diff --git a/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp b/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp
index 85a7663993a04..5973b63b5a802 100644
--- a/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp
@@ -1099,6 +1099,36 @@ bool AArch64RegisterInfo::getRegAllocationHints(
     const VirtRegMap *VRM, const LiveRegMatrix *Matrix) const {
   const MachineRegisterInfo &MRI = MF.getRegInfo();
 
+  // The SVE calling convention preserves registers Z8-Z23. As a result, there
+  // are no ZPR2Strided or ZPR4Strided registers that do not overlap with the
+  // callee-saved registers and so by default these will be pushed to the back
+  // of the allocation order for the ZPRStridedOrContiguous classes.
+  // If any of the instructions which define VirtReg are used by the
+  // FORM_TRANSPOSED_REG_TUPLE pseudo, we want to favour reducing copy
+  // instructions over reducing the number of clobbered callee-save registers,
+  // so we add the strided registers as a hint.
+  unsigned RegID = MRI.getRegClass(VirtReg)->getID();
+  // Look through uses of the register for FORM_TRANSPOSED_REG_TUPLE.
+  if ((RegID == AArch64::ZPR2StridedOrContiguousRegClassID ||
+       RegID == AArch64::ZPR4StridedOrContiguousRegClassID) &&
+      any_of(MRI.use_nodbg_instructions(VirtReg), [](const MachineInstr &Use) {
+        return Use.getOpcode() ==
+                   AArch64::FORM_TRANSPOSED_REG_TUPLE_X2_PSEUDO ||
+               Use.getOpcode() == AArch64::FORM_TRANSPOSED_REG_TUPLE_X4_PSEUDO;
+      })) {
+    const TargetRegisterClass *StridedRC =
+        RegID == AArch64::ZPR2StridedOrContiguousRegClassID
+            ? &AArch64::ZPR2StridedRegClass
+            : &AArch64::ZPR4StridedRegClass;
+
+    for (MCPhysReg Reg : Order)
+      if (StridedRC->contains(Reg))
+        Hints.push_back(Reg);
+
+    return TargetRegisterInfo::getRegAllocationHints(VirtReg, Order, Hints, MF,
+                                                     VRM);
+  }
+
   for (MachineInstr &MI : MRI.def_instructions(VirtReg)) {
     if (MI.getOpcode() != AArch64::FORM_TRANSPOSED_REG_TUPLE_X2_PSEUDO &&
         MI.getOpcode() != AArch64::FORM_TRANSPOSED_REG_TUPLE_X4_PSEUDO)
diff --git a/llvm/test/CodeGen/AArch64/sme2-intrinsics-int-dots.ll b/llvm/test/CodeGen/AArch64/sme2-intrinsics-int-dots.ll
index ef569e480ea3d..86ed63d743713 100644
--- a/llvm/test/CodeGen/AArch64/sme2-intrinsics-int-dots.ll
+++ b/llvm/test/CodeGen/AArch64/sme2-intrinsics-int-dots.ll
@@ -602,8 +602,42 @@ entry:
   %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
   %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
   %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
-  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @udot_form_2x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: udot_form_2x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-3
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z9, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z8.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z9.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    udot za.s[w8, 0, vgx2], { z0.b, z1.b }, z0.b[0]
+; CHECK-NEXT:    udot za.s[w8, 0, vgx2], { z8.b, z9.b }, z0.b[0]
+; CHECK-NEXT:    ldr z9, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #3
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
+  %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -650,10 +684,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @udot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: udot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    udot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    udot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    udot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    udot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.udot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -744,8 +850,42 @@ entry:
   %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
   %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
   %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
-  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @usdot_form_2x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: usdot_form_2x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-3
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z9, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z8.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z9.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    usdot za.s[w8, 0, vgx2], { z0.b, z1.b }, z0.b[0]
+; CHECK-NEXT:    usdot za.s[w8, 0, vgx2], { z8.b, z9.b }, z0.b[0]
+; CHECK-NEXT:    ldr z9, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #3
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
+  %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -792,10 +932,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @usdot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: usdot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    usdot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    usdot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    usdot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    usdot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -888,8 +1100,42 @@ entry:
   %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
   %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
   %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
-  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @sdot_form_2x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: sdot_form_2x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-3
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z9, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z8.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z9.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    sdot za.s[w8, 0, vgx2], { z0.b, z1.b }, z0.b[0]
+; CHECK-NEXT:    sdot za.s[w8, 0, vgx2], { z8.b, z9.b }, z0.b[0]
+; CHECK-NEXT:    ldr z9, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #3
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
+  %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -936,10 +1182,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @sdot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: sdot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    sdot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    sdot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    sdot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    sdot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -1032,8 +1350,42 @@ entry:
