diff --git a/src/core/cpu_manager.cpp b/src/core/cpu_manager.cpp
index 4ad9cc69d5..2cbbf7010c 100644
--- a/src/core/cpu_manager.cpp
+++ b/src/core/cpu_manager.cpp
@@ -104,12 +104,10 @@ void CpuManager::MultiCoreRunGuestThread() {
     kernel.CurrentScheduler()->OnThreadStart();
 
     while (true) {
-        auto* physical_core = &kernel.CurrentPhysicalCore();
-        while (!physical_core->IsInterrupted()) {
-            physical_core->RunThread(thread);
-            physical_core = &kernel.CurrentPhysicalCore();
+        auto& physical_core = kernel.CurrentPhysicalCore();
+        while (!physical_core.IsInterrupted()) {
+            physical_core.RunThread(thread);
         }
-
         HandleInterrupt();
     }
 }
diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp
index 5c2b6a4a88..91c6f8e853 100644
--- a/src/core/hle/kernel/kernel.cpp
+++ b/src/core/hle/kernel/kernel.cpp
@@ -355,19 +355,8 @@ struct KernelCore::Impl {
         application_process->Open();
     }
 
-    /// Sets the host thread ID for the caller.
-    u32 SetHostThreadId(std::size_t core_id) {
-        // This should only be called during core init.
-        ASSERT(tls_data.host_thread_id == UINT8_MAX);
-
-        // The first four slots are reserved for CPU core threads
-        ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
-        tls_data.host_thread_id = u8(core_id);
-        return tls_data.host_thread_id;
-    }
-
     /// Gets the host thread ID for the caller
-    u32 GetHostThreadId() const {
+    [[nodiscard]] inline u32 GetHostThreadId() const noexcept {
         return tls_data.host_thread_id;
     }
 
@@ -386,9 +375,12 @@ struct KernelCore::Impl {
     }
 
     /// Registers a CPU core thread by allocating a host thread ID for it
-    void RegisterCoreThread(std::size_t core_id) {
+    void RegisterCoreThread(std::size_t core_id) noexcept {
+        // This should only be called during core init.
+        ASSERT(tls_data.host_thread_id == UINT8_MAX);
+        // The first four slots are reserved for CPU core threads
         ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
-        const auto this_id = SetHostThreadId(core_id);
+        const auto this_id = tls_data.host_thread_id = u8(core_id);
         if (!is_multicore)
             single_core_thread_id = this_id;
     }
@@ -398,7 +390,7 @@ struct KernelCore::Impl {
         (void)GetHostDummyThread(existing_thread);
     }
 
-    [[nodiscard]] u32 GetCurrentHostThreadID() {
+    [[nodiscard]] inline u32 GetCurrentHostThreadID() noexcept {
         auto const this_id = GetHostThreadId();
         if (!is_multicore && single_core_thread_id == this_id)
             return u32(system.GetCpuManager().CurrentCore());
@@ -930,11 +922,7 @@ const Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) const {
 }
 
 size_t KernelCore::CurrentPhysicalCoreIndex() const {
-    const u32 core_id = impl->GetCurrentHostThreadID();
-    if (core_id >= Core::Hardware::NUM_CPU_CORES) {
-        return Core::Hardware::NUM_CPU_CORES - 1;
-    }
-    return core_id;
+    return impl->GetCurrentHostThreadID();
 }
 
 Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() {
diff --git a/src/core/hle/kernel/physical_core.cpp b/src/core/hle/kernel/physical_core.cpp
index e083ee1b0a..59b6c32b16 100644
--- a/src/core/hle/kernel/physical_core.cpp
+++ b/src/core/hle/kernel/physical_core.cpp
@@ -207,7 +207,7 @@ void PhysicalCore::Idle() {
     m_on_interrupt.wait(lk, [this] { return m_is_interrupted; });
 }
 
-bool PhysicalCore::IsInterrupted() const {
+[[noinline]] bool PhysicalCore::IsInterrupted() const {
     return m_is_interrupted;
 }
 
diff --git a/src/dynarmic/src/dynarmic/backend/x64/a32_interface.cpp b/src/dynarmic/src/dynarmic/backend/x64/a32_interface.cpp
index 5afc41b864..7260542c3b 100644
--- a/src/dynarmic/src/dynarmic/backend/x64/a32_interface.cpp
+++ b/src/dynarmic/src/dynarmic/backend/x64/a32_interface.cpp
@@ -120,11 +120,13 @@ struct Jit::Impl {
     }
 
     void HaltExecution(HaltReason hr) {
-        Atomic::Or(&jit_state.halt_reason, static_cast<u32>(hr));
+        Atomic::Or(&jit_state.halt_reason, u32(hr));
+        Atomic::Barrier();
     }
 
     void ClearHalt(HaltReason hr) {
-        Atomic::And(&jit_state.halt_reason, ~static_cast<u32>(hr));
+        Atomic::And(&jit_state.halt_reason, ~u32(hr));
+        Atomic::Barrier();
     }
 
     void ClearExclusiveState() {
diff --git a/src/dynarmic/src/dynarmic/backend/x64/a64_interface.cpp b/src/dynarmic/src/dynarmic/backend/x64/a64_interface.cpp
index b6240fdb89..e1c29feb48 100644
--- a/src/dynarmic/src/dynarmic/backend/x64/a64_interface.cpp
+++ b/src/dynarmic/src/dynarmic/backend/x64/a64_interface.cpp
@@ -122,10 +122,12 @@ public:
 
     void HaltExecution(HaltReason hr) {
         Atomic::Or(&jit_state.halt_reason, u32(hr));
+        Atomic::Barrier();
     }
 
     void ClearHalt(HaltReason hr) {
         Atomic::And(&jit_state.halt_reason, ~u32(hr));
+        Atomic::Barrier();
     }
 
     u64 GetSP() const {