBuiltin<"__nvvm_d2ui_rp">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_i2d_rn : GCCBuiltin<"__nvvm_i2d_rn">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_i2d_rz : GCCBuiltin<"__nvvm_i2d_rz">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_i2d_rm : GCCBuiltin<"__nvvm_i2d_rm">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_i2d_rp : GCCBuiltin<"__nvvm_i2d_rp">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_ui2d_rn : GCCBuiltin<"__nvvm_ui2d_rn">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ui2d_rz : GCCBuiltin<"__nvvm_ui2d_rz">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ui2d_rm : GCCBuiltin<"__nvvm_ui2d_rm">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ui2d_rp : GCCBuiltin<"__nvvm_ui2d_rp">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_f2i_rn_ftz : GCCBuiltin<"__nvvm_f2i_rn_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2i_rn : GCCBuiltin<"__nvvm_f2i_rn">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2i_rz_ftz : GCCBuiltin<"__nvvm_f2i_rz_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2i_rz : GCCBuiltin<"__nvvm_f2i_rz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2i_rm_ftz : GCCBuiltin<"__nvvm_f2i_rm_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2i_rm : GCCBuiltin<"__nvvm_f2i_rm">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2i_rp_ftz : GCCBuiltin<"__nvvm_f2i_rp_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2i_rp : GCCBuiltin<"__nvvm_f2i_rp">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_f2ui_rn_ftz : GCCBuiltin<"__nvvm_f2ui_rn_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ui_rn : GCCBuiltin<"__nvvm_f2ui_rn">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ui_rz_ftz : GCCBuiltin<"__nvvm_f2ui_rz_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ui_rz : GCCBuiltin<"__nvvm_f2ui_rz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ui_rm_ftz : GCCBuiltin<"__nvvm_f2ui_rm_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ui_rm : GCCBuiltin<"__nvvm_f2ui_rm">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ui_rp_ftz : GCCBuiltin<"__nvvm_f2ui_rp_ftz">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ui_rp : GCCBuiltin<"__nvvm_f2ui_rp">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_i2f_rn : GCCBuiltin<"__nvvm_i2f_rn">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_i2f_rz : GCCBuiltin<"__nvvm_i2f_rz">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_i2f_rm : GCCBuiltin<"__nvvm_i2f_rm">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_i2f_rp : GCCBuiltin<"__nvvm_i2f_rp">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_ui2f_rn : GCCBuiltin<"__nvvm_ui2f_rn">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ui2f_rz : GCCBuiltin<"__nvvm_ui2f_rz">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ui2f_rm : GCCBuiltin<"__nvvm_ui2f_rm">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ui2f_rp : GCCBuiltin<"__nvvm_ui2f_rp">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_lohi_i2d : GCCBuiltin<"__nvvm_lohi_i2d">,
-      Intrinsic<[llvm_double_ty], [llvm_i32_ty, llvm_i32_ty],
-        [IntrNoMem, Commutative]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i32_ty, llvm_i32_ty],
+        [IntrNoMem, IntrSpeculatable, Commutative]>;
 
