[RISCV] Expand scalable-vector truncstores and extloads

[frasercrmck]

Summary

Caught in internal testing, these operations are assumed legal by
default, even for scalable vector types. Expand them back into separate
truncations and stores, or loads and extensions.

Also add explicit fixed-length vector tests for these operations, even
though they should have been correct already.

[frasercrmck]

• rebase to hopefully fix the buildbots

[craig.topper]

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

508

You might be better off using integer_scalable_vector_valuetypes(). We need to block mask vectors too.

566

LoadExt for integer fixed vectors is missing.

578

Is the explicit makeArrayRef needed?

[frasercrmck]

• expand operations also for bool and fixed-length types
• remove makearrayref

[frasercrmck]

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

508

Good catch, thanks.

566

Done.

578

Turns out it's not, thanks.

[craig.topper]

LGTM

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

for (MVT VT : BoolVecVTs) {
       setOperationAction(ISD::SPLAT_VECTOR, VT, Legal);
 
       // Mask VTs are custom-expanded into a series of standard nodes
       setOperationAction(ISD::TRUNCATE, VT, Custom);
       setOperationAction(ISD::INSERT_SUBVECTOR, VT, Custom);
       setOperationAction(ISD::EXTRACT_SUBVECTOR, VT, Custom);
 
       setOperationAction(ISD::EXTRACT_VECTOR_ELT, VT, Custom);
+
+      // Expand all extending loads to types larger than this, and truncating
+      // stores from types larger than this.
+      for (MVT OtherVT : MVT::integer_scalable_vector_valuetypes()) {
+        setTruncStoreAction(OtherVT, VT, Expand);
+        setLoadExtAction(ISD::EXTLOAD, OtherVT, VT, Expand);
+        setLoadExtAction(ISD::SEXTLOAD, OtherVT, VT, Expand);
+        setLoadExtAction(ISD::ZEXTLOAD, OtherVT, VT, Expand);
+      }
     }
 
     for (MVT VT : IntVecVTs) {
       setOperationAction(ISD::SPLAT_VECTOR, VT, Legal);
       setOperationAction(ISD::SPLAT_VECTOR_PARTS, VT, Custom);
 
       setOperationAction(ISD::SMIN, VT, Legal);
       setOperationAction(ISD::SMAX, VT, Legal);
for (MVT VT : IntVecVTs) {
       setOperationAction(ISD::MSCATTER, VT, Custom);
 
       setOperationAction(ISD::CONCAT_VECTORS, VT, Custom);
       setOperationAction(ISD::INSERT_SUBVECTOR, VT, Custom);
       setOperationAction(ISD::EXTRACT_SUBVECTOR, VT, Custom);
 
       setOperationAction(ISD::STEP_VECTOR, VT, Custom);
       setOperationAction(ISD::VECTOR_REVERSE, VT, Custom);
+
+      for (MVT OtherVT : MVT::integer_scalable_vector_valuetypes()) {
+        setTruncStoreAction(VT, OtherVT, Expand);
+        setLoadExtAction(ISD::EXTLOAD, OtherVT, VT, Expand);
+        setLoadExtAction(ISD::SEXTLOAD, OtherVT, VT, Expand);
+        setLoadExtAction(ISD::ZEXTLOAD, OtherVT, VT, Expand);
+      }
     }
 
     // Expand various CCs to best match the RVV ISA, which natively supports UNE
     // but no other unordered comparisons, and supports all ordered comparisons
     // except ONE. Additionally, we expand GT,OGT,GE,OGE for optimization
     // purposes; they are expanded to their swapped-operand CCs (LT,OLT,LE,OLE),
     // and we pattern-match those back to the "original", swapping operands once
     // more. This way we catch both operations and both "vf" and "fv" forms with
const auto SetCommonVFPActions = [&](MVT VT) {
 
       setOperationAction(ISD::CONCAT_VECTORS, VT, Custom);
       setOperationAction(ISD::INSERT_SUBVECTOR, VT, Custom);
       setOperationAction(ISD::EXTRACT_SUBVECTOR, VT, Custom);
 
       setOperationAction(ISD::VECTOR_REVERSE, VT, Custom);
     };
 
+    // Sets common extload/truncstore actions on RVV floating-point vector
+    // types.
+    const auto SetCommonVFPExtLoadTruncStoreActions =
+        [&](MVT VT, ArrayRef<MVT::SimpleValueType> SmallerVTs) {
+          for (auto SmallVT : SmallerVTs) {
+            setTruncStoreAction(VT, SmallVT, Expand);
+            setLoadExtAction(ISD::EXTLOAD, VT, SmallVT, Expand);
+          }
+        };
+
     if (Subtarget.hasStdExtZfh())
       for (MVT VT : F16VecVTs)
         SetCommonVFPActions(VT);
 
+    for (MVT VT : F32VecVTs) {
       if (Subtarget.hasStdExtF())
-      for (MVT VT : F32VecVTs)
         SetCommonVFPActions(VT);
+      SetCommonVFPExtLoadTruncStoreActions(VT, F16VecVTs);
+    }
 
+    for (MVT VT : F64VecVTs) {
       if (Subtarget.hasStdExtD())
-      for (MVT VT : F64VecVTs)
         SetCommonVFPActions(VT);
+      SetCommonVFPExtLoadTruncStoreActions(VT, F16VecVTs);
+      SetCommonVFPExtLoadTruncStoreActions(VT, F32VecVTs);
+    }
 
     if (Subtarget.useRVVForFixedLengthVectors()) {
       for (MVT VT : MVT::integer_fixedlen_vector_valuetypes()) {
         if (!useRVVForFixedLengthVectorVT(VT))
           continue;
 
         // By default everything must be expanded.
         for (unsigned Op = 0; Op < ISD::BUILTIN_OP_END; ++Op)
           setOperationAction(Op, VT, Expand);
-        for (MVT OtherVT : MVT::fixedlen_vector_valuetypes())
+        for (MVT OtherVT : MVT::integer_fixedlen_vector_valuetypes()) {
           setTruncStoreAction(VT, OtherVT, Expand);
+          setLoadExtAction(ISD::EXTLOAD, OtherVT, VT, Expand);
+          setLoadExtAction(ISD::SEXTLOAD, OtherVT, VT, Expand);
+          setLoadExtAction(ISD::ZEXTLOAD, OtherVT, VT, Expand);
+        }
 
         // We use EXTRACT_SUBVECTOR as a "cast" from scalable to fixed.
         setOperationAction(ISD::INSERT_SUBVECTOR, VT, Custom);
         setOperationAction(ISD::EXTRACT_SUBVECTOR, VT, Custom);
 
         setOperationAction(ISD::BUILD_VECTOR, VT, Custom);
         setOperationAction(ISD::CONCAT_VECTORS, VT, Custom);
 

