diff --git a/include/physics.hpp b/include/physics.hpp
index ca4a8a3..56f59a1 100644
--- a/include/physics.hpp
+++ b/include/physics.hpp
@@ -2,6 +2,7 @@
 #define __PHYSICS_HPP_
 
 #include <atomic>
+#include <condition_variable>
 #include <mutex>
 #include <queue>
 #include <raylib.h>
@@ -150,7 +151,11 @@ class ThreadedPhysics {
     TracyLockable(std::mutex, command_mtx);
     std::queue<Command> pending_commands;
 
-    TracyLockable(std::mutex, pos_mtx);
+    TracyLockable(std::mutex, data_mtx);
+    std::condition_variable_any data_ready_cnd;
+    std::condition_variable_any data_consumed_cnd;
+    bool data_ready = false;
+    bool data_consumed = true;
     std::vector<Mass> masses;                    // Read by renderer
     std::unordered_map<State, int> state_masses; // Read by renderer
     std::vector<Spring> springs;                 // Read by renderer
diff --git a/include/renderer.hpp b/include/renderer.hpp
index 642d391..334d350 100644
--- a/include/renderer.hpp
+++ b/include/renderer.hpp
@@ -61,10 +61,9 @@ public:
   }
 
 private:
-  auto AllocateGraphInstancing(const std::vector<Mass> &masses) -> void;
+  auto AllocateGraphInstancing(std::size_t size) -> void;
 
-  auto ReallocateGraphInstancingIfNecessary(const std::vector<Mass> &masses)
-      -> void;
+  auto ReallocateGraphInstancingIfNecessary(std::size_t size) -> void;
 
 public:
   auto UpdateTextureSizes() -> void;
diff --git a/src/main.cpp b/src/main.cpp
index fc8c98b..7ca8fe9 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -81,12 +81,19 @@ auto main(int argc, char *argv[]) -> int {
     // TODO: Don't download each frame if nothing changed, check for update
     //       first
     {
-      std::lock_guard<LockableBase(std::mutex)> lock(physics.state.pos_mtx