   def int_nvvm_d2i_lo : GCCBuiltin<"__nvvm_d2i_lo">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2i_hi : GCCBuiltin<"__nvvm_d2i_hi">,
-      Intrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_f2ll_rn_ftz : GCCBuiltin<"__nvvm_f2ll_rn_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ll_rn : GCCBuiltin<"__nvvm_f2ll_rn">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ll_rz_ftz : GCCBuiltin<"__nvvm_f2ll_rz_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ll_rz : GCCBuiltin<"__nvvm_f2ll_rz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ll_rm_ftz : GCCBuiltin<"__nvvm_f2ll_rm_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ll_rm : GCCBuiltin<"__nvvm_f2ll_rm">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ll_rp_ftz : GCCBuiltin<"__nvvm_f2ll_rp_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ll_rp : GCCBuiltin<"__nvvm_f2ll_rp">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_f2ull_rn_ftz : GCCBuiltin<"__nvvm_f2ull_rn_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ull_rn : GCCBuiltin<"__nvvm_f2ull_rn">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ull_rz_ftz : GCCBuiltin<"__nvvm_f2ull_rz_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ull_rz : GCCBuiltin<"__nvvm_f2ull_rz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ull_rm_ftz : GCCBuiltin<"__nvvm_f2ull_rm_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ull_rm : GCCBuiltin<"__nvvm_f2ull_rm">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ull_rp_ftz : GCCBuiltin<"__nvvm_f2ull_rp_ftz">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2ull_rp : GCCBuiltin<"__nvvm_f2ull_rp">,
-      Intrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_d2ll_rn : GCCBuiltin<"__nvvm_d2ll_rn">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ll_rz : GCCBuiltin<"__nvvm_d2ll_rz">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ll_rm : GCCBuiltin<"__nvvm_d2ll_rm">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ll_rp : GCCBuiltin<"__nvvm_d2ll_rp">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_d2ull_rn : GCCBuiltin<"__nvvm_d2ull_rn">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ull_rz : GCCBuiltin<"__nvvm_d2ull_rz">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ull_rm : GCCBuiltin<"__nvvm_d2ull_rm">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_d2ull_rp : GCCBuiltin<"__nvvm_d2ull_rp">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_ll2f_rn : GCCBuiltin<"__nvvm_ll2f_rn">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ll2f_rz : GCCBuiltin<"__nvvm_ll2f_rz">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ll2f_rm : GCCBuiltin<"__nvvm_ll2f_rm">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ll2f_rp : GCCBuiltin<"__nvvm_ll2f_rp">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2f_rn : GCCBuiltin<"__nvvm_ull2f_rn">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2f_rz : GCCBuiltin<"__nvvm_ull2f_rz">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2f_rm : GCCBuiltin<"__nvvm_ull2f_rm">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2f_rp : GCCBuiltin<"__nvvm_ull2f_rp">,
-      Intrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_ll2d_rn : GCCBuiltin<"__nvvm_ll2d_rn">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ll2d_rz : GCCBuiltin<"__nvvm_ll2d_rz">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ll2d_rm : GCCBuiltin<"__nvvm_ll2d_rm">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ll2d_rp : GCCBuiltin<"__nvvm_ll2d_rp">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2d_rn : GCCBuiltin<"__nvvm_ull2d_rn">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2d_rz : GCCBuiltin<"__nvvm_ull2d_rz">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2d_rm : GCCBuiltin<"__nvvm_ull2d_rm">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_ull2d_rp : GCCBuiltin<"__nvvm_ull2d_rp">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_f2h_rn_ftz : GCCBuiltin<"__nvvm_f2h_rn_ftz">,
-      Intrinsic<[llvm_i16_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i16_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_f2h_rn : GCCBuiltin<"__nvvm_f2h_rn">,
-      Intrinsic<[llvm_i16_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i16_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
 
 //
 // Bitcast
 //
 
   def int_nvvm_bitcast_f2i : GCCBuiltin<"__nvvm_bitcast_f2i">,
-      Intrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_float_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_bitcast_i2f : GCCBuiltin<"__nvvm_bitcast_i2f">,
-      Intrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_float_ty], [llvm_i32_ty], [IntrNoMem, IntrSpeculatable]>;
 
   def int_nvvm_bitcast_ll2d : GCCBuiltin<"__nvvm_bitcast_ll2d">,
-      Intrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_double_ty], [llvm_i64_ty], [IntrNoMem, IntrSpeculatable]>;
   def int_nvvm_bitcast_d2ll : GCCBuiltin<"__nvvm_bitcast_d2ll">,
-      Intrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem]>;
+      DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_double_ty], [IntrNoMem, IntrSpeculatable]>;
 
 // FNS
 
   def int_nvvm_fns : GCCBuiltin<"__nvvm_fns">,
-      Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+      DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
                 [IntrNoMem]>;
 
 // Atomics not available as llvm intrinsics.
   def int_nvvm_atomic_load_inc_32 : Intrinsic<[llvm_i32_ty],
           [LLVMAnyPointerType<llvm_i32_ty>, llvm_i32_ty],
                                       [IntrArgMemOnly, NoCapture<ArgIndex<0>>]>;
   def int_nvvm_atomic_load_dec_32 : Intrinsic<[llvm_i32_ty],
           [LLVMAnyPointerType<llvm_i32_ty>, llvm_i32_ty],

 // Use for generic pointers
 // - These intrinsics are used to convert address spaces.
 // - The input pointer and output pointer must have the same type, except for
 //   the address-space. (This restriction is not enforced here as there is
 //   currently no way to describe it).
 // - This complements the llvm bitcast, which can be used to cast one type
 //   of pointer to another type of pointer, while the address space remains
 //   the same.
-def int_nvvm_ptr_local_to_gen: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_local_to_gen: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.local.to.gen">;
-def int_nvvm_ptr_shared_to_gen: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_shared_to_gen: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.shared.to.gen">;
-def int_nvvm_ptr_global_to_gen: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_global_to_gen: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.global.to.gen">;
-def int_nvvm_ptr_constant_to_gen: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_constant_to_gen: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.constant.to.gen">;
 