llvm/test/CodeGen/RISCV/rvv/extload-truncstore.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -verify-machineinstrs < %s | FileCheck %s
+
+define <vscale x 1 x i8> @sextload_nxv1i1_nxv1i8(<vscale x 1 x i1>* %x) {
+; CHECK-LABEL: sextload_nxv1i1_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vle1.v v0, (a0)
+; CHECK-NEXT:    vmv.v.i v25, 0
+; CHECK-NEXT:    vmerge.vim v8, v25, -1, v0
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i1>, <vscale x 1 x i1>* %x
+  %z = sext <vscale x 1 x i1> %y to <vscale x 1 x i8>
+  ret <vscale x 1 x i8> %z
+}
+
+define <vscale x 1 x i16> @sextload_nxv1i8_nxv1i16(<vscale x 1 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv1i8_nxv1i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i8>, <vscale x 1 x i8>* %x
+  %z = sext <vscale x 1 x i8> %y to <vscale x 1 x i16>
+  ret <vscale x 1 x i16> %z
+}
+
+define <vscale x 1 x i16> @zextload_nxv1i8_nxv1i16(<vscale x 1 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv1i8_nxv1i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i8>, <vscale x 1 x i8>* %x
+  %z = zext <vscale x 1 x i8> %y to <vscale x 1 x i16>
+  ret <vscale x 1 x i16> %z
+}
+
+define <vscale x 1 x i32> @sextload_nxv1i8_nxv1i32(<vscale x 1 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv1i8_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i8>, <vscale x 1 x i8>* %x
+  %z = sext <vscale x 1 x i8> %y to <vscale x 1 x i32>
+  ret <vscale x 1 x i32> %z
+}
+
+define <vscale x 1 x i32> @zextload_nxv1i8_nxv1i32(<vscale x 1 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv1i8_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i8>, <vscale x 1 x i8>* %x
+  %z = zext <vscale x 1 x i8> %y to <vscale x 1 x i32>
+  ret <vscale x 1 x i32> %z
+}
+
+define <vscale x 1 x i64> @sextload_nxv1i8_nxv1i64(<vscale x 1 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv1i8_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vsext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i8>, <vscale x 1 x i8>* %x
+  %z = sext <vscale x 1 x i8> %y to <vscale x 1 x i64>
+  ret <vscale x 1 x i64> %z
+}
+
+define <vscale x 1 x i64> @zextload_nxv1i8_nxv1i64(<vscale x 1 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv1i8_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vzext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i8>, <vscale x 1 x i8>* %x
+  %z = zext <vscale x 1 x i8> %y to <vscale x 1 x i64>
+  ret <vscale x 1 x i64> %z
+}
+
+define <vscale x 2 x i16> @sextload_nxv2i8_nxv2i16(<vscale x 2 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv2i8_nxv2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i8>, <vscale x 2 x i8>* %x
+  %z = sext <vscale x 2 x i8> %y to <vscale x 2 x i16>
+  ret <vscale x 2 x i16> %z
+}
+
+define <vscale x 2 x i16> @zextload_nxv2i8_nxv2i16(<vscale x 2 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv2i8_nxv2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i8>, <vscale x 2 x i8>* %x
+  %z = zext <vscale x 2 x i8> %y to <vscale x 2 x i16>
+  ret <vscale x 2 x i16> %z
+}
+
+define <vscale x 2 x i32> @sextload_nxv2i8_nxv2i32(<vscale x 2 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv2i8_nxv2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i8>, <vscale x 2 x i8>* %x
+  %z = sext <vscale x 2 x i8> %y to <vscale x 2 x i32>
+  ret <vscale x 2 x i32> %z
+}
+
+define <vscale x 2 x i32> @zextload_nxv2i8_nxv2i32(<vscale x 2 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv2i8_nxv2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i8>, <vscale x 2 x i8>* %x
+  %z = zext <vscale x 2 x i8> %y to <vscale x 2 x i32>
+  ret <vscale x 2 x i32> %z
+}
+
+define <vscale x 2 x i64> @sextload_nxv2i8_nxv2i64(<vscale x 2 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv2i8_nxv2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m2,ta,mu
+; CHECK-NEXT:    vsext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i8>, <vscale x 2 x i8>* %x
+  %z = sext <vscale x 2 x i8> %y to <vscale x 2 x i64>
+  ret <vscale x 2 x i64> %z
+}
+
+define <vscale x 2 x i64> @zextload_nxv2i8_nxv2i64(<vscale x 2 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv2i8_nxv2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m2,ta,mu
+; CHECK-NEXT:    vzext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i8>, <vscale x 2 x i8>* %x
+  %z = zext <vscale x 2 x i8> %y to <vscale x 2 x i64>
+  ret <vscale x 2 x i64> %z
+}
+
+define <vscale x 4 x i16> @sextload_nxv4i8_nxv4i16(<vscale x 4 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv4i8_nxv4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i8>, <vscale x 4 x i8>* %x
+  %z = sext <vscale x 4 x i8> %y to <vscale x 4 x i16>
+  ret <vscale x 4 x i16> %z
+}
+
+define <vscale x 4 x i16> @zextload_nxv4i8_nxv4i16(<vscale x 4 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv4i8_nxv4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i8>, <vscale x 4 x i8>* %x
+  %z = zext <vscale x 4 x i8> %y to <vscale x 4 x i16>
+  ret <vscale x 4 x i16> %z
+}
+
+define <vscale x 4 x i32> @sextload_nxv4i8_nxv4i32(<vscale x 4 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv4i8_nxv4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i8>, <vscale x 4 x i8>* %x
+  %z = sext <vscale x 4 x i8> %y to <vscale x 4 x i32>
+  ret <vscale x 4 x i32> %z
+}
+
+define <vscale x 4 x i32> @zextload_nxv4i8_nxv4i32(<vscale x 4 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv4i8_nxv4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i8>, <vscale x 4 x i8>* %x
+  %z = zext <vscale x 4 x i8> %y to <vscale x 4 x i32>
+  ret <vscale x 4 x i32> %z
+}
+
+define <vscale x 4 x i64> @sextload_nxv4i8_nxv4i64(<vscale x 4 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv4i8_nxv4i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m4,ta,mu
+; CHECK-NEXT:    vsext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i8>, <vscale x 4 x i8>* %x
+  %z = sext <vscale x 4 x i8> %y to <vscale x 4 x i64>
+  ret <vscale x 4 x i64> %z
+}
+
+define <vscale x 4 x i64> @zextload_nxv4i8_nxv4i64(<vscale x 4 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv4i8_nxv4i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m4,ta,mu
+; CHECK-NEXT:    vzext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i8>, <vscale x 4 x i8>* %x
+  %z = zext <vscale x 4 x i8> %y to <vscale x 4 x i64>
+  ret <vscale x 4 x i64> %z
+}
+
+define <vscale x 8 x i16> @sextload_nxv8i8_nxv8i16(<vscale x 8 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv8i8_nxv8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1r.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i8>, <vscale x 8 x i8>* %x
+  %z = sext <vscale x 8 x i8> %y to <vscale x 8 x i16>
+  ret <vscale x 8 x i16> %z
+}
+
+define <vscale x 8 x i16> @zextload_nxv8i8_nxv8i16(<vscale x 8 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv8i8_nxv8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1r.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i8>, <vscale x 8 x i8>* %x
+  %z = zext <vscale x 8 x i8> %y to <vscale x 8 x i16>
+  ret <vscale x 8 x i16> %z
+}
+
+define <vscale x 8 x i32> @sextload_nxv8i8_nxv8i32(<vscale x 8 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv8i8_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1r.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i8>, <vscale x 8 x i8>* %x
+  %z = sext <vscale x 8 x i8> %y to <vscale x 8 x i32>
+  ret <vscale x 8 x i32> %z
+}
+
+define <vscale x 8 x i32> @zextload_nxv8i8_nxv8i32(<vscale x 8 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv8i8_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1r.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i8>, <vscale x 8 x i8>* %x
+  %z = zext <vscale x 8 x i8> %y to <vscale x 8 x i32>
+  ret <vscale x 8 x i32> %z
+}
+
+define <vscale x 8 x i64> @sextload_nxv8i8_nxv8i64(<vscale x 8 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv8i8_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1r.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m8,ta,mu
+; CHECK-NEXT:    vsext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i8>, <vscale x 8 x i8>* %x
+  %z = sext <vscale x 8 x i8> %y to <vscale x 8 x i64>
+  ret <vscale x 8 x i64> %z
+}
+
+define <vscale x 8 x i64> @zextload_nxv8i8_nxv8i64(<vscale x 8 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv8i8_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1r.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m8,ta,mu
+; CHECK-NEXT:    vzext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i8>, <vscale x 8 x i8>* %x
+  %z = zext <vscale x 8 x i8> %y to <vscale x 8 x i64>
+  ret <vscale x 8 x i64> %z
+}
+
+define <vscale x 16 x i16> @sextload_nxv16i8_nxv16i16(<vscale x 16 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv16i8_nxv16i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2r.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x i8>, <vscale x 16 x i8>* %x
+  %z = sext <vscale x 16 x i8> %y to <vscale x 16 x i16>
+  ret <vscale x 16 x i16> %z
+}
+
+define <vscale x 16 x i16> @zextload_nxv16i8_nxv16i16(<vscale x 16 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv16i8_nxv16i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2r.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x i8>, <vscale x 16 x i8>* %x
+  %z = zext <vscale x 16 x i8> %y to <vscale x 16 x i16>
+  ret <vscale x 16 x i16> %z
+}
+
+define <vscale x 16 x i32> @sextload_nxv16i8_nxv16i32(<vscale x 16 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv16i8_nxv16i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2r.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m8,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x i8>, <vscale x 16 x i8>* %x
+  %z = sext <vscale x 16 x i8> %y to <vscale x 16 x i32>
+  ret <vscale x 16 x i32> %z
+}
+
+define <vscale x 16 x i32> @zextload_nxv16i8_nxv16i32(<vscale x 16 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv16i8_nxv16i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2r.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m8,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x i8>, <vscale x 16 x i8>* %x
+  %z = zext <vscale x 16 x i8> %y to <vscale x 16 x i32>
+  ret <vscale x 16 x i32> %z
+}
+
+define <vscale x 32 x i16> @sextload_nxv32i8_nxv32i16(<vscale x 32 x i8>* %x) {
+; CHECK-LABEL: sextload_nxv32i8_nxv32i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4r.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m8,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 32 x i8>, <vscale x 32 x i8>* %x
+  %z = sext <vscale x 32 x i8> %y to <vscale x 32 x i16>
+  ret <vscale x 32 x i16> %z
+}
+
+define <vscale x 32 x i16> @zextload_nxv32i8_nxv32i16(<vscale x 32 x i8>* %x) {
+; CHECK-LABEL: zextload_nxv32i8_nxv32i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4r.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m8,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 32 x i8>, <vscale x 32 x i8>* %x
+  %z = zext <vscale x 32 x i8> %y to <vscale x 32 x i16>
+  ret <vscale x 32 x i16> %z
+}
+
+define void @truncstore_nxv1i8_nxv1i1(<vscale x 1 x i8> %x, <vscale x 1 x i1> *%z) {
+; CHECK-LABEL: truncstore_nxv1i8_nxv1i1:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vand.vi v25, v8, 1
+; CHECK-NEXT:    vmsne.vi v26, v25, 0
+; CHECK-NEXT:    vse1.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 1 x i8> %x to <vscale x 1 x i1>
+  store <vscale x 1 x i1> %y, <vscale x 1 x i1>* %z
+  ret void
+}
+
+define void @truncstore_nxv1i16_nxv1i8(<vscale x 1 x i16> %x, <vscale x 1 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv1i16_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 1 x i16> %x to <vscale x 1 x i8>
+  store <vscale x 1 x i8> %y, <vscale x 1 x i8>* %z
+  ret void
+}
+
+define <vscale x 1 x i32> @sextload_nxv1i16_nxv1i32(<vscale x 1 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv1i16_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i16>, <vscale x 1 x i16>* %x
+  %z = sext <vscale x 1 x i16> %y to <vscale x 1 x i32>
+  ret <vscale x 1 x i32> %z
+}
+
+define <vscale x 1 x i32> @zextload_nxv1i16_nxv1i32(<vscale x 1 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv1i16_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i16>, <vscale x 1 x i16>* %x
+  %z = zext <vscale x 1 x i16> %y to <vscale x 1 x i32>
+  ret <vscale x 1 x i32> %z
+}
+
+define <vscale x 1 x i64> @sextload_nxv1i16_nxv1i64(<vscale x 1 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv1i16_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i16>, <vscale x 1 x i16>* %x
+  %z = sext <vscale x 1 x i16> %y to <vscale x 1 x i64>
+  ret <vscale x 1 x i64> %z
+}
+
+define <vscale x 1 x i64> @zextload_nxv1i16_nxv1i64(<vscale x 1 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv1i16_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i16>, <vscale x 1 x i16>* %x
+  %z = zext <vscale x 1 x i16> %y to <vscale x 1 x i64>
+  ret <vscale x 1 x i64> %z
+}
+
+define void @truncstore_nxv2i16_nxv2i8(<vscale x 2 x i16> %x, <vscale x 2 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv2i16_nxv2i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 2 x i16> %x to <vscale x 2 x i8>
