diff --git a/Source/Core/Common/Arm64Emitter.cpp b/Source/Core/Common/Arm64Emitter.cpp
index 4cb4086d7c..355275caf8 100644
--- a/Source/Core/Common/Arm64Emitter.cpp
+++ b/Source/Core/Common/Arm64Emitter.cpp
@@ -2874,7 +2874,7 @@ void ARM64FloatEmitter::FMUL(u8 size, ARM64Reg Rd, ARM64Reg Rn, ARM64Reg Rm, u8
 	EmitVectorxElement(0, 2 | (size >> 6), L, 0b1001, H, Rd, Rn, Rm);
 }
 
-void ARM64FloatEmitter::ABI_PushRegisters(BitSet32 registers)
+void ARM64FloatEmitter::ABI_PushRegisters(BitSet32 registers, ARM64Reg tmp)
 {
 	bool bundled_loadstore = false;
 	int num_regs = registers.Count();
@@ -2904,15 +2904,10 @@ void ARM64FloatEmitter::ABI_PushRegisters(BitSet32 registers)
 		}
 	}
 
-	if (!bundled_loadstore)
+	if (bundled_loadstore && tmp != INVALID_REG)
 	{
-		for (auto it : registers)
-			STR(128, INDEX_PRE, (ARM64Reg)(Q0 + it), SP, -16);
-	}
-	else
-	{
-		// Violating the AAPCS64 never felt so right.
 		m_emit->SUB(SP, SP, num_regs * 16);
+		m_emit->ADD(tmp, SP, 0);
 		for (int i = 0; i < 32; ++i)
 		{
 			if (!registers[i])
@@ -2927,14 +2922,18 @@ void ARM64FloatEmitter::ABI_PushRegisters(BitSet32 registers)
 			// 4 < 4 && registers[i + 4] false!
 			while (++count < 4 && (i + count) < 32 && registers[i + count]) {}
 
-			ST1(64, count, INDEX_POST, (ARM64Reg)(Q0 + i), SP);
+			ST1(64, count, INDEX_POST, (ARM64Reg)(Q0 + i), tmp);
 
 			i += count - 1;
 		}
-		m_emit->SUB(SP, SP, num_regs * 16);
+	}
+	else
+	{
+		for (auto it : registers)
+			STR(128, INDEX_PRE, (ARM64Reg)(Q0 + it), SP, -16);
 	}
 }
-void ARM64FloatEmitter::ABI_PopRegisters(BitSet32 registers)
+void ARM64FloatEmitter::ABI_PopRegisters(BitSet32 registers, ARM64Reg tmp)
 {
 	bool bundled_loadstore = false;
 	int num_regs = registers.Count();
@@ -2964,18 +2963,9 @@ void ARM64FloatEmitter::ABI_PopRegisters(BitSet32 registers)
 		}
 	}
 
-	if (!bundled_loadstore)
-	{
-		for (int i = 31; i >= 0; --i)
-		{
-			if (!registers[i])
-				continue;
-
-			LDR(128, INDEX_POST, (ARM64Reg)(Q0 + i), SP, 16);
-		}
-	}
-	else
+	if (bundled_loadstore && tmp != INVALID_REG)
 	{
+		// The temporary register is only used to indicate that we can use this code path
 		for (int i = 0; i < 32; ++i)
 		{
 			if (!registers[i])
@@ -2989,6 +2979,16 @@ void ARM64FloatEmitter::ABI_PopRegisters(BitSet32 registers)
 			i += count - 1;
 		}
 	}
+	else
+	{
+		for (int i = 31; i >= 0; --i)
+		{
+			if (!registers[i])
+				continue;
+
+			LDR(128, INDEX_POST, (ARM64Reg)(Q0 + i), SP, 16);
+		}
+	}
 }
 
 }
diff --git a/Source/Core/Common/Arm64Emitter.h b/Source/Core/Common/Arm64Emitter.h
index 3fd3d390c7..ab933225c5 100644
--- a/Source/Core/Common/Arm64Emitter.h
+++ b/Source/Core/Common/Arm64Emitter.h
@@ -755,8 +755,8 @@ public:
 	void FMUL(u8 size, ARM64Reg Rd, ARM64Reg Rn, ARM64Reg Rm, u8 index);
 
 	// ABI related
-	void ABI_PushRegisters(BitSet32 registers);
-	void ABI_PopRegisters(BitSet32 registers);
+	void ABI_PushRegisters(BitSet32 registers, ARM64Reg tmp = INVALID_REG);
+	void ABI_PopRegisters(BitSet32 registers, ARM64Reg tmp = INVALID_REG);
 
 private:
 	ARM64XEmitter* m_emit;
diff --git a/Source/Core/Core/PowerPC/JitArm64/JitAsm.cpp b/Source/Core/Core/PowerPC/JitArm64/JitAsm.cpp
index 62cce8374b..e82367873b 100644
--- a/Source/Core/Core/PowerPC/JitArm64/JitAsm.cpp
+++ b/Source/Core/Core/PowerPC/JitArm64/JitAsm.cpp
@@ -279,12 +279,12 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 		SetJumpTarget(argh);
 
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.UMOV(64, X0, Q0, 0);
 		ORR(X0, SP, X0, ArithOption(X0, ST_ROR, 32));
 		MOVI2R(X30, (u64)PowerPC::Write_U64);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -310,12 +310,12 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.UMOV(16, W0, Q0, 0);
 		REV16(W0, W0);
 		MOVI2R(X30, (u64)PowerPC::Write_U16);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -341,12 +341,12 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.UMOV(16, W0, Q0, 0);
 		REV16(W0, W0);
 		MOVI2R(X30, (u64)PowerPC::Write_U16);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -372,12 +372,12 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.REV32(8, D0, D0);
 		float_emit.UMOV(32, W0, Q0, 0);
 		MOVI2R(X30, (u64)PowerPC::Write_U32);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -402,12 +402,12 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.REV32(8, D0, D0);
 		float_emit.UMOV(32, W0, Q0, 0);
 		MOVI2R(X30, (u64)PowerPC::Write_U32);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -428,11 +428,11 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 		SetJumpTarget(argh);
 
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.UMOV(32, W0, Q0, 0);
 		MOVI2R(X30, (u64)&PowerPC::Write_U32);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -457,11 +457,11 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.UMOV(32, W0, Q0, 0);
 		MOVI2R(X30, (u64)&PowerPC::Write_U8);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -486,11 +486,11 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.SMOV(32, W0, Q0, 0);
 		MOVI2R(X30, (u64)&PowerPC::Write_U8);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -515,11 +515,11 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.UMOV(32, W0, Q0, 0);
 		MOVI2R(X30, (u64)&PowerPC::Write_U16);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}
@@ -544,11 +544,11 @@ void JitArm64AsmRoutineManager::GenerateCommon()
 
 		SetJumpTarget(argh);
 		ABI_PushRegisters(gprs);
-		float_emit.ABI_PushRegisters(fprs);
+		float_emit.ABI_PushRegisters(fprs, X3);
 		float_emit.SMOV(32, W0, Q0, 0);
 		MOVI2R(X30, (u64)&PowerPC::Write_U16);
 		BLR(X30);
-		float_emit.ABI_PopRegisters(fprs);
+		float_emit.ABI_PopRegisters(fprs, X3);
 		ABI_PopRegisters(gprs);
 		RET(X30);
 	}