-def int_nvvm_ptr_gen_to_global: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_gen_to_global: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.gen.to.global">;
-def int_nvvm_ptr_gen_to_shared: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_gen_to_shared: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.gen.to.shared">;
-def int_nvvm_ptr_gen_to_local: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_gen_to_local: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.gen.to.local">;
-def int_nvvm_ptr_gen_to_constant: Intrinsic<[llvm_anyptr_ty],
-                 [llvm_anyptr_ty], [IntrNoMem],
+def int_nvvm_ptr_gen_to_constant: DefaultAttrsIntrinsic<[llvm_anyptr_ty],
+                 [llvm_anyptr_ty], [IntrNoMem, IntrSpeculatable],
                  "llvm.nvvm.ptr.gen.to.constant">;
 
 // Used in nvvm internally to help address space opt and ptx code generation
 // This is for params that are passed to kernel functions by pointer by-val.
 def int_nvvm_ptr_gen_to_param: Intrinsic<[llvm_anyptr_ty],
                                      [llvm_anyptr_ty],
-                                   [IntrNoMem],
+                                   [IntrNoMem, IntrSpeculatable],
                                    "llvm.nvvm.ptr.gen.to.param">;
 
 // Move intrinsics, used in nvvm internally
 
 def int_nvvm_move_i16 : Intrinsic<[llvm_i16_ty], [llvm_i16_ty], [IntrNoMem],
   "llvm.nvvm.move.i16">;
 def int_nvvm_move_i32 : Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem],
   "llvm.nvvm.move.i32">;

 def int_nvvm_compiler_warn :
     Intrinsic<[], [llvm_anyptr_ty], [], "llvm.nvvm.compiler.warn">;
 
 def int_nvvm_reflect :
   Intrinsic<[llvm_i32_ty], [llvm_anyptr_ty], [IntrNoMem], "llvm.nvvm.reflect">;
 
 // isspacep.{const, global, local, shared}
 def int_nvvm_isspacep_const
-  : Intrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.isspacep.const">,
     GCCBuiltin<"__nvvm_isspacep_const">;
 def int_nvvm_isspacep_global
-  : Intrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.isspacep.global">,
     GCCBuiltin<"__nvvm_isspacep_global">;
 def int_nvvm_isspacep_local
-  : Intrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.isspacep.local">,
     GCCBuiltin<"__nvvm_isspacep_local">;
 def int_nvvm_isspacep_shared
-  : Intrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i1_ty], [llvm_ptr_ty], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.isspacep.shared">,
     GCCBuiltin<"__nvvm_isspacep_shared">;
 
 // Environment register read
 def int_nvvm_read_ptx_sreg_envreg0
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg0">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg0">;
 def int_nvvm_read_ptx_sreg_envreg1
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg1">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg1">;
 def int_nvvm_read_ptx_sreg_envreg2
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg2">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg2">;
 def int_nvvm_read_ptx_sreg_envreg3
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg3">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg3">;
 def int_nvvm_read_ptx_sreg_envreg4
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg4">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg4">;
 def int_nvvm_read_ptx_sreg_envreg5
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg5">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg5">;
 def int_nvvm_read_ptx_sreg_envreg6
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg6">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg6">;
 def int_nvvm_read_ptx_sreg_envreg7
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg7">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg7">;
 def int_nvvm_read_ptx_sreg_envreg8
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg8">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg8">;
 def int_nvvm_read_ptx_sreg_envreg9
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg9">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg9">;
 def int_nvvm_read_ptx_sreg_envreg10
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg10">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg10">;
 def int_nvvm_read_ptx_sreg_envreg11
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg11">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg11">;
 def int_nvvm_read_ptx_sreg_envreg12
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg12">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg12">;
 def int_nvvm_read_ptx_sreg_envreg13
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg13">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg13">;
 def int_nvvm_read_ptx_sreg_envreg14
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg14">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg14">;
 def int_nvvm_read_ptx_sreg_envreg15
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg15">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg15">;
 def int_nvvm_read_ptx_sreg_envreg16
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg16">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg16">;
 def int_nvvm_read_ptx_sreg_envreg17
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg17">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg17">;
 def int_nvvm_read_ptx_sreg_envreg18
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg18">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg18">;
 def int_nvvm_read_ptx_sreg_envreg19
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg19">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg19">;
 def int_nvvm_read_ptx_sreg_envreg20
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg20">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg20">;
 def int_nvvm_read_ptx_sreg_envreg21
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg21">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg21">;
 def int_nvvm_read_ptx_sreg_envreg22
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg22">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg22">;
 def int_nvvm_read_ptx_sreg_envreg23
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg23">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg23">;
 def int_nvvm_read_ptx_sreg_envreg24
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg24">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg24">;
 def int_nvvm_read_ptx_sreg_envreg25
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg25">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg25">;
 def int_nvvm_read_ptx_sreg_envreg26
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg26">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg26">;
 def int_nvvm_read_ptx_sreg_envreg27
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg27">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg27">;
 def int_nvvm_read_ptx_sreg_envreg28
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg28">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg28">;
 def int_nvvm_read_ptx_sreg_envreg29
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg29">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg29">;
 def int_nvvm_read_ptx_sreg_envreg30
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg30">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg30">;
 def int_nvvm_read_ptx_sreg_envreg31
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem],
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable],
               "llvm.nvvm.read.ptx.sreg.envreg31">,
     GCCBuiltin<"__nvvm_read_ptx_sreg_envreg31">;
 