+  store <vscale x 2 x i8> %y, <vscale x 2 x i8>* %z
+  ret void
+}
+
+define <vscale x 2 x i32> @sextload_nxv2i16_nxv2i32(<vscale x 2 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv2i16_nxv2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i16>, <vscale x 2 x i16>* %x
+  %z = sext <vscale x 2 x i16> %y to <vscale x 2 x i32>
+  ret <vscale x 2 x i32> %z
+}
+
+define <vscale x 2 x i32> @zextload_nxv2i16_nxv2i32(<vscale x 2 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv2i16_nxv2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i16>, <vscale x 2 x i16>* %x
+  %z = zext <vscale x 2 x i16> %y to <vscale x 2 x i32>
+  ret <vscale x 2 x i32> %z
+}
+
+define <vscale x 2 x i64> @sextload_nxv2i16_nxv2i64(<vscale x 2 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv2i16_nxv2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m2,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i16>, <vscale x 2 x i16>* %x
+  %z = sext <vscale x 2 x i16> %y to <vscale x 2 x i64>
+  ret <vscale x 2 x i64> %z
+}
+
+define <vscale x 2 x i64> @zextload_nxv2i16_nxv2i64(<vscale x 2 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv2i16_nxv2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m2,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i16>, <vscale x 2 x i16>* %x
+  %z = zext <vscale x 2 x i16> %y to <vscale x 2 x i64>
+  ret <vscale x 2 x i64> %z
+}
+
+define void @truncstore_nxv4i16_nxv4i8(<vscale x 4 x i16> %x, <vscale x 4 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv4i16_nxv4i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 4 x i16> %x to <vscale x 4 x i8>
+  store <vscale x 4 x i8> %y, <vscale x 4 x i8>* %z
+  ret void
+}
+
+define <vscale x 4 x i32> @sextload_nxv4i16_nxv4i32(<vscale x 4 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv4i16_nxv4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i16>, <vscale x 4 x i16>* %x
+  %z = sext <vscale x 4 x i16> %y to <vscale x 4 x i32>
+  ret <vscale x 4 x i32> %z
+}
+
+define <vscale x 4 x i32> @zextload_nxv4i16_nxv4i32(<vscale x 4 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv4i16_nxv4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i16>, <vscale x 4 x i16>* %x
+  %z = zext <vscale x 4 x i16> %y to <vscale x 4 x i32>
+  ret <vscale x 4 x i32> %z
+}
+
+define <vscale x 4 x i64> @sextload_nxv4i16_nxv4i64(<vscale x 4 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv4i16_nxv4i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m4,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i16>, <vscale x 4 x i16>* %x
+  %z = sext <vscale x 4 x i16> %y to <vscale x 4 x i64>
+  ret <vscale x 4 x i64> %z
+}
+
+define <vscale x 4 x i64> @zextload_nxv4i16_nxv4i64(<vscale x 4 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv4i16_nxv4i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m4,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i16>, <vscale x 4 x i16>* %x
+  %z = zext <vscale x 4 x i16> %y to <vscale x 4 x i64>
+  ret <vscale x 4 x i64> %z
+}
+
+define void @truncstore_nxv8i16_nxv8i8(<vscale x 8 x i16> %x, <vscale x 8 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv8i16_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 8 x i16> %x to <vscale x 8 x i8>
+  store <vscale x 8 x i8> %y, <vscale x 8 x i8>* %z
+  ret void
+}
+
+define <vscale x 8 x i32> @sextload_nxv8i16_nxv8i32(<vscale x 8 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv8i16_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i16>, <vscale x 8 x i16>* %x
+  %z = sext <vscale x 8 x i16> %y to <vscale x 8 x i32>
+  ret <vscale x 8 x i32> %z
+}
+
+define <vscale x 8 x i32> @zextload_nxv8i16_nxv8i32(<vscale x 8 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv8i16_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i16>, <vscale x 8 x i16>* %x
+  %z = zext <vscale x 8 x i16> %y to <vscale x 8 x i32>
+  ret <vscale x 8 x i32> %z
+}
+
+define <vscale x 8 x i64> @sextload_nxv8i16_nxv8i64(<vscale x 8 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv8i16_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m8,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i16>, <vscale x 8 x i16>* %x
+  %z = sext <vscale x 8 x i16> %y to <vscale x 8 x i64>
+  ret <vscale x 8 x i64> %z
+}
+
+define <vscale x 8 x i64> @zextload_nxv8i16_nxv8i64(<vscale x 8 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv8i16_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m8,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i16>, <vscale x 8 x i16>* %x
+  %z = zext <vscale x 8 x i16> %y to <vscale x 8 x i64>
+  ret <vscale x 8 x i64> %z
+}
+
+define void @truncstore_nxv16i16_nxv16i8(<vscale x 16 x i16> %x, <vscale x 16 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv16i16_nxv16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v8, 0
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 16 x i16> %x to <vscale x 16 x i8>
+  store <vscale x 16 x i8> %y, <vscale x 16 x i8>* %z
+  ret void
+}
+
+define <vscale x 16 x i32> @sextload_nxv16i16_nxv16i32(<vscale x 16 x i16>* %x) {
+; CHECK-LABEL: sextload_nxv16i16_nxv16i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re16.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m8,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x i16>, <vscale x 16 x i16>* %x
+  %z = sext <vscale x 16 x i16> %y to <vscale x 16 x i32>
+  ret <vscale x 16 x i32> %z
+}
+
+define <vscale x 16 x i32> @zextload_nxv16i16_nxv16i32(<vscale x 16 x i16>* %x) {
+; CHECK-LABEL: zextload_nxv16i16_nxv16i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re16.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m8,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x i16>, <vscale x 16 x i16>* %x
+  %z = zext <vscale x 16 x i16> %y to <vscale x 16 x i32>
+  ret <vscale x 16 x i32> %z
+}
+
+define void @truncstore_nxv32i16_nxv32i8(<vscale x 32 x i16> %x, <vscale x 32 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv32i16_nxv32i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8,m4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v28, v8, 0
+; CHECK-NEXT:    vs4r.v v28, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 32 x i16> %x to <vscale x 32 x i8>
+  store <vscale x 32 x i8> %y, <vscale x 32 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv1i32_nxv1i8(<vscale x 1 x i32> %x, <vscale x 1 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv1i32_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vse8.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 1 x i32> %x to <vscale x 1 x i8>
+  store <vscale x 1 x i8> %y, <vscale x 1 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv1i32_nxv1i16(<vscale x 1 x i32> %x, <vscale x 1 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv1i32_nxv1i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 1 x i32> %x to <vscale x 1 x i16>
+  store <vscale x 1 x i16> %y, <vscale x 1 x i16>* %z
+  ret void
+}
+
+define <vscale x 1 x i64> @sextload_nxv1i32_nxv1i64(<vscale x 1 x i32>* %x) {
+; CHECK-LABEL: sextload_nxv1i32_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i32>, <vscale x 1 x i32>* %x
+  %z = sext <vscale x 1 x i32> %y to <vscale x 1 x i64>
+  ret <vscale x 1 x i64> %z
+}
+
+define <vscale x 1 x i64> @zextload_nxv1i32_nxv1i64(<vscale x 1 x i32>* %x) {
+; CHECK-LABEL: zextload_nxv1i32_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x i32>, <vscale x 1 x i32>* %x
+  %z = zext <vscale x 1 x i32> %y to <vscale x 1 x i64>
+  ret <vscale x 1 x i64> %z
+}
+
+define void @truncstore_nxv2i32_nxv2i8(<vscale x 2 x i32> %x, <vscale x 2 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv2i32_nxv2i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vse8.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 2 x i32> %x to <vscale x 2 x i8>
+  store <vscale x 2 x i8> %y, <vscale x 2 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv2i32_nxv2i16(<vscale x 2 x i32> %x, <vscale x 2 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv2i32_nxv2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 2 x i32> %x to <vscale x 2 x i16>
+  store <vscale x 2 x i16> %y, <vscale x 2 x i16>* %z
+  ret void
+}
+
+define <vscale x 2 x i64> @sextload_nxv2i32_nxv2i64(<vscale x 2 x i32>* %x) {
+; CHECK-LABEL: sextload_nxv2i32_nxv2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re32.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m2,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i32>, <vscale x 2 x i32>* %x
+  %z = sext <vscale x 2 x i32> %y to <vscale x 2 x i64>
+  ret <vscale x 2 x i64> %z
+}
+
+define <vscale x 2 x i64> @zextload_nxv2i32_nxv2i64(<vscale x 2 x i32>* %x) {
+; CHECK-LABEL: zextload_nxv2i32_nxv2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re32.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m2,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x i32>, <vscale x 2 x i32>* %x
+  %z = zext <vscale x 2 x i32> %y to <vscale x 2 x i64>
+  ret <vscale x 2 x i64> %z
+}
+
+define void @truncstore_nxv4i32_nxv4i8(<vscale x 4 x i32> %x, <vscale x 4 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv4i32_nxv4i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vse8.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 4 x i32> %x to <vscale x 4 x i8>
+  store <vscale x 4 x i8> %y, <vscale x 4 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv4i32_nxv4i16(<vscale x 4 x i32> %x, <vscale x 4 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv4i32_nxv4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 4 x i32> %x to <vscale x 4 x i16>
+  store <vscale x 4 x i16> %y, <vscale x 4 x i16>* %z
+  ret void
+}
+
+define <vscale x 4 x i64> @sextload_nxv4i32_nxv4i64(<vscale x 4 x i32>* %x) {
+; CHECK-LABEL: sextload_nxv4i32_nxv4i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re32.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m4,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i32>, <vscale x 4 x i32>* %x
+  %z = sext <vscale x 4 x i32> %y to <vscale x 4 x i64>
+  ret <vscale x 4 x i64> %z
+}
+
+define <vscale x 4 x i64> @zextload_nxv4i32_nxv4i64(<vscale x 4 x i32>* %x) {
+; CHECK-LABEL: zextload_nxv4i32_nxv4i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re32.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m4,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x i32>, <vscale x 4 x i32>* %x
+  %z = zext <vscale x 4 x i32> %y to <vscale x 4 x i64>
+  ret <vscale x 4 x i64> %z
+}
+
+define void @truncstore_nxv8i32_nxv8i8(<vscale x 8 x i32> %x, <vscale x 8 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv8i32_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v26, 0
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 8 x i32> %x to <vscale x 8 x i8>
+  store <vscale x 8 x i8> %y, <vscale x 8 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv8i32_nxv8i16(<vscale x 8 x i32> %x, <vscale x 8 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv8i32_nxv8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v8, 0
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 8 x i32> %x to <vscale x 8 x i16>
+  store <vscale x 8 x i16> %y, <vscale x 8 x i16>* %z
+  ret void
+}
+
+define <vscale x 8 x i64> @sextload_nxv8i32_nxv8i64(<vscale x 8 x i32>* %x) {
+; CHECK-LABEL: sextload_nxv8i32_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re32.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m8,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i32>, <vscale x 8 x i32>* %x
+  %z = sext <vscale x 8 x i32> %y to <vscale x 8 x i64>
+  ret <vscale x 8 x i64> %z
+}
+
+define <vscale x 8 x i64> @zextload_nxv8i32_nxv8i64(<vscale x 8 x i32>* %x) {
+; CHECK-LABEL: zextload_nxv8i32_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re32.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64,m8,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x i32>, <vscale x 8 x i32>* %x
+  %z = zext <vscale x 8 x i32> %y to <vscale x 8 x i64>
+  ret <vscale x 8 x i64> %z
+}
+
+define void @truncstore_nxv16i32_nxv16i8(<vscale x 16 x i32> %x, <vscale x 16 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv16i32_nxv16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v28, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v28, 0
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 16 x i32> %x to <vscale x 16 x i8>
+  store <vscale x 16 x i8> %y, <vscale x 16 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv16i32_nxv16i16(<vscale x 16 x i32> %x, <vscale x 16 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv16i32_nxv16i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v28, v8, 0
+; CHECK-NEXT:    vs4r.v v28, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 16 x i32> %x to <vscale x 16 x i16>
+  store <vscale x 16 x i16> %y, <vscale x 16 x i16>* %z
+  ret void
+}
+
+define void @truncstore_nxv1i64_nxv1i8(<vscale x 1 x i64> %x, <vscale x 1 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv1i64_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v26, 0
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 1 x i64> %x to <vscale x 1 x i8>
