diff --git a/src/xenia/cpu/backend/machine_info.h b/src/xenia/cpu/backend/machine_info.h
index 7734b2a48..ebfe2abca 100644
--- a/src/xenia/cpu/backend/machine_info.h
+++ b/src/xenia/cpu/backend/machine_info.h
@@ -17,6 +17,8 @@ namespace cpu {
 namespace backend {
 
 struct MachineInfo {
+  bool supports_extended_load_store;
+
   struct RegisterSet {
     enum Types {
       INT_TYPES = (1 << 1),
diff --git a/src/xenia/cpu/backend/x64/x64_backend.cc b/src/xenia/cpu/backend/x64/x64_backend.cc
index e02d0a0bb..d77ff60aa 100644
--- a/src/xenia/cpu/backend/x64/x64_backend.cc
+++ b/src/xenia/cpu/backend/x64/x64_backend.cc
@@ -14,6 +14,11 @@
 #include "xenia/cpu/backend/x64/x64_sequences.h"
 #include "xenia/cpu/backend/x64/x64_thunk_emitter.h"
 #include "xenia/cpu/processor.h"
+#include "third_party/xbyak/xbyak/xbyak_util.h"
+
+DEFINE_bool(
+    enable_haswell_instructions, true,
+    "Uses the AVX2/FMA/etc instructions on Haswell processors, if available.");
 
 namespace xe {
 namespace cpu {
@@ -38,6 +43,15 @@ bool X64Backend::Initialize() {
 
   RegisterSequences();
 
+  // Need movbe to do advanced LOAD/STORE tricks.
+  if (FLAGS_enable_haswell_instructions) {
+    Xbyak::util::Cpu cpu;
+    machine_info_.supports_extended_load_store =
+        cpu.has(Xbyak::util::Cpu::tMOVBE);
+  } else {
+    machine_info_.supports_extended_load_store = false;
+  }
+
   machine_info_.register_sets[0] = {
       0, "gpr", MachineInfo::RegisterSet::INT_TYPES, X64Emitter::GPR_COUNT,
   };
diff --git a/src/xenia/cpu/backend/x64/x64_backend.h b/src/xenia/cpu/backend/x64/x64_backend.h
index 6c85d834e..0bbd5c01b 100644
--- a/src/xenia/cpu/backend/x64/x64_backend.h
+++ b/src/xenia/cpu/backend/x64/x64_backend.h
@@ -10,8 +10,12 @@
 #ifndef XENIA_BACKEND_X64_X64_BACKEND_H_
 #define XENIA_BACKEND_X64_X64_BACKEND_H_
 
+#include <gflags/gflags.h>
+
 #include "xenia/cpu/backend/backend.h"
 
+DECLARE_bool(enable_haswell_instructions);
+
 namespace xe {
 namespace cpu {
 namespace backend {
diff --git a/src/xenia/cpu/backend/x64/x64_emitter.cc b/src/xenia/cpu/backend/x64/x64_emitter.cc
index 247a1fb9c..d923806bd 100644
--- a/src/xenia/cpu/backend/x64/x64_emitter.cc
+++ b/src/xenia/cpu/backend/x64/x64_emitter.cc
@@ -31,10 +31,6 @@
 #include "xenia/cpu/thread_state.h"
 #include "xenia/profiling.h"
 
-DEFINE_bool(
-    enable_haswell_instructions, true,
-    "Uses the AVX2/FMA/etc instructions on Haswell processors, if available.");
-
 DEFINE_bool(enable_debugprint_log, false,
             "Log debugprint traps to the active debugger");
 
@@ -87,10 +83,10 @@ X64Emitter::X64Emitter(X64Backend* backend, XbyakAllocator* allocator)
     feature_flags_ |= cpu_.has(Xbyak::util::Cpu::tLZCNT) ? kX64EmitLZCNT : 0;
     feature_flags_ |= cpu_.has(Xbyak::util::Cpu::tBMI2) ? kX64EmitBMI2 : 0;
     feature_flags_ |= cpu_.has(Xbyak::util::Cpu::tF16C) ? kX64EmitF16C : 0;
+    feature_flags_ |= cpu_.has(Xbyak::util::Cpu::tMOVBE) ? kX64EmitMovbe : 0;
   }
 