 
 // Texture Fetch
 // texmode_independent
 def int_nvvm_tex_1d_v4f32_s32
   : Intrinsic<[llvm_float_ty, llvm_float_ty, llvm_float_ty, llvm_float_ty],

 def int_nvvm_sust_p_3d_v4i32_trap
   : Intrinsic<[], [llvm_i64_ty, llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
                    llvm_i32_ty, llvm_i32_ty, llvm_i32_ty, llvm_i32_ty], [],
               "llvm.nvvm.sust.p.3d.v4i32.trap">,
     GCCBuiltin<"__nvvm_sust_p_3d_v4i32_trap">;
 
 
 def int_nvvm_rotate_b32
-  : Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
-              [IntrNoMem], "llvm.nvvm.rotate.b32">,
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
+              [IntrNoMem, IntrSpeculatable], "llvm.nvvm.rotate.b32">,
               GCCBuiltin<"__nvvm_rotate_b32">;
 
 def int_nvvm_rotate_b64
-  :Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i32_ty],
-             [IntrNoMem], "llvm.nvvm.rotate.b64">,
+  : DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i32_ty],
+             [IntrNoMem, IntrSpeculatable], "llvm.nvvm.rotate.b64">,
              GCCBuiltin<"__nvvm_rotate_b64">;
 
 def int_nvvm_rotate_right_b64
-  : Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i32_ty],
-              [IntrNoMem], "llvm.nvvm.rotate.right.b64">,
+  : DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i32_ty],
+              [IntrNoMem, IntrSpeculatable], "llvm.nvvm.rotate.right.b64">,
               GCCBuiltin<"__nvvm_rotate_right_b64">;
 
 def int_nvvm_swap_lo_hi_b64
-  : Intrinsic<[llvm_i64_ty], [llvm_i64_ty],
-              [IntrNoMem], "llvm.nvvm.swap.lo.hi.b64">,
+  : DefaultAttrsIntrinsic<[llvm_i64_ty], [llvm_i64_ty],
+              [IntrNoMem, IntrSpeculatable], "llvm.nvvm.swap.lo.hi.b64">,
               GCCBuiltin<"__nvvm_swap_lo_hi_b64">;
 
 
 // Accessing special registers.
 multiclass PTXReadSRegIntrinsic_v4i32<string regname> {
 // FIXME: Do we need the 128-bit integer type version?
-//    def _r64   : Intrinsic<[llvm_i128_ty],   [], [IntrNoMem]>;
+//    def _r64   : Intrinsic<[llvm_i128_ty],   [], [IntrNoMem, IntrSpeculatable]>;
 
 // FIXME: Enable this once v4i32 support is enabled in back-end.
-//    def _v4i16 : Intrinsic<[llvm_v4i32_ty], [], [IntrNoMem]>;
+//    def _v4i16 : Intrinsic<[llvm_v4i32_ty], [], [IntrNoMem, IntrSpeculatable]>;
 