+  store <vscale x 1 x i8> %y, <vscale x 1 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv1i64_nxv1i16(<vscale x 1 x i64> %x, <vscale x 1 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv1i64_nxv1i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vse16.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 1 x i64> %x to <vscale x 1 x i16>
+  store <vscale x 1 x i16> %y, <vscale x 1 x i16>* %z
+  ret void
+}
+
+define void @truncstore_nxv1i64_nxv1i32(<vscale x 1 x i64> %x, <vscale x 1 x i32>* %z) {
+; CHECK-LABEL: truncstore_nxv1i64_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 1 x i64> %x to <vscale x 1 x i32>
+  store <vscale x 1 x i32> %y, <vscale x 1 x i32>* %z
+  ret void
+}
+
+define void @truncstore_nxv2i64_nxv2i8(<vscale x 2 x i64> %x, <vscale x 2 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv2i64_nxv2i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v26, 0
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 2 x i64> %x to <vscale x 2 x i8>
+  store <vscale x 2 x i8> %y, <vscale x 2 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv2i64_nxv2i16(<vscale x 2 x i64> %x, <vscale x 2 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv2i64_nxv2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vse16.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 2 x i64> %x to <vscale x 2 x i16>
+  store <vscale x 2 x i16> %y, <vscale x 2 x i16>* %z
+  ret void
+}
+
+define void @truncstore_nxv2i64_nxv2i32(<vscale x 2 x i64> %x, <vscale x 2 x i32>* %z) {
+; CHECK-LABEL: truncstore_nxv2i64_nxv2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 2 x i64> %x to <vscale x 2 x i32>
+  store <vscale x 2 x i32> %y, <vscale x 2 x i32>* %z
+  ret void
+}
+
+define void @truncstore_nxv4i64_nxv4i8(<vscale x 4 x i64> %x, <vscale x 4 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv4i64_nxv4i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v26, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vse8.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 4 x i64> %x to <vscale x 4 x i8>
+  store <vscale x 4 x i8> %y, <vscale x 4 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv4i64_nxv4i16(<vscale x 4 x i64> %x, <vscale x 4 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv4i64_nxv4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v26, 0
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 4 x i64> %x to <vscale x 4 x i16>
+  store <vscale x 4 x i16> %y, <vscale x 4 x i16>* %z
+  ret void
+}
+
+define void @truncstore_nxv4i64_nxv4i32(<vscale x 4 x i64> %x, <vscale x 4 x i32>* %z) {
+; CHECK-LABEL: truncstore_nxv4i64_nxv4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v8, 0
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 4 x i64> %x to <vscale x 4 x i32>
+  store <vscale x 4 x i32> %y, <vscale x 4 x i32>* %z
+  ret void
+}
+
+define void @truncstore_nxv8i64_nxv8i8(<vscale x 8 x i64> %x, <vscale x 8 x i8>* %z) {
+; CHECK-LABEL: truncstore_nxv8i64_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v28, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v28, 0
+; CHECK-NEXT:    vsetvli a1, zero, e8,m1,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v26, 0
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 8 x i64> %x to <vscale x 8 x i8>
+  store <vscale x 8 x i8> %y, <vscale x 8 x i8>* %z
+  ret void
+}
+
+define void @truncstore_nxv8i64_nxv8i16(<vscale x 8 x i64> %x, <vscale x 8 x i16>* %z) {
+; CHECK-LABEL: truncstore_nxv8i64_nxv8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v28, v8, 0
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v28, 0
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 8 x i64> %x to <vscale x 8 x i16>
+  store <vscale x 8 x i16> %y, <vscale x 8 x i16>* %z
+  ret void
+}
+
+define void @truncstore_nxv8i64_nxv8i32(<vscale x 8 x i64> %x, <vscale x 8 x i32>* %z) {
+; CHECK-LABEL: truncstore_nxv8i64_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v28, v8, 0
+; CHECK-NEXT:    vs4r.v v28, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <vscale x 8 x i64> %x to <vscale x 8 x i32>
+  store <vscale x 8 x i32> %y, <vscale x 8 x i32>* %z
+  ret void
+}
+
+define <vscale x 1 x float> @extload_nxv1f16_nxv1f32(<vscale x 1 x half>* %x) {
+; CHECK-LABEL: extload_nxv1f16_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x half>, <vscale x 1 x half>* %x
+  %z = fpext <vscale x 1 x half> %y to <vscale x 1 x float>
+  ret <vscale x 1 x float> %z
+}
+
+define <vscale x 1 x double> @extload_nxv1f16_nxv1f64(<vscale x 1 x half>* %x) {
+; CHECK-LABEL: extload_nxv1f16_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfwcvt.f.f.v v26, v25
+; CHECK-NEXT:    vsetvli a0, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x half>, <vscale x 1 x half>* %x
+  %z = fpext <vscale x 1 x half> %y to <vscale x 1 x double>
+  ret <vscale x 1 x double> %z
+}
+
+define <vscale x 2 x float> @extload_nxv2f16_nxv2f32(<vscale x 2 x half>* %x) {
+; CHECK-LABEL: extload_nxv2f16_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x half>, <vscale x 2 x half>* %x
+  %z = fpext <vscale x 2 x half> %y to <vscale x 2 x float>
+  ret <vscale x 2 x float> %z
+}
+
+define <vscale x 2 x double> @extload_nxv2f16_nxv2f64(<vscale x 2 x half>* %x) {
+; CHECK-LABEL: extload_nxv2f16_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfwcvt.f.f.v v26, v25
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x half>, <vscale x 2 x half>* %x
+  %z = fpext <vscale x 2 x half> %y to <vscale x 2 x double>
+  ret <vscale x 2 x double> %z
+}
+
+define <vscale x 4 x float> @extload_nxv4f16_nxv4f32(<vscale x 4 x half>* %x) {
+; CHECK-LABEL: extload_nxv4f16_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x half>, <vscale x 4 x half>* %x
+  %z = fpext <vscale x 4 x half> %y to <vscale x 4 x float>
+  ret <vscale x 4 x float> %z
+}
+
+define <vscale x 4 x double> @extload_nxv4f16_nxv4f64(<vscale x 4 x half>* %x) {
+; CHECK-LABEL: extload_nxv4f16_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v26, v25
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x half>, <vscale x 4 x half>* %x
+  %z = fpext <vscale x 4 x half> %y to <vscale x 4 x double>
+  ret <vscale x 4 x double> %z
+}
+
+define <vscale x 8 x float> @extload_nxv8f16_nxv8f32(<vscale x 8 x half>* %x) {
+; CHECK-LABEL: extload_nxv8f16_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x half>, <vscale x 8 x half>* %x
+  %z = fpext <vscale x 8 x half> %y to <vscale x 8 x float>
+  ret <vscale x 8 x float> %z
+}
+
+define <vscale x 8 x double> @extload_nxv8f16_nxv8f64(<vscale x 8 x half>* %x) {
+; CHECK-LABEL: extload_nxv8f16_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v28, v26
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x half>, <vscale x 8 x half>* %x
+  %z = fpext <vscale x 8 x half> %y to <vscale x 8 x double>
+  ret <vscale x 8 x double> %z
+}
+
+define <vscale x 16 x float> @extload_nxv16f16_nxv16f32(<vscale x 16 x half>* %x) {
+; CHECK-LABEL: extload_nxv16f16_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re16.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x half>, <vscale x 16 x half>* %x
+  %z = fpext <vscale x 16 x half> %y to <vscale x 16 x float>
+  ret <vscale x 16 x float> %z
+}
+
+define void @truncstore_nxv1f32_nxv1f16(<vscale x 1 x float> %x, <vscale x 1 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv1f32_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 1 x float> %x to <vscale x 1 x half>
+  store <vscale x 1 x half> %y, <vscale x 1 x half>* %z
+  ret void
+}
+
+define <vscale x 1 x double> @extload_nxv1f32_nxv1f64(<vscale x 1 x float>* %x) {
+; CHECK-LABEL: extload_nxv1f32_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 1 x float>, <vscale x 1 x float>* %x
+  %z = fpext <vscale x 1 x float> %y to <vscale x 1 x double>
+  ret <vscale x 1 x double> %z
+}
+
+define void @truncstore_nxv2f32_nxv2f16(<vscale x 2 x float> %x, <vscale x 2 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv2f32_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 2 x float> %x to <vscale x 2 x half>
+  store <vscale x 2 x half> %y, <vscale x 2 x half>* %z
+  ret void
+}
+
+define <vscale x 2 x double> @extload_nxv2f32_nxv2f64(<vscale x 2 x float>* %x) {
+; CHECK-LABEL: extload_nxv2f32_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re32.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x float>, <vscale x 2 x float>* %x
+  %z = fpext <vscale x 2 x float> %y to <vscale x 2 x double>
+  ret <vscale x 2 x double> %z
+}
+
+define void @truncstore_nxv4f32_nxv4f16(<vscale x 4 x float> %x, <vscale x 4 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv4f32_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 4 x float> %x to <vscale x 4 x half>
+  store <vscale x 4 x half> %y, <vscale x 4 x half>* %z
+  ret void
+}
+
+define <vscale x 4 x double> @extload_nxv4f32_nxv4f64(<vscale x 4 x float>* %x) {
+; CHECK-LABEL: extload_nxv4f32_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re32.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x float>, <vscale x 4 x float>* %x
+  %z = fpext <vscale x 4 x float> %y to <vscale x 4 x double>
+  ret <vscale x 4 x double> %z
+}
+
+define void @truncstore_nxv8f32_nxv8f16(<vscale x 8 x float> %x, <vscale x 8 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv8f32_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v8
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 8 x float> %x to <vscale x 8 x half>
+  store <vscale x 8 x half> %y, <vscale x 8 x half>* %z
+  ret void
+}
+
+define <vscale x 8 x double> @extload_nxv8f32_nxv8f64(<vscale x 8 x float>* %x) {
+; CHECK-LABEL: extload_nxv8f32_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re32.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x float>, <vscale x 8 x float>* %x
+  %z = fpext <vscale x 8 x float> %y to <vscale x 8 x double>
+  ret <vscale x 8 x double> %z
+}
+
+define void @truncstore_nxv16f32_nxv16f16(<vscale x 16 x float> %x, <vscale x 16 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv16f32_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v28, v8
+; CHECK-NEXT:    vs4r.v v28, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 16 x float> %x to <vscale x 16 x half>
+  store <vscale x 16 x half> %y, <vscale x 16 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv1f64_nxv1f16(<vscale x 1 x double> %x, <vscale x 1 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv1f64_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v25, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v25
+; CHECK-NEXT:    vse16.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 1 x double> %x to <vscale x 1 x half>
+  store <vscale x 1 x half> %y, <vscale x 1 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv1f64_nxv1f32(<vscale x 1 x double> %x, <vscale x 1 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv1f64_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 1 x double> %x to <vscale x 1 x float>
+  store <vscale x 1 x float> %y, <vscale x 1 x float>* %z
+  ret void
+}
+
+define void @truncstore_nxv2f64_nxv2f16(<vscale x 2 x double> %x, <vscale x 2 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv2f64_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v25, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v25
+; CHECK-NEXT:    vse16.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 2 x double> %x to <vscale x 2 x half>
+  store <vscale x 2 x half> %y, <vscale x 2 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv2f64_nxv2f32(<vscale x 2 x double> %x, <vscale x 2 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv2f64_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 2 x double> %x to <vscale x 2 x float>
+  store <vscale x 2 x float> %y, <vscale x 2 x float>* %z
+  ret void
+}
+
+define void @truncstore_nxv4f64_nxv4f16(<vscale x 4 x double> %x, <vscale x 4 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv4f64_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v26, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v26
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 4 x double> %x to <vscale x 4 x half>
+  store <vscale x 4 x half> %y, <vscale x 4 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv4f64_nxv4f32(<vscale x 4 x double> %x, <vscale x 4 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv4f64_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v8
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 4 x double> %x to <vscale x 4 x float>
+  store <vscale x 4 x float> %y, <vscale x 4 x float>* %z
+  ret void
+}
+
+define void @truncstore_nxv8f64_nxv8f16(<vscale x 8 x double> %x, <vscale x 8 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv8f64_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v28, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v28
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 8 x double> %x to <vscale x 8 x half>
+  store <vscale x 8 x half> %y, <vscale x 8 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv8f64_nxv8f32(<vscale x 8 x double> %x, <vscale x 8 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv8f64_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v28, v8
+; CHECK-NEXT:    vs4r.v v28, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 8 x double> %x to <vscale x 8 x float>
+  store <vscale x 8 x float> %y, <vscale x 8 x float>* %z
+  ret void
+}