   %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
   %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
   %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
-  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @sudot_form_2x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: sudot_form_2x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-3
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z9, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z8.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z9.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    sudot za.s[w8, 0, vgx2], { z0.b, z1.b }, z0.b[0]
+; CHECK-NEXT:    sudot za.s[w8, 0, vgx2], { z8.b, z9.b }, z0.b[0]
+; CHECK-NEXT:    ldr z9, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #3
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %4 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x2.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 0
+  %6 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } %4, 1
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %5, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x2.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %6, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -1080,10 +1432,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @sudot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: sudot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    sudot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    sudot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    sudot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    sudot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.sudot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
diff --git a/llvm/test/CodeGen/AArch64/sme2-intrinsics-vdot.ll b/llvm/test/CodeGen/AArch64/sme2-intrinsics-vdot.ll
index 49106e12378be..e7d1050b60799 100644
--- a/llvm/test/CodeGen/AArch64/sme2-intrinsics-vdot.ll
+++ b/llvm/test/CodeGen/AArch64/sme2-intrinsics-vdot.ll
@@ -94,8 +94,43 @@ entry:
   %4 = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld1.pn.x2.nxv8i16(target("aarch64.svcount") %0, ptr %arrayidx2)
   %5 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 0
   %6 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 1
-  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %2, <vscale x 8 x i16> %5, <vscale x 8 x i16> undef, i32 0)
-  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %3, <vscale x 8 x i16> %6, <vscale x 8 x i16> undef, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %2, <vscale x 8 x i16> %5, <vscale x 8 x i16> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %3, <vscale x 8 x i16> %6, <vscale x 8 x i16> poison, i32 0)
+  ret void
+}
+
+define void @svdot_form_2x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: svdot_form_2x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-3
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    add x9, x0, x1
+; CHECK-NEXT:    str z9, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z8, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1h { z0.h, z8.h }, pn8/z, [x0]
+; CHECK-NEXT:    ld1h { z1.h, z9.h }, pn8/z, [x9]
+; CHECK-NEXT:    svdot za.s[w8, 0, vgx2], { z0.h, z1.h }, z0.h[0]
+; CHECK-NEXT:    svdot za.s[w8, 0, vgx2], { z8.h, z9.h }, z0.h[0]
+; CHECK-NEXT:    ldr z9, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #3
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld1.pn.x2.nxv8i16(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %1, 0
+  %3 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %1, 1
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %4 = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld1.pn.x2.nxv8i16(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %5 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 0
+  %6 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 1
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %2, <vscale x 8 x i16> %5, <vscale x 8 x i16> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %3, <vscale x 8 x i16> %6, <vscale x 8 x i16> poison, i32 0)
   ret void
 }
 
@@ -142,10 +177,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @svdot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: svdot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    svdot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    svdot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    svdot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    svdot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.svdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -210,8 +317,43 @@ entry:
   %4 = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld1.pn.x2.nxv8i16(target("aarch64.svcount") %0, ptr %arrayidx2)
   %5 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 0
   %6 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 1
-  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %2, <vscale x 8 x i16> %5, <vscale x 8 x i16> undef, i32 0)
-  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %3, <vscale x 8 x i16> %6, <vscale x 8 x i16> undef, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %2, <vscale x 8 x i16> %5, <vscale x 8 x i16> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %3, <vscale x 8 x i16> %6, <vscale x 8 x i16> poison, i32 0)
+  ret void
+}
+
+define void @uvdot_form_2x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: uvdot_form_2x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-3
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    add x9, x0, x1
+; CHECK-NEXT:    str z9, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z8, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1h { z0.h, z8.h }, pn8/z, [x0]
+; CHECK-NEXT:    ld1h { z1.h, z9.h }, pn8/z, [x9]
+; CHECK-NEXT:    uvdot za.s[w8, 0, vgx2], { z0.h, z1.h }, z0.h[0]
+; CHECK-NEXT:    uvdot za.s[w8, 0, vgx2], { z8.h, z9.h }, z0.h[0]
+; CHECK-NEXT:    ldr z9, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #3
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld1.pn.x2.nxv8i16(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %1, 0
+  %3 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %1, 1