-  def _x     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem]>,
+  def _x     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable]>,
                GCCBuiltin<"__nvvm_read_ptx_sreg_" # regname # "_x">;
-  def _y     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem]>,
+  def _y     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable]>,
                GCCBuiltin<"__nvvm_read_ptx_sreg_" # regname # "_y">;
-  def _z     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem]>,
+  def _z     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable]>,
                GCCBuiltin<"__nvvm_read_ptx_sreg_" # regname # "_z">;
-  def _w     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem]>,
+  def _w     : Intrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable]>,
                GCCBuiltin<"__nvvm_read_ptx_sreg_" # regname # "_w">;
 }
 
 class PTXReadSRegIntrinsic_r32<string name>
-  : Intrinsic<[llvm_i32_ty], [], [IntrNoMem]>,
+  : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrNoMem, IntrSpeculatable]>,
     GCCBuiltin<"__nvvm_read_ptx_sreg_" # name>;
 class PTXReadSRegIntrinsic_r64<string name>
-  : Intrinsic<[llvm_i64_ty], [], [IntrNoMem]>,
+  : DefaultAttrsIntrinsic<[llvm_i64_ty], [], [IntrNoMem, IntrSpeculatable]>,
     GCCBuiltin<"__nvvm_read_ptx_sreg_" # name>;
 
 // Intrinsics to read registers with non-constant values. E.g. the values that
 // do change over the kernel lifetime. Such reads should not be CSE'd.
 class PTXReadNCSRegIntrinsic_r32<string name>
   : Intrinsic<[llvm_i32_ty], [], [IntrInaccessibleMemOnly]>,
     GCCBuiltin<"__nvvm_read_ptx_sreg_" # name>;
 class PTXReadNCSRegIntrinsic_r64<string name>