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-extload-truncstore.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=4 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX4
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=4 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX4
+
+define <2 x i16> @sextload_v2i1_v2i16(<2 x i1>* %x) {
+; CHECK-LABEL: sextload_v2i1_v2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vle1.v v0, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vmv.v.i v25, 0
+; CHECK-NEXT:    vmerge.vim v8, v25, -1, v0
+; CHECK-NEXT:    ret
+  %y = load <2 x i1>, <2 x i1>* %x
+  %z = sext <2 x i1> %y to <2 x i16>
+  ret <2 x i16> %z
+}
+
+define <2 x i16> @sextload_v2i8_v2i16(<2 x i8>* %x) {
+; CHECK-LABEL: sextload_v2i8_v2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i8>, <2 x i8>* %x
+  %z = sext <2 x i8> %y to <2 x i16>
+  ret <2 x i16> %z
+}
+
+define <2 x i16> @zextload_v2i8_v2i16(<2 x i8>* %x) {
+; CHECK-LABEL: zextload_v2i8_v2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i8>, <2 x i8>* %x
+  %z = zext <2 x i8> %y to <2 x i16>
+  ret <2 x i16> %z
+}
+
+define <2 x i32> @sextload_v2i8_v2i32(<2 x i8>* %x) {
+; CHECK-LABEL: sextload_v2i8_v2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i8>, <2 x i8>* %x
+  %z = sext <2 x i8> %y to <2 x i32>
+  ret <2 x i32> %z
+}
+
+define <2 x i32> @zextload_v2i8_v2i32(<2 x i8>* %x) {
+; CHECK-LABEL: zextload_v2i8_v2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i8>, <2 x i8>* %x
+  %z = zext <2 x i8> %y to <2 x i32>
+  ret <2 x i32> %z
+}
+
+define <2 x i64> @sextload_v2i8_v2i64(<2 x i8>* %x) {
+; CHECK-LABEL: sextload_v2i8_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; CHECK-NEXT:    vsext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i8>, <2 x i8>* %x
+  %z = sext <2 x i8> %y to <2 x i64>
+  ret <2 x i64> %z
+}
+
+define <2 x i64> @zextload_v2i8_v2i64(<2 x i8>* %x) {
+; CHECK-LABEL: zextload_v2i8_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; CHECK-NEXT:    vzext.vf8 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i8>, <2 x i8>* %x
+  %z = zext <2 x i8> %y to <2 x i64>
+  ret <2 x i64> %z
+}
+
+define <4 x i16> @sextload_v4i8_v4i16(<4 x i8>* %x) {
+; CHECK-LABEL: sextload_v4i8_v4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <4 x i8>, <4 x i8>* %x
+  %z = sext <4 x i8> %y to <4 x i16>
+  ret <4 x i16> %z
+}
+
+define <4 x i16> @zextload_v4i8_v4i16(<4 x i8>* %x) {
+; CHECK-LABEL: zextload_v4i8_v4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <4 x i8>, <4 x i8>* %x
+  %z = zext <4 x i8> %y to <4 x i16>
+  ret <4 x i16> %z
+}
+
+define <4 x i32> @sextload_v4i8_v4i32(<4 x i8>* %x) {
+; CHECK-LABEL: sextload_v4i8_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <4 x i8>, <4 x i8>* %x
+  %z = sext <4 x i8> %y to <4 x i32>
+  ret <4 x i32> %z
+}
+
+define <4 x i32> @zextload_v4i8_v4i32(<4 x i8>* %x) {
+; CHECK-LABEL: zextload_v4i8_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <4 x i8>, <4 x i8>* %x
+  %z = zext <4 x i8> %y to <4 x i32>
+  ret <4 x i32> %z
+}
+
+define <4 x i64> @sextload_v4i8_v4i64(<4 x i8>* %x) {
+; LMULMAX1-LABEL: sextload_v4i8_v4i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v4i8_v4i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
+; LMULMAX4-NEXT:    vsext.vf8 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <4 x i8>, <4 x i8>* %x
+  %z = sext <4 x i8> %y to <4 x i64>
+  ret <4 x i64> %z
+}
+
+define <4 x i64> @zextload_v4i8_v4i64(<4 x i8>* %x) {
+; LMULMAX1-LABEL: zextload_v4i8_v4i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v4i8_v4i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
+; LMULMAX4-NEXT:    vzext.vf8 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <4 x i8>, <4 x i8>* %x
+  %z = zext <4 x i8> %y to <4 x i64>
+  ret <4 x i64> %z
+}
+
+define <8 x i16> @sextload_v8i8_v8i16(<8 x i8>* %x) {
+; CHECK-LABEL: sextload_v8i8_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <8 x i8>, <8 x i8>* %x
+  %z = sext <8 x i8> %y to <8 x i16>
+  ret <8 x i16> %z
+}
+
+define <8 x i16> @zextload_v8i8_v8i16(<8 x i8>* %x) {
+; CHECK-LABEL: zextload_v8i8_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <8 x i8>, <8 x i8>* %x
+  %z = zext <8 x i8> %y to <8 x i16>
+  ret <8 x i16> %z
+}
+
+define <8 x i32> @sextload_v8i8_v8i32(<8 x i8>* %x) {
+; LMULMAX1-LABEL: sextload_v8i8_v8i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v8i8_v8i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vsext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i8>, <8 x i8>* %x
+  %z = sext <8 x i8> %y to <8 x i32>
+  ret <8 x i32> %z
+}
+
+define <8 x i32> @zextload_v8i8_v8i32(<8 x i8>* %x) {
+; LMULMAX1-LABEL: zextload_v8i8_v8i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v8i8_v8i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vzext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i8>, <8 x i8>* %x
+  %z = zext <8 x i8> %y to <8 x i32>
+  ret <8 x i32> %z
+}
+
+define <8 x i64> @sextload_v8i8_v8i64(<8 x i8>* %x) {
+; LMULMAX1-LABEL: sextload_v8i8_v8i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v11, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v8i8_v8i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf8 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i8>, <8 x i8>* %x
+  %z = sext <8 x i8> %y to <8 x i64>
+  ret <8 x i64> %z
+}
+
+define <8 x i64> @zextload_v8i8_v8i64(<8 x i8>* %x) {
+; LMULMAX1-LABEL: zextload_v8i8_v8i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v11, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v8i8_v8i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf8 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i8>, <8 x i8>* %x
+  %z = zext <8 x i8> %y to <8 x i64>
+  ret <8 x i64> %z
+}
+
+define <16 x i16> @sextload_v16i8_v16i16(<16 x i8>* %x) {
+; LMULMAX1-LABEL: sextload_v16i8_v16i16:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 8
+; LMULMAX1-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v16i8_v16i16:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vsext.vf2 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i8>, <16 x i8>* %x
+  %z = sext <16 x i8> %y to <16 x i16>
+  ret <16 x i16> %z
+}
+
+define <16 x i16> @zextload_v16i8_v16i16(<16 x i8>* %x) {
+; LMULMAX1-LABEL: zextload_v16i8_v16i16:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 8
+; LMULMAX1-NEXT:    vsetivli a0, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v16i8_v16i16:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vzext.vf2 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i8>, <16 x i8>* %x
+  %z = zext <16 x i8> %y to <16 x i16>
+  ret <16 x i16> %z
+}
+
+define <16 x i32> @sextload_v16i8_v16i32(<16 x i8>* %x) {
+; LMULMAX1-LABEL: sextload_v16i8_v16i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 8
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v11, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v16i8_v16i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i8>, <16 x i8>* %x
+  %z = sext <16 x i8> %y to <16 x i32>
+  ret <16 x i32> %z
+}
+
+define <16 x i32> @zextload_v16i8_v16i32(<16 x i8>* %x) {
+; LMULMAX1-LABEL: zextload_v16i8_v16i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 8
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v11, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v16i8_v16i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i8>, <16 x i8>* %x
+  %z = zext <16 x i8> %y to <16 x i32>
+  ret <16 x i32> %z
+}
+
+define <16 x i64> @sextload_v16i8_v16i64(<16 x i8>* %x) {
+; LMULMAX1-LABEL: sextload_v16i8_v16i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 8
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v12, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v14, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v15, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v9, v27
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v10, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v11, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf8 v13, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v16i8_v16i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf8 v8, v25
+; LMULMAX4-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vslidedown.vi v25, v25, 8
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf8 v12, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i8>, <16 x i8>* %x
+  %z = sext <16 x i8> %y to <16 x i64>
+  ret <16 x i64> %z
+}
+
+define <16 x i64> @zextload_v16i8_v16i64(<16 x i8>* %x) {
+; LMULMAX1-LABEL: zextload_v16i8_v16i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vle8.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 8
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v12, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v14, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v15, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v9, v27
+; LMULMAX1-NEXT:    vsetivli a0, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v10, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v11, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf8 v13, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v16i8_v16i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vle8.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf8 v8, v25
+; LMULMAX4-NEXT:    vsetivli a0, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vslidedown.vi v25, v25, 8
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf8 v12, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i8>, <16 x i8>* %x
+  %z = zext <16 x i8> %y to <16 x i64>
+  ret <16 x i64> %z
+}
+
+define void @truncstore_v2i8_v2i1(<2 x i8> %x, <2 x i1>* %z) {
+; CHECK-LABEL: truncstore_v2i8_v2i1:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vand.vi v25, v8, 1
+; CHECK-NEXT:    vmsne.vi v26, v25, 0
+; CHECK-NEXT:    vse1.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <2 x i8> %x to <2 x i1>
+  store <2 x i1> %y, <2 x i1>* %z
+  ret void
+}
+
+define void @truncstore_v2i16_v2i8(<2 x i16> %x, <2 x i8>* %z) {
+; CHECK-LABEL: truncstore_v2i16_v2i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e8,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <2 x i16> %x to <2 x i8>
+  store <2 x i8> %y, <2 x i8>* %z
+  ret void
+}
+
+define <2 x i32> @sextload_v2i16_v2i32(<2 x i16>* %x) {
+; CHECK-LABEL: sextload_v2i16_v2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i16>, <2 x i16>* %x
+  %z = sext <2 x i16> %y to <2 x i32>
+  ret <2 x i32> %z
+}
+
+define <2 x i32> @zextload_v2i16_v2i32(<2 x i16>* %x) {
+; CHECK-LABEL: zextload_v2i16_v2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i16>, <2 x i16>* %x
+  %z = zext <2 x i16> %y to <2 x i32>
+  ret <2 x i32> %z
+}
+
+define <2 x i64> @sextload_v2i16_v2i64(<2 x i16>* %x) {
+; CHECK-LABEL: sextload_v2i16_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; CHECK-NEXT:    vsext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i16>, <2 x i16>* %x
+  %z = sext <2 x i16> %y to <2 x i64>
+  ret <2 x i64> %z
+}
+
+define <2 x i64> @zextload_v2i16_v2i64(<2 x i16>* %x) {
+; CHECK-LABEL: zextload_v2i16_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; CHECK-NEXT:    vzext.vf4 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i16>, <2 x i16>* %x
+  %z = zext <2 x i16> %y to <2 x i64>
+  ret <2 x i64> %z
+}
+
+define void @truncstore_v4i16_v4i8(<4 x i16> %x, <4 x i8>* %z) {
+; CHECK-LABEL: truncstore_v4i16_v4i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e8,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <4 x i16> %x to <4 x i8>
+  store <4 x i8> %y, <4 x i8>* %z
+  ret void
+}
+
+define <4 x i32> @sextload_v4i16_v4i32(<4 x i16>* %x) {
+; CHECK-LABEL: sextload_v4i16_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <4 x i16>, <4 x i16>* %x
+  %z = sext <4 x i16> %y to <4 x i32>
+  ret <4 x i32> %z
+}
+
+define <4 x i32> @zextload_v4i16_v4i32(<4 x i16>* %x) {
+; CHECK-LABEL: zextload_v4i16_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <4 x i16>, <4 x i16>* %x
+  %z = zext <4 x i16> %y to <4 x i32>
+  ret <4 x i32> %z
+}
+
+define <4 x i64> @sextload_v4i16_v4i64(<4 x i16>* %x) {
+; LMULMAX1-LABEL: sextload_v4i16_v4i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v4i16_v4i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vle16.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
+; LMULMAX4-NEXT:    vsext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <4 x i16>, <4 x i16>* %x
+  %z = sext <4 x i16> %y to <4 x i64>
+  ret <4 x i64> %z
+}
+
+define <4 x i64> @zextload_v4i16_v4i64(<4 x i16>* %x) {
+; LMULMAX1-LABEL: zextload_v4i16_v4i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v4i16_v4i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vle16.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
+; LMULMAX4-NEXT:    vzext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <4 x i16>, <4 x i16>* %x
