diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 9679b0e5..19d53cc6 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -167,6 +167,7 @@ if (ARCHITECTURE_x86_64)
          backend_x64/emit_x64_packed.cpp
          backend_x64/emit_x64_saturation.cpp
          backend_x64/emit_x64_vector.cpp
+         backend_x64/emit_x64_vector_floating_point.cpp
          backend_x64/hostloc.cpp
          backend_x64/hostloc.h
          backend_x64/jitstate_info.h
diff --git a/src/backend_x64/emit_x64_vector_floating_point.cpp b/src/backend_x64/emit_x64_vector_floating_point.cpp
new file mode 100644
index 00000000..53003f1f
--- /dev/null
+++ b/src/backend_x64/emit_x64_vector_floating_point.cpp
@@ -0,0 +1,36 @@
+/* This file is part of the dynarmic project.
+ * Copyright (c) 2016 MerryMage
+ * This software may be used and distributed according to the terms of the GNU
+ * General Public License version 2 or any later version.
+ */
+
+#include "backend_x64/block_of_code.h"
+#include "backend_x64/emit_x64.h"
+#include "frontend/ir/basic_block.h"
+#include "frontend/ir/microinstruction.h"
+
+namespace Dynarmic::BackendX64 {
+
+using namespace Xbyak::util;
+
+template <typename Function>
+static void EmitVectorOperation(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst, Function fn) {
+    auto args = ctx.reg_alloc.GetArgumentInfo(inst);
+
+    Xbyak::Xmm xmm_a = ctx.reg_alloc.UseScratchXmm(args[0]);
+    Xbyak::Xmm xmm_b = ctx.reg_alloc.UseXmm(args[1]);
+
+    (code.*fn)(xmm_a, xmm_b);
+
+    ctx.reg_alloc.DefineValue(inst, xmm_a);
+}
+
+void EmitX64::EmitFPVectorSub32(EmitContext& ctx, IR::Inst* inst) {
+    EmitVectorOperation(code, ctx, inst, &Xbyak::CodeGenerator::subps);
+}
+
+void EmitX64::EmitFPVectorSub64(EmitContext& ctx, IR::Inst* inst) {
+    EmitVectorOperation(code, ctx, inst, &Xbyak::CodeGenerator::subpd);
+}
+
+} // namespace Dynarmic::BackendX64
diff --git a/src/frontend/ir/ir_emitter.cpp b/src/frontend/ir/ir_emitter.cpp
index 402319e3..1c79797c 100644
--- a/src/frontend/ir/ir_emitter.cpp
+++ b/src/frontend/ir/ir_emitter.cpp
@@ -1139,6 +1139,17 @@ U64 IREmitter::FPU32ToDouble(const U32& a, bool round_to_nearest, bool fpscr_con
     return Inst<U64>(Opcode::FPU32ToDouble, a, Imm1(round_to_nearest));
 }
 
+U128 IREmitter::FPVectorSub(size_t esize, const U128& a, const U128& b) {
+    switch (esize) {
+    case 32:
+        return Inst<U128>(Opcode::FPVectorSub32, a, b);
+    case 64:
+        return Inst<U128>(Opcode::FPVectorSub64, a, b);
+    }
+    UNREACHABLE();
+    return {};
+}
+
 void IREmitter::Breakpoint() {
     Inst(Opcode::Breakpoint);
 }
diff --git a/src/frontend/ir/ir_emitter.h b/src/frontend/ir/ir_emitter.h
index 436f8d11..c25d891f 100644
--- a/src/frontend/ir/ir_emitter.h
+++ b/src/frontend/ir/ir_emitter.h
@@ -247,6 +247,8 @@ public:
     U64 FPS32ToDouble(const U32& a, bool round_to_nearest, bool fpscr_controlled);
     U64 FPU32ToDouble(const U32& a, bool round_to_nearest, bool fpscr_controlled);
 
+    U128 FPVectorSub(size_t esize, const U128& a, const U128& b);
+
     void Breakpoint();
 
     void SetTerm(const Terminal& terminal);
diff --git a/src/frontend/ir/opcodes.inc b/src/frontend/ir/opcodes.inc
index 54690b71..85bd88b7 100644
--- a/src/frontend/ir/opcodes.inc
+++ b/src/frontend/ir/opcodes.inc
@@ -289,6 +289,10 @@ OPCODE(FPS32ToSingle,               T::U32,         T::U32,         T::U1
 OPCODE(FPU32ToDouble,               T::U64,         T::U32,         T::U1                           )
 OPCODE(FPS32ToDouble,               T::U64,         T::U32,         T::U1                           )
 
+// Floating-point vector instructions
+OPCODE(FPVectorSub32,               T::U128,        T::U128,        T::U128                         )
+OPCODE(FPVectorSub64,               T::U128,        T::U128,        T::U128                         )
+
 // A32 Memory access
 A32OPC(ClearExclusive,              T::Void,                                                        )
 A32OPC(SetExclusive,                T::Void,        T::U32,         T::U8                           )