llvm/test/Transforms/OpenMP/replace_globalization.ll

 ; CHECK: @[[GLOB0:[0-9]+]] = private unnamed_addr constant [113 x i8] c"
 ; CHECK: @[[GLOB1:[0-9]+]] = private unnamed_addr constant [[STRUCT_IDENT_T:%.*]] { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([113 x i8], [113 x i8]* @[[GLOB0]], i32 0, i32 0) }, align 8
 ; CHECK: @[[X:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [16 x i8] undef, align 32
 ; CHECK: @[[Y:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [4 x i8] undef, align 32
 ;.
 ; CHECK-LABEL: define {{[^@]+}}@foo() {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true)
-; CHECK-NEXT:    [[X:%.*]] = call i8* @__kmpc_alloc_shared(i64 4) #[[ATTR4:[0-9]+]]
+; CHECK-NEXT:    [[X:%.*]] = call i8* @__kmpc_alloc_shared(i64 4) #[[ATTR5:[0-9]+]]
 ; CHECK-NEXT:    call void @unknown_no_openmp()
-; CHECK-NEXT:    call void @use.internalized(i8* nofree writeonly [[X]]) #[[ATTR5:[0-9]+]]
-; CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[X]], i64 4) #[[ATTR4]]
+; CHECK-NEXT:    call void @use.internalized(i8* nofree writeonly [[X]]) #[[ATTR6:[0-9]+]]
+; CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[X]], i64 4) #[[ATTR5]]
 ; CHECK-NEXT:    call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
 ; CHECK-NEXT:    ret void
 ;
 ;
 ; CHECK-LABEL: define {{[^@]+}}@bar() {
 ; CHECK-NEXT:    [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true)
 ; CHECK-NEXT:    call void @unknown_no_openmp()
 ; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[C]], -1
 ; CHECK-NEXT:    br i1 [[CMP]], label [[MASTER1:%.*]], label [[EXIT:%.*]]
 ; CHECK:       master1:
-; CHECK-NEXT:    call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([16 x i8], [16 x i8] addrspace(3)* @x, i32 0, i32 0) to i8*)) #[[ATTR5]]
+; CHECK-NEXT:    call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([16 x i8], [16 x i8] addrspace(3)* @x, i32 0, i32 0) to i8*)) #[[ATTR6]]
 ; CHECK-NEXT:    br label [[NEXT:%.*]]
 ; CHECK:       next:
 ; CHECK-NEXT:    call void @unknown_no_openmp()
 ; CHECK-NEXT:    br label [[MASTER2:%.*]]
 ; CHECK:       master2:
-; CHECK-NEXT:    call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([4 x i8], [4 x i8] addrspace(3)* @y, i32 0, i32 0) to i8*)) #[[ATTR5]]
+; CHECK-NEXT:    call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([4 x i8], [4 x i8] addrspace(3)* @y, i32 0, i32 0) to i8*)) #[[ATTR6]]
 ; CHECK-NEXT:    br label [[EXIT]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
 ; CHECK-NEXT:    ret void
 ;
 ;
 ; CHECK-LABEL: define {{[^@]+}}@baz_spmd() {
 ; CHECK-NEXT:    [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 true, i1 true)
 ; CHECK-NEXT:    call void @unknown_no_openmp()
 ; CHECK-NEXT:    [[C0:%.*]] = icmp eq i32 [[C]], -1
 ; CHECK-NEXT:    br i1 [[C0]], label [[MASTER3:%.*]], label [[EXIT:%.*]]
 ; CHECK:       master3:
-; CHECK-NEXT:    [[Z:%.*]] = call i8* @__kmpc_alloc_shared(i64 24) #[[ATTR4]], !dbg [[DBG9:![0-9]+]]
-; CHECK-NEXT:    call void @use.internalized(i8* nofree [[Z]]) #[[ATTR5]]
-; CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[Z]], i64 24) #[[ATTR4]]
+; CHECK-NEXT:    [[Z:%.*]] = call i8* @__kmpc_alloc_shared(i64 24) #[[ATTR5]], !dbg [[DBG9:![0-9]+]]
+; CHECK-NEXT:    call void @use.internalized(i8* nofree [[Z]]) #[[ATTR6]]
+; CHECK-NEXT:    call void @__kmpc_free_shared(i8* [[Z]], i64 24) #[[ATTR5]]
 ; CHECK-NEXT:    br label [[EXIT]]
 ; CHECK:       exit:
 ; CHECK-NEXT:    call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
 ; CHECK-NEXT:    ret void
 ;
 ;
 ; CHECK: Function Attrs: nofree nosync nounwind willreturn writeonly
 ; CHECK-LABEL: define {{[^@]+}}@use.internalized
 ; CHECK-SAME: (i8* nofree writeonly [[X:%.*]]) #[[ATTR0:[0-9]+]] {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    store i8* [[X]], i8** @S, align 8
 ; CHECK-NEXT:    ret void
 ;
 ;
 ; CHECK-LABEL: define {{[^@]+}}@use
 ; CHECK-SAME: (i8* [[X:%.*]]) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    store i8* [[X]], i8** @S, align 8
 ; CHECK-NEXT:    ret void
 ;
 ;.
 ; CHECK: attributes #[[ATTR0]] = { nofree nosync nounwind willreturn writeonly }
 ; CHECK: attributes #[[ATTR1:[0-9]+]] = { nosync nounwind }
-; CHECK: attributes #[[ATTR2:[0-9]+]] = { nounwind readnone }
-; CHECK: attributes #[[ATTR3:[0-9]+]] = { "llvm.assume"="omp_no_openmp" }
-; CHECK: attributes #[[ATTR4]] = { nounwind }
-; CHECK: attributes #[[ATTR5]] = { nounwind writeonly }
+; CHECK: attributes #[[ATTR2:[0-9]+]] = { nounwind readnone speculatable }
+; CHECK: attributes #[[ATTR3:[0-9]+]] = { nofree nosync nounwind readnone speculatable willreturn }
+; CHECK: attributes #[[ATTR4:[0-9]+]] = { "llvm.assume"="omp_no_openmp" }
+; CHECK: attributes #[[ATTR5]] = { nounwind }
+; CHECK: attributes #[[ATTR6]] = { nounwind writeonly }
 ;.
 ; CHECK: [[META0:![0-9]+]] = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None)
 ; CHECK: [[META1:![0-9]+]] = !DIFile(filename: "replace_globalization.c", directory: "/tmp/replace_globalization.c")
 ; CHECK: [[META2:![0-9]+]] = !{}
 ; CHECK: [[META3:![0-9]+]] = !{i32 2, !"Debug Info Version", i32 3}
 ; CHECK: [[META4:![0-9]+]] = !{i32 1, !"wchar_size", i32 4}
 ; CHECK: [[META5:![0-9]+]] = !{i32 7, !"openmp", i32 50}
 ; CHECK: [[META6:![0-9]+]] = !{i32 7, !"openmp-device", i32 50}
 ; CHECK: [[META7:![0-9]+]] = !{void ()* @foo, !"kernel", i32 1}
 ; CHECK: [[META8:![0-9]+]] = !{void ()* @bar, !"kernel", i32 1}
 ; CHECK: [[DBG9]] = !DILocation(line: 5, column: 14, scope: !10)
 ; CHECK: [[META10:![0-9]+]] = distinct !DISubprogram(name: "bar", scope: !1, file: !1, line: 1, type: !11, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2)
 ; CHECK: [[META11:![0-9]+]] = !DISubroutineType(types: !2)
 ;.