+  %z = zext <4 x i16> %y to <4 x i64>
+  ret <4 x i64> %z
+}
+
+define void @truncstore_v8i16_v8i8(<8 x i16> %x, <8 x i8>* %z) {
+; CHECK-LABEL: truncstore_v8i16_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 8, e8,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <8 x i16> %x to <8 x i8>
+  store <8 x i8> %y, <8 x i8>* %z
+  ret void
+}
+
+define <8 x i32> @sextload_v8i16_v8i32(<8 x i16>* %x) {
+; LMULMAX1-LABEL: sextload_v8i16_v8i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v8i16_v8i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vle16.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vsext.vf2 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i16>, <8 x i16>* %x
+  %z = sext <8 x i16> %y to <8 x i32>
+  ret <8 x i32> %z
+}
+
+define <8 x i32> @zextload_v8i16_v8i32(<8 x i16>* %x) {
+; LMULMAX1-LABEL: zextload_v8i16_v8i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v8i16_v8i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vle16.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vzext.vf2 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i16>, <8 x i16>* %x
+  %z = zext <8 x i16> %y to <8 x i32>
+  ret <8 x i32> %z
+}
+
+define <8 x i64> @sextload_v8i16_v8i64(<8 x i16>* %x) {
+; LMULMAX1-LABEL: sextload_v8i16_v8i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v11, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v8i16_v8i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vle16.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i16>, <8 x i16>* %x
+  %z = sext <8 x i16> %y to <8 x i64>
+  ret <8 x i64> %z
+}
+
+define <8 x i64> @zextload_v8i16_v8i64(<8 x i16>* %x) {
+; LMULMAX1-LABEL: zextload_v8i16_v8i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v11, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v8i16_v8i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vle16.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf4 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i16>, <8 x i16>* %x
+  %z = zext <8 x i16> %y to <8 x i64>
+  ret <8 x i64> %z
+}
+
+define void @truncstore_v16i16_v16i8(<16 x i16> %x, <16 x i8>* %z) {
+; LMULMAX1-LABEL: truncstore_v16i16_v16i8:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 8
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vse8.v v26, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v16i16_v16i8:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX4-NEXT:    vse8.v v25, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <16 x i16> %x to <16 x i8>
+  store <16 x i8> %y, <16 x i8>* %z
+  ret void
+}
+
+define <16 x i32> @sextload_v16i16_v16i32(<16 x i16>* %x) {
+; LMULMAX1-LABEL: sextload_v16i16_v16i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle16.v v26, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v9, v25
+; LMULMAX1-NEXT:    vsext.vf2 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v11, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v16i16_v16i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vle16.v v26, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf2 v8, v26
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i16>, <16 x i16>* %x
+  %z = sext <16 x i16> %y to <16 x i32>
+  ret <16 x i32> %z
+}
+
+define <16 x i32> @zextload_v16i16_v16i32(<16 x i16>* %x) {
+; LMULMAX1-LABEL: zextload_v16i16_v16i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle16.v v26, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v9, v25
+; LMULMAX1-NEXT:    vzext.vf2 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v11, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v16i16_v16i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vle16.v v26, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf2 v8, v26
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i16>, <16 x i16>* %x
+  %z = zext <16 x i16> %y to <16 x i32>
+  ret <16 x i32> %z
+}
+
+define <16 x i64> @sextload_v16i16_v16i64(<16 x i16>* %x) {
+; LMULMAX1-LABEL: sextload_v16i16_v16i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle16.v v26, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v10, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v11, v27
+; LMULMAX1-NEXT:    vsext.vf4 v12, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v14, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v15, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v9, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf4 v13, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v16i16_v16i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vle16.v v26, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf4 v8, v26
+; LMULMAX4-NEXT:    vsetivli a0, 8, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vslidedown.vi v26, v26, 8
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf4 v12, v26
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i16>, <16 x i16>* %x
+  %z = sext <16 x i16> %y to <16 x i64>
+  ret <16 x i64> %z
+}
+
+define <16 x i64> @zextload_v16i16_v16i64(<16 x i16>* %x) {
+; LMULMAX1-LABEL: zextload_v16i16_v16i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vle16.v v25, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle16.v v26, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v25, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v10, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v11, v27
+; LMULMAX1-NEXT:    vzext.vf4 v12, v26
+; LMULMAX1-NEXT:    vsetivli a0, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v26, 4
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v14, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v15, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v9, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf4 v13, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v16i16_v16i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vle16.v v26, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf4 v8, v26
+; LMULMAX4-NEXT:    vsetivli a0, 8, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vslidedown.vi v26, v26, 8
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf4 v12, v26
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i16>, <16 x i16>* %x
+  %z = zext <16 x i16> %y to <16 x i64>
+  ret <16 x i64> %z
+}
+
+define void @truncstore_v2i32_v2i8(<2 x i32> %x, <2 x i8>* %z) {
+; CHECK-LABEL: truncstore_v2i32_v2i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 2, e8,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vse8.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <2 x i32> %x to <2 x i8>
+  store <2 x i8> %y, <2 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v2i32_v2i16(<2 x i32> %x, <2 x i16>* %z) {
+; CHECK-LABEL: truncstore_v2i32_v2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <2 x i32> %x to <2 x i16>
+  store <2 x i16> %y, <2 x i16>* %z
+  ret void
+}
+
+define <2 x i64> @sextload_v2i32_v2i64(<2 x i32>* %x) {
+; CHECK-LABEL: sextload_v2i32_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; CHECK-NEXT:    vsext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i32>, <2 x i32>* %x
+  %z = sext <2 x i32> %y to <2 x i64>
+  ret <2 x i64> %z
+}
+
+define <2 x i64> @zextload_v2i32_v2i64(<2 x i32>* %x) {
+; CHECK-LABEL: zextload_v2i32_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; CHECK-NEXT:    vzext.vf2 v8, v25
+; CHECK-NEXT:    ret
+  %y = load <2 x i32>, <2 x i32>* %x
+  %z = zext <2 x i32> %y to <2 x i64>
+  ret <2 x i64> %z
+}
+
+define void @truncstore_v4i32_v4i8(<4 x i32> %x, <4 x i8>* %z) {
+; CHECK-LABEL: truncstore_v4i32_v4i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; CHECK-NEXT:    vse8.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <4 x i32> %x to <4 x i8>
+  store <4 x i8> %y, <4 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v4i32_v4i16(<4 x i32> %x, <4 x i16>* %z) {
+; CHECK-LABEL: truncstore_v4i32_v4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <4 x i32> %x to <4 x i16>
+  store <4 x i16> %y, <4 x i16>* %z
+  ret void
+}
+
+define <4 x i64> @sextload_v4i32_v4i64(<4 x i32>* %x) {
+; LMULMAX1-LABEL: sextload_v4i32_v4i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vle32.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v4i32_v4i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX4-NEXT:    vle32.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
+; LMULMAX4-NEXT:    vsext.vf2 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <4 x i32>, <4 x i32>* %x
+  %z = sext <4 x i32> %y to <4 x i64>
+  ret <4 x i64> %z
+}
+
+define <4 x i64> @zextload_v4i32_v4i64(<4 x i32>* %x) {
+; LMULMAX1-LABEL: zextload_v4i32_v4i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vle32.v v25, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v9, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v4i32_v4i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX4-NEXT:    vle32.v v25, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 4, e64,m2,ta,mu
+; LMULMAX4-NEXT:    vzext.vf2 v8, v25
+; LMULMAX4-NEXT:    ret
+  %y = load <4 x i32>, <4 x i32>* %x
+  %z = zext <4 x i32> %y to <4 x i64>
+  ret <4 x i64> %z
+}
+
+define void @truncstore_v8i32_v8i8(<8 x i32> %x, <8 x i8>* %z) {
+; LMULMAX1-LABEL: truncstore_v8i32_v8i8:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v27, 4
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vse8.v v25, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v8i32_v8i8:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,mf2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vse8.v v26, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <8 x i32> %x to <8 x i8>
+  store <8 x i8> %y, <8 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v8i32_v8i16(<8 x i32> %x, <8 x i16>* %z) {
+; LMULMAX1-LABEL: truncstore_v8i32_v8i16:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vse16.v v26, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v8i32_v8i16:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX4-NEXT:    vse16.v v25, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <8 x i32> %x to <8 x i16>
+  store <8 x i16> %y, <8 x i16>* %z
+  ret void
+}
+
+define <8 x i64> @sextload_v8i32_v8i64(<8 x i32>* %x) {
+; LMULMAX1-LABEL: sextload_v8i32_v8i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vle32.v v25, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle32.v v26, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v9, v25
+; LMULMAX1-NEXT:    vsext.vf2 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v11, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v8i32_v8i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vle32.v v26, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf2 v8, v26
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i32>, <8 x i32>* %x
+  %z = sext <8 x i32> %y to <8 x i64>
+  ret <8 x i64> %z
+}
+
+define <8 x i64> @zextload_v8i32_v8i64(<8 x i32>* %x) {
+; LMULMAX1-LABEL: zextload_v8i32_v8i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vle32.v v25, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle32.v v26, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v8, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v9, v25
+; LMULMAX1-NEXT:    vzext.vf2 v10, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v11, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v8i32_v8i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vle32.v v26, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf2 v8, v26
+; LMULMAX4-NEXT:    ret
+  %y = load <8 x i32>, <8 x i32>* %x
+  %z = zext <8 x i32> %y to <8 x i64>
+  ret <8 x i64> %z
+}
+
+define void @truncstore_v16i32_v16i8(<16 x i32> %x, <16 x i8>* %z) {
+; LMULMAX1-LABEL: truncstore_v16i32_v16i8:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v27, v25
+; LMULMAX1-NEXT:    vslideup.vi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v28, 4
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v28, 4
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v27, 8
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vse8.v v26, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v16i32_v16i8:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v26, 0
+; LMULMAX4-NEXT:    vse8.v v25, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <16 x i32> %x to <16 x i8>
+  store <16 x i8> %y, <16 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v16i32_v16i16(<16 x i32> %x, <16 x i16>* %z) {
+; LMULMAX1-LABEL: truncstore_v16i32_v16i16:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v27, v26
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 4
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 4
+; LMULMAX1-NEXT:    addi a1, a0, 16
+; LMULMAX1-NEXT:    vsetivli a2, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vse16.v v26, (a1)