-  if (!cpu_.has(Xbyak::util::Cpu::tAVX) ||
-      !cpu_.has(Xbyak::util::Cpu::tMOVBE)) {
+  if (!cpu_.has(Xbyak::util::Cpu::tAVX)) {
     XEFATAL(
         "Your CPU is too old to support Xenia. See the FAQ for system "
         "requirements at http://xenia.jp");
diff --git a/src/xenia/cpu/backend/x64/x64_emitter.h b/src/xenia/cpu/backend/x64/x64_emitter.h
index 9038e6ea1..c5e895472 100644
--- a/src/xenia/cpu/backend/x64/x64_emitter.h
+++ b/src/xenia/cpu/backend/x64/x64_emitter.h
@@ -104,6 +104,7 @@ enum X64EmitterFeatureFlags {
   kX64EmitLZCNT = 1 << 3,
   kX64EmitBMI2 = 1 << 4,
   kX64EmitF16C = 1 << 5,
+  kX64EmitMovbe = 1 << 6,
 };
 
 class X64Emitter : public Xbyak::CodeGenerator {
diff --git a/src/xenia/cpu/backend/x64/x64_sequences.cc b/src/xenia/cpu/backend/x64/x64_sequences.cc
index b8ab2456d..b2cc4a7de 100644
--- a/src/xenia/cpu/backend/x64/x64_sequences.cc
+++ b/src/xenia/cpu/backend/x64/x64_sequences.cc
@@ -1524,7 +1524,12 @@ EMITTER(LOAD_I16, MATCH(I<OPCODE_LOAD, I16<>, I64<>>)) {
   static void Emit(X64Emitter& e, const EmitArgType& i) {
     auto addr = ComputeMemoryAddress(e, i.src1);
     if (i.instr->flags & LoadStoreFlags::LOAD_STORE_BYTE_SWAP) {
-      e.movbe(i.dest, e.word[addr]);
+      if (e.IsFeatureEnabled(kX64EmitMovbe)) {
+        e.movbe(i.dest, e.word[addr]);
+      } else {
+        e.mov(i.dest, e.word[addr]);
+        e.ror(i.dest, 8);
+      }
     } else {
       e.mov(i.dest, e.word[addr]);
     }
@@ -1539,7 +1544,12 @@ EMITTER(LOAD_I32, MATCH(I<OPCODE_LOAD, I32<>, I64<>>)) {
   static void Emit(X64Emitter& e, const EmitArgType& i) {
     auto addr = ComputeMemoryAddress(e, i.src1);
     if (i.instr->flags & LoadStoreFlags::LOAD_STORE_BYTE_SWAP) {
-      e.movbe(i.dest, e.dword[addr]);
+      if (e.IsFeatureEnabled(kX64EmitMovbe)) {
+        e.movbe(i.dest, e.dword[addr]);
+      } else {
+        e.mov(i.dest, e.dword[addr]);
+        e.bswap(i.dest);
+      }
     } else {
       e.mov(i.dest, e.dword[addr]);
     }
@@ -1554,7 +1564,12 @@ EMITTER(LOAD_I64, MATCH(I<OPCODE_LOAD, I64<>, I64<>>)) {
   static void Emit(X64Emitter& e, const EmitArgType& i) {
     auto addr = ComputeMemoryAddress(e, i.src1);
     if (i.instr->flags & LoadStoreFlags::LOAD_STORE_BYTE_SWAP) {
-      e.movbe(i.dest, e.qword[addr]);
+      if (e.IsFeatureEnabled(kX64EmitMovbe)) {
+        e.movbe(i.dest, e.qword[addr]);
+      } else {
+        e.mov(i.dest, e.qword[addr]);
+        e.bswap(i.dest);
+      }
     } else {
       e.mov(i.dest, e.qword[addr]);
     }
@@ -1645,7 +1660,11 @@ EMITTER(STORE_I16, MATCH(I<OPCODE_STORE, VoidOp, I64<>, I16<>>)) {
     auto addr = ComputeMemoryAddress(e, i.src1);
     if (i.instr->flags & LoadStoreFlags::LOAD_STORE_BYTE_SWAP) {