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %4 = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.ld1.pn.x2.nxv8i16(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %5 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 0
+  %6 = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } %4, 1
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %2, <vscale x 8 x i16> %5, <vscale x 8 x i16> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x2.nxv8i16(i32 0, <vscale x 8 x i16> %3, <vscale x 8 x i16> %6, <vscale x 8 x i16> poison, i32 0)
   ret void
 }
 
@@ -258,10 +400,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @uvdot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: uvdot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    uvdot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    uvdot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    uvdot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    uvdot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.uvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -323,10 +537,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @suvdot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: suvdot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    suvdot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    suvdot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    suvdot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    suvdot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.suvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }
 
@@ -388,10 +674,82 @@ entry:
   %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
   %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
   %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
-  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> undef, i32 0)
-  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> undef, i32 0)
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
+  ret void
+}
+
+define void @usvdot_form_4x_tuple_svecc(ptr %ptr, i64 %stride, <vscale x 16 x i8> %scalable_arg) #0 {
+; CHECK-LABEL: usvdot_form_4x_tuple_svecc:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
+; CHECK-NEXT:    addvl sp, sp, #-9
+; CHECK-NEXT:    lsl x9, x1, #1
+; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Folded Spill
+; CHECK-NEXT:    ptrue pn8.b
+; CHECK-NEXT:    str z15, [sp, #1, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    mov w8, wzr
+; CHECK-NEXT:    str z14, [sp, #2, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    add x10, x9, x1
+; CHECK-NEXT:    str z13, [sp, #3, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z12, [sp, #4, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z11, [sp, #5, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z10, [sp, #6, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z9, [sp, #7, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    str z8, [sp, #8, mul vl] // 16-byte Folded Spill
+; CHECK-NEXT:    ld1b { z0.b, z4.b, z8.b, z12.b }, pn8/z, [x0]
+; CHECK-NEXT:    ld1b { z1.b, z5.b, z9.b, z13.b }, pn8/z, [x0, x1]
+; CHECK-NEXT:    ld1b { z2.b, z6.b, z10.b, z14.b }, pn8/z, [x0, x9]
+; CHECK-NEXT:    ld1b { z3.b, z7.b, z11.b, z15.b }, pn8/z, [x0, x10]
+; CHECK-NEXT:    usvdot za.s[w8, 0, vgx4], { z0.b - z3.b }, z0.b[0]
+; CHECK-NEXT:    usvdot za.s[w8, 0, vgx4], { z4.b - z7.b }, z0.b[0]
+; CHECK-NEXT:    usvdot za.s[w8, 0, vgx4], { z8.b - z11.b }, z0.b[0]
+; CHECK-NEXT:    usvdot za.s[w8, 0, vgx4], { z12.b - z15.b }, z0.b[0]
+; CHECK-NEXT:    ldr z15, [sp, #1, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z14, [sp, #2, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z13, [sp, #3, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z12, [sp, #4, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z11, [sp, #5, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z10, [sp, #6, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z9, [sp, #7, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr z8, [sp, #8, mul vl] // 16-byte Folded Reload
+; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Folded Reload
+; CHECK-NEXT:    addvl sp, sp, #9
+; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
+; CHECK-NEXT:    ret
+entry:
+  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
+  %1 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %ptr)
+  %2 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 0
+  %3 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 1
+  %4 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 2
+  %5 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %1, 3
+  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
+  %6 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx2)
+  %7 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 0
+  %8 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 1
+  %9 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 2
+  %10 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %6, 3
+  %mul3 = shl i64 %stride, 1
+  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
+  %11 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx4)
+  %12 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 0
+  %13 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 1
+  %14 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 2
+  %15 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %11, 3
+  %mul5 = mul i64 %stride, 3
+  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
+  %16 = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.ld1.pn.x4.nxv16i8(target("aarch64.svcount") %0, ptr %arrayidx6)
+  %17 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 0
+  %18 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 1
+  %19 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 2
+  %20 = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %16, 3
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %2, <vscale x 16 x i8> %7, <vscale x 16 x i8> %12, <vscale x 16 x i8> %17, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %3, <vscale x 16 x i8> %8, <vscale x 16 x i8> %13, <vscale x 16 x i8> %18, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %4, <vscale x 16 x i8> %9, <vscale x 16 x i8> %14, <vscale x 16 x i8> %19, <vscale x 16 x i8> poison, i32 0)
+  tail call void @llvm.aarch64.sme.usvdot.lane.za32.vg1x4.nxv16i8(i32 0, <vscale x 16 x i8> %5, <vscale x 16 x i8> %10, <vscale x 16 x i8> %15, <vscale x 16 x i8> %20, <vscale x 16 x i8> poison, i32 0)
   ret void
 }