+; LMULMAX1-NEXT:    vse16.v v27, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v16i32_v16i16:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX4-NEXT:    vse16.v v26, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <16 x i32> %x to <16 x i16>
+  store <16 x i16> %y, <16 x i16>* %z
+  ret void
+}
+
+define <16 x i64> @sextload_v16i32_v16i64(<16 x i32>* %x) {
+; LMULMAX1-LABEL: sextload_v16i32_v16i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    addi a1, a0, 48
+; LMULMAX1-NEXT:    vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vle32.v v25, (a1)
+; LMULMAX1-NEXT:    addi a1, a0, 32
+; LMULMAX1-NEXT:    vle32.v v26, (a1)
+; LMULMAX1-NEXT:    vle32.v v27, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle32.v v28, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v8, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v9, v27
+; LMULMAX1-NEXT:    vsext.vf2 v10, v28
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v28, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v11, v27
+; LMULMAX1-NEXT:    vsext.vf2 v12, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v13, v26
+; LMULMAX1-NEXT:    vsext.vf2 v14, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vsext.vf2 v15, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: sextload_v16i32_v16i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vle32.v v28, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf2 v8, v28
+; LMULMAX4-NEXT:    vsetivli a0, 8, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vslidedown.vi v28, v28, 8
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vsext.vf2 v12, v28
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i32>, <16 x i32>* %x
+  %z = sext <16 x i32> %y to <16 x i64>
+  ret <16 x i64> %z
+}
+
+define <16 x i64> @zextload_v16i32_v16i64(<16 x i32>* %x) {
+; LMULMAX1-LABEL: zextload_v16i32_v16i64:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    addi a1, a0, 48
+; LMULMAX1-NEXT:    vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vle32.v v25, (a1)
+; LMULMAX1-NEXT:    addi a1, a0, 32
+; LMULMAX1-NEXT:    vle32.v v26, (a1)
+; LMULMAX1-NEXT:    vle32.v v27, (a0)
+; LMULMAX1-NEXT:    addi a0, a0, 16
+; LMULMAX1-NEXT:    vle32.v v28, (a0)
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v8, v27
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v27, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v9, v27
+; LMULMAX1-NEXT:    vzext.vf2 v10, v28
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v27, v28, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v11, v27
+; LMULMAX1-NEXT:    vzext.vf2 v12, v26
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v26, v26, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v13, v26
+; LMULMAX1-NEXT:    vzext.vf2 v14, v25
+; LMULMAX1-NEXT:    vsetivli a0, 2, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vslidedown.vi v25, v25, 2
+; LMULMAX1-NEXT:    vsetivli a0, 2, e64,m1,ta,mu
+; LMULMAX1-NEXT:    vzext.vf2 v15, v25
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: zextload_v16i32_v16i64:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vle32.v v28, (a0)
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf2 v8, v28
+; LMULMAX4-NEXT:    vsetivli a0, 8, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vslidedown.vi v28, v28, 8
+; LMULMAX4-NEXT:    vsetivli a0, 8, e64,m4,ta,mu
+; LMULMAX4-NEXT:    vzext.vf2 v12, v28
+; LMULMAX4-NEXT:    ret
+  %y = load <16 x i32>, <16 x i32>* %x
+  %z = zext <16 x i32> %y to <16 x i64>
+  ret <16 x i64> %z
+}
+
+define void @truncstore_v2i64_v2i8(<2 x i64> %x, <2 x i8>* %z) {
+; CHECK-LABEL: truncstore_v2i64_v2i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v26, 0
+; CHECK-NEXT:    vsetivli a1, 2, e8,m1,ta,mu
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <2 x i64> %x to <2 x i8>
+  store <2 x i8> %y, <2 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v2i64_v2i16(<2 x i64> %x, <2 x i16>* %z) {
+; CHECK-LABEL: truncstore_v2i64_v2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; CHECK-NEXT:    vnsrl.wi v26, v25, 0
+; CHECK-NEXT:    vsetivli a1, 2, e16,m1,ta,mu
+; CHECK-NEXT:    vse16.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <2 x i64> %x to <2 x i16>
+  store <2 x i16> %y, <2 x i16>* %z
+  ret void
+}
+
+define void @truncstore_v2i64_v2i32(<2 x i64> %x, <2 x i32>* %z) {
+; CHECK-LABEL: truncstore_v2i64_v2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; CHECK-NEXT:    vnsrl.wi v25, v8, 0
+; CHECK-NEXT:    vsetivli a1, 2, e32,m1,ta,mu
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = trunc <2 x i64> %x to <2 x i32>
+  store <2 x i32> %y, <2 x i32>* %z
+  ret void
+}
+
+define void @truncstore_v4i64_v4i8(<4 x i64> %x, <4 x i8>* %z) {
+; LMULMAX1-LABEL: truncstore_v4i64_v4i8:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vse8.v v26, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v4i64_v4i8:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX4-NEXT:    vsetivli a1, 4, e8,mf4,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v26, 0
+; LMULMAX4-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vse8.v v25, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <4 x i64> %x to <4 x i8>
+  store <4 x i8> %y, <4 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v4i64_v4i16(<4 x i64> %x, <4 x i16>* %z) {
+; LMULMAX1-LABEL: truncstore_v4i64_v4i16:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v27, 2
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vse16.v v25, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v4i64_v4i16:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 4, e16,mf2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX4-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vse16.v v26, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <4 x i64> %x to <4 x i16>
+  store <4 x i16> %y, <4 x i16>* %z
+  ret void
+}
+
+define void @truncstore_v4i64_v4i32(<4 x i64> %x, <4 x i32>* %z) {
+; LMULMAX1-LABEL: truncstore_v4i64_v4i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vse32.v v26, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v4i64_v4i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX4-NEXT:    vse32.v v25, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <4 x i64> %x to <4 x i32>
+  store <4 x i32> %y, <4 x i32>* %z
+  ret void
+}
+
+define void @truncstore_v8i64_v8i8(<8 x i64> %x, <8 x i8>* %z) {
+; LMULMAX1-LABEL: truncstore_v8i64_v8i8:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v27, v26
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v27, 4
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vse8.v v25, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v8i64_v8i8:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v26, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,mf2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vse8.v v26, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <8 x i64> %x to <8 x i8>
+  store <8 x i8> %y, <8 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v8i64_v8i16(<8 x i64> %x, <8 x i16>* %z) {
+; LMULMAX1-LABEL: truncstore_v8i64_v8i16:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v27, v25
+; LMULMAX1-NEXT:    vslideup.vi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v28, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v28, 2
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v27, 4
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vse16.v v26, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v8i64_v8i16:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v26, 0
+; LMULMAX4-NEXT:    vse16.v v25, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <8 x i64> %x to <8 x i16>
+  store <8 x i16> %y, <8 x i16>* %z
+  ret void
+}
+
+define void @truncstore_v8i64_v8i32(<8 x i64> %x, <8 x i32>* %z) {
+; LMULMAX1-LABEL: truncstore_v8i64_v8i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v27, v26
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
+; LMULMAX1-NEXT:    addi a1, a0, 16
+; LMULMAX1-NEXT:    vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vse32.v v26, (a1)
+; LMULMAX1-NEXT:    vse32.v v27, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v8i64_v8i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX4-NEXT:    vse32.v v26, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <8 x i64> %x to <8 x i32>
+  store <8 x i32> %y, <8 x i32>* %z
+  ret void
+}
+
+define void @truncstore_v16i64_v16i8(<16 x i64> %x, <16 x i8>* %z) {
+; LMULMAX1-LABEL: truncstore_v16i64_v16i8:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v14, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v26, v25
+; LMULMAX1-NEXT:    vslideup.vi v26, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v15, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v27, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v12, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v28, v25
+; LMULMAX1-NEXT:    vslideup.vi v28, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v13, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v29, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v29, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v28, v27, 2
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v29, v27
+; LMULMAX1-NEXT:    vslideup.vi v29, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v29, v26, 4
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v28, v25
+; LMULMAX1-NEXT:    vslideup.vi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v30, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v30, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v28, v26, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v30, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v30, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v30, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e8,mf8,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v30, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v26, 2
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v28, 4
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v29, 8
+; LMULMAX1-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX1-NEXT:    vse8.v v25, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v16i64_v16i8:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v25, v26, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,mf2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vmv.v.i v25, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,m1,tu,mu
+; LMULMAX4-NEXT:    vslideup.vi v25, v26, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v12, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e8,mf2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v28, 0
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,tu,mu
+; LMULMAX4-NEXT:    vslideup.vi v25, v26, 8
+; LMULMAX4-NEXT:    vsetivli a1, 16, e8,m1,ta,mu
+; LMULMAX4-NEXT:    vse8.v v25, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <16 x i64> %x to <16 x i8>
+  store <16 x i8> %y, <16 x i8>* %z
+  ret void
+}
+
+define void @truncstore_v16i64_v16i16(<16 x i64> %x, <16 x i16>* %z) {
+; LMULMAX1-LABEL: truncstore_v16i64_v16i16:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v27, v25
+; LMULMAX1-NEXT:    vslideup.vi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v28, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v26, v25
+; LMULMAX1-NEXT:    vslideup.vi v26, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v28, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v29, v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v29, 2
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v28, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v29, v28
+; LMULMAX1-NEXT:    vslideup.vi v29, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v29, v27, 4
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v26, v14, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v26, v25
+; LMULMAX1-NEXT:    vslideup.vi v26, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v15, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v30, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v30, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v12, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v30, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v30, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v27, v13, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e16,mf4,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v30, v27, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v25, v30, 2
+; LMULMAX1-NEXT:    vslideup.vi v28, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 8, e16,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v28, v26, 4
+; LMULMAX1-NEXT:    addi a1, a0, 16
+; LMULMAX1-NEXT:    vsetivli a2, 8, e16,m1,ta,mu
+; LMULMAX1-NEXT:    vse16.v v28, (a1)
+; LMULMAX1-NEXT:    vse16.v v29, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v16i64_v16i16:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v26, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v28, v26, 0