       assert_false(i.src2.is_constant);
-      e.movbe(e.word[addr], i.src2);
+      if (e.IsFeatureEnabled(kX64EmitMovbe)) {
+        e.movbe(e.word[addr], i.src2);
+      } else {
+        assert_always("not implemented");
+      }
     } else {
       if (i.src2.is_constant) {
         e.mov(e.word[addr], i.src2.constant());
@@ -1666,7 +1685,11 @@ EMITTER(STORE_I32, MATCH(I<OPCODE_STORE, VoidOp, I64<>, I32<>>)) {
     auto addr = ComputeMemoryAddress(e, i.src1);
     if (i.instr->flags & LoadStoreFlags::LOAD_STORE_BYTE_SWAP) {
       assert_false(i.src2.is_constant);
-      e.movbe(e.dword[addr], i.src2);
+      if (e.IsFeatureEnabled(kX64EmitMovbe)) {
+        e.movbe(e.dword[addr], i.src2);
+      } else {
+        assert_always("not implemented");
+      }
     } else {
       if (i.src2.is_constant) {
         e.mov(e.dword[addr], i.src2.constant());
@@ -1687,7 +1710,11 @@ EMITTER(STORE_I64, MATCH(I<OPCODE_STORE, VoidOp, I64<>, I64<>>)) {
     auto addr = ComputeMemoryAddress(e, i.src1);
     if (i.instr->flags & LoadStoreFlags::LOAD_STORE_BYTE_SWAP) {
       assert_false(i.src2.is_constant);
-      e.movbe(e.qword[addr], i.src2);
+      if (e.IsFeatureEnabled(kX64EmitMovbe)) {
+        e.movbe(e.qword[addr], i.src2);
+      } else {
+        assert_always("not implemented");
+      }
     } else {
       if (i.src2.is_constant) {
         e.MovMem64(addr, i.src2.constant());
diff --git a/src/xenia/cpu/frontend/ppc_translator.cc b/src/xenia/cpu/frontend/ppc_translator.cc
index e962839ef..2e51280ee 100644
--- a/src/xenia/cpu/frontend/ppc_translator.cc
+++ b/src/xenia/cpu/frontend/ppc_translator.cc
@@ -60,8 +60,12 @@ PPCTranslator::PPCTranslator(PPCFrontend* frontend) : frontend_(frontend) {
   if (validate) compiler_->AddPass(std::make_unique<passes::ValidationPass>());
   compiler_->AddPass(std::make_unique<passes::ConstantPropagationPass>());
   if (validate) compiler_->AddPass(std::make_unique<passes::ValidationPass>());
-  compiler_->AddPass(std::make_unique<passes::MemorySequenceCombinationPass>());
-  if (validate) compiler_->AddPass(std::make_unique<passes::ValidationPass>());
+  if (backend->machine_info()->supports_extended_load_store) {
+    // Backend supports the advanced LOAD/STORE instructions.
+    // These will save us a lot of HIR opcodes.
+    compiler_->AddPass(std::make_unique<passes::MemorySequenceCombinationPass>());
+    if (validate) compiler_->AddPass(std::make_unique<passes::ValidationPass>());
+  }
   compiler_->AddPass(std::make_unique<passes::SimplificationPass>());
   if (validate) compiler_->AddPass(std::make_unique<passes::ValidationPass>());
   // compiler_->AddPass(std::make_unique<passes::DeadStoreEliminationPass>());