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vmv.v.i v26, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m2,tu,mu
+; LMULMAX4-NEXT:    vslideup.vi v26, v28, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v28, v12, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v30, v28, 0
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,tu,mu
+; LMULMAX4-NEXT:    vslideup.vi v26, v30, 8
+; LMULMAX4-NEXT:    vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX4-NEXT:    vse16.v v26, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <16 x i64> %x to <16 x i16>
+  store <16 x i16> %y, <16 x i16>* %z
+  ret void
+}
+
+define void @truncstore_v16i64_v16i32(<16 x i64> %x, <16 x i32>* %z) {
+; LMULMAX1-LABEL: truncstore_v16i64_v16i32:
+; LMULMAX1:       # %bb.0:
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v8, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vmv.v.i v26, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v27, v26
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v9, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v27, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v10, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v28, v26
+; LMULMAX1-NEXT:    vslideup.vi v28, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v11, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v28, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v12, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vmv1r.v v29, v26
+; LMULMAX1-NEXT:    vslideup.vi v29, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v13, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v29, v25, 2
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v14, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 0
+; LMULMAX1-NEXT:    vsetivli a1, 2, e32,mf2,ta,mu
+; LMULMAX1-NEXT:    vnsrl.wi v25, v15, 0
+; LMULMAX1-NEXT:    vsetivli a1, 4, e32,m1,tu,mu
+; LMULMAX1-NEXT:    vslideup.vi v26, v25, 2
+; LMULMAX1-NEXT:    addi a1, a0, 48
+; LMULMAX1-NEXT:    vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-NEXT:    vse32.v v26, (a1)
+; LMULMAX1-NEXT:    addi a1, a0, 32
+; LMULMAX1-NEXT:    vse32.v v29, (a1)
+; LMULMAX1-NEXT:    addi a1, a0, 16
+; LMULMAX1-NEXT:    vse32.v v28, (a1)
+; LMULMAX1-NEXT:    vse32.v v27, (a0)
+; LMULMAX1-NEXT:    ret
+;
+; LMULMAX4-LABEL: truncstore_v16i64_v16i32:
+; LMULMAX4:       # %bb.0:
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v28, v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vmv.v.i v8, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m4,tu,mu
+; LMULMAX4-NEXT:    vslideup.vi v8, v28, 0
+; LMULMAX4-NEXT:    vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX4-NEXT:    vnsrl.wi v28, v12, 0
+; LMULMAX4-NEXT:    vsetivli a1, 16, e32,m4,tu,mu
+; LMULMAX4-NEXT:    vslideup.vi v8, v28, 8
+; LMULMAX4-NEXT:    vsetivli a1, 16, e32,m4,ta,mu
+; LMULMAX4-NEXT:    vse32.v v8, (a0)
+; LMULMAX4-NEXT:    ret
+  %y = trunc <16 x i64> %x to <16 x i32>
+  store <16 x i32> %y, <16 x i32>* %z
+  ret void
+}
+
+define <vscale x 2 x float> @extload_nxv2f16_nxv2f32(<vscale x 2 x half>* %x) {
+; CHECK-LABEL: extload_nxv2f16_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x half>, <vscale x 2 x half>* %x
+  %z = fpext <vscale x 2 x half> %y to <vscale x 2 x float>
+  ret <vscale x 2 x float> %z
+}
+
+define <vscale x 2 x double> @extload_nxv2f16_nxv2f64(<vscale x 2 x half>* %x) {
+; CHECK-LABEL: extload_nxv2f16_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfwcvt.f.f.v v26, v25
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x half>, <vscale x 2 x half>* %x
+  %z = fpext <vscale x 2 x half> %y to <vscale x 2 x double>
+  ret <vscale x 2 x double> %z
+}
+
+define <vscale x 4 x float> @extload_nxv4f16_nxv4f32(<vscale x 4 x half>* %x) {
+; CHECK-LABEL: extload_nxv4f16_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x half>, <vscale x 4 x half>* %x
+  %z = fpext <vscale x 4 x half> %y to <vscale x 4 x float>
+  ret <vscale x 4 x float> %z
+}
+
+define <vscale x 4 x double> @extload_nxv4f16_nxv4f64(<vscale x 4 x half>* %x) {
+; CHECK-LABEL: extload_nxv4f16_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re16.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v26, v25
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x half>, <vscale x 4 x half>* %x
+  %z = fpext <vscale x 4 x half> %y to <vscale x 4 x double>
+  ret <vscale x 4 x double> %z
+}
+
+define <vscale x 8 x float> @extload_nxv8f16_nxv8f32(<vscale x 8 x half>* %x) {
+; CHECK-LABEL: extload_nxv8f16_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x half>, <vscale x 8 x half>* %x
+  %z = fpext <vscale x 8 x half> %y to <vscale x 8 x float>
+  ret <vscale x 8 x float> %z
+}
+
+define <vscale x 8 x double> @extload_nxv8f16_nxv8f64(<vscale x 8 x half>* %x) {
+; CHECK-LABEL: extload_nxv8f16_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re16.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v28, v26
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x half>, <vscale x 8 x half>* %x
+  %z = fpext <vscale x 8 x half> %y to <vscale x 8 x double>
+  ret <vscale x 8 x double> %z
+}
+
+define <vscale x 16 x float> @extload_nxv16f16_nxv16f32(<vscale x 16 x half>* %x) {
+; CHECK-LABEL: extload_nxv16f16_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re16.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x half>, <vscale x 16 x half>* %x
+  %z = fpext <vscale x 16 x half> %y to <vscale x 16 x float>
+  ret <vscale x 16 x float> %z
+}
+
+define <vscale x 16 x double> @extload_nxv16f16_nxv16f64(<vscale x 16 x half>* %x) {
+; CHECK-LABEL: extload_nxv16f16_nxv16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re16.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v16, v28
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v16
+; CHECK-NEXT:    vsetvli a0, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v24, v30
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v16, v24
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x half>, <vscale x 16 x half>* %x
+  %z = fpext <vscale x 16 x half> %y to <vscale x 16 x double>
+  ret <vscale x 16 x double> %z
+}
+
+define void @truncstore_nxv2f32_nxv2f16(<vscale x 2 x float> %x, <vscale x 2 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv2f32_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 2 x float> %x to <vscale x 2 x half>
+  store <vscale x 2 x half> %y, <vscale x 2 x half>* %z
+  ret void
+}
+
+define <vscale x 2 x double> @extload_nxv2f32_nxv2f64(<vscale x 2 x float>* %x) {
+; CHECK-LABEL: extload_nxv2f32_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl1re32.v v25, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v25
+; CHECK-NEXT:    ret
+  %y = load <vscale x 2 x float>, <vscale x 2 x float>* %x
+  %z = fpext <vscale x 2 x float> %y to <vscale x 2 x double>
+  ret <vscale x 2 x double> %z
+}
+
+define void @truncstore_nxv4f32_nxv4f16(<vscale x 4 x float> %x, <vscale x 4 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv4f32_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 4 x float> %x to <vscale x 4 x half>
+  store <vscale x 4 x half> %y, <vscale x 4 x half>* %z
+  ret void
+}
+
+define <vscale x 4 x double> @extload_nxv4f32_nxv4f64(<vscale x 4 x float>* %x) {
+; CHECK-LABEL: extload_nxv4f32_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl2re32.v v26, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v26
+; CHECK-NEXT:    ret
+  %y = load <vscale x 4 x float>, <vscale x 4 x float>* %x
+  %z = fpext <vscale x 4 x float> %y to <vscale x 4 x double>
+  ret <vscale x 4 x double> %z
+}
+
+define void @truncstore_nxv8f32_nxv8f16(<vscale x 8 x float> %x, <vscale x 8 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv8f32_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v8
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 8 x float> %x to <vscale x 8 x half>
+  store <vscale x 8 x half> %y, <vscale x 8 x half>* %z
+  ret void
+}
+
+define <vscale x 8 x double> @extload_nxv8f32_nxv8f64(<vscale x 8 x float>* %x) {
+; CHECK-LABEL: extload_nxv8f32_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl4re32.v v28, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 8 x float>, <vscale x 8 x float>* %x
+  %z = fpext <vscale x 8 x float> %y to <vscale x 8 x double>
+  ret <vscale x 8 x double> %z
+}
+
+define void @truncstore_nxv16f32_nxv16f16(<vscale x 16 x float> %x, <vscale x 16 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv16f32_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v28, v8
+; CHECK-NEXT:    vs4r.v v28, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 16 x float> %x to <vscale x 16 x half>
+  store <vscale x 16 x half> %y, <vscale x 16 x half>* %z
+  ret void
+}
+
+define <vscale x 16 x double> @extload_nxv16f32_nxv16f64(<vscale x 16 x float>* %x) {
+; CHECK-LABEL: extload_nxv16f32_nxv16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re32.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfwcvt.f.f.v v8, v24
+; CHECK-NEXT:    vfwcvt.f.f.v v16, v28
+; CHECK-NEXT:    ret
+  %y = load <vscale x 16 x float>, <vscale x 16 x float>* %x
+  %z = fpext <vscale x 16 x float> %y to <vscale x 16 x double>
+  ret <vscale x 16 x double> %z
+}
+
+define void @truncstore_nxv2f64_nxv2f16(<vscale x 2 x double> %x, <vscale x 2 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv2f64_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v25, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v25
+; CHECK-NEXT:    vse16.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 2 x double> %x to <vscale x 2 x half>
+  store <vscale x 2 x half> %y, <vscale x 2 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv2f64_nxv2f32(<vscale x 2 x double> %x, <vscale x 2 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv2f64_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v8
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 2 x double> %x to <vscale x 2 x float>
+  store <vscale x 2 x float> %y, <vscale x 2 x float>* %z
+  ret void
+}
+
+define void @truncstore_nxv4f64_nxv4f16(<vscale x 4 x double> %x, <vscale x 4 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv4f64_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v26, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v25, v26
+; CHECK-NEXT:    vs1r.v v25, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 4 x double> %x to <vscale x 4 x half>
+  store <vscale x 4 x half> %y, <vscale x 4 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv4f64_nxv4f32(<vscale x 4 x double> %x, <vscale x 4 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv4f64_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v8
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 4 x double> %x to <vscale x 4 x float>
+  store <vscale x 4 x float> %y, <vscale x 4 x float>* %z
+  ret void
+}
+
+define void @truncstore_nxv8f64_nxv8f16(<vscale x 8 x double> %x, <vscale x 8 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv8f64_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v28, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v26, v28
+; CHECK-NEXT:    vs2r.v v26, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 8 x double> %x to <vscale x 8 x half>
+  store <vscale x 8 x half> %y, <vscale x 8 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv8f64_nxv8f32(<vscale x 8 x double> %x, <vscale x 8 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv8f64_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v28, v8
+; CHECK-NEXT:    vs4r.v v28, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 8 x double> %x to <vscale x 8 x float>
+  store <vscale x 8 x float> %y, <vscale x 8 x float>* %z
+  ret void
+}
+
+define void @truncstore_nxv16f64_nxv16f16(<vscale x 16 x double> %x, <vscale x 16 x half>* %z) {
+; CHECK-LABEL: truncstore_nxv16f64_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v28, v8
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v8, v28
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfncvt.rod.f.f.w v28, v16
+; CHECK-NEXT:    vsetvli a1, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v10, v28
+; CHECK-NEXT:    vs4r.v v8, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 16 x double> %x to <vscale x 16 x half>
+  store <vscale x 16 x half> %y, <vscale x 16 x half>* %z
+  ret void
+}
+
+define void @truncstore_nxv16f64_nxv16f32(<vscale x 16 x double> %x, <vscale x 16 x float>* %z) {
+; CHECK-LABEL: truncstore_nxv16f64_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vfncvt.f.f.w v24, v8
+; CHECK-NEXT:    vfncvt.f.f.w v28, v16
+; CHECK-NEXT:    vs8r.v v24, (a0)
+; CHECK-NEXT:    ret
+  %y = fptrunc <vscale x 16 x double> %x to <vscale x 16 x float>
+  store <vscale x 16 x float> %y, <vscale x 16 x float>* %z
+  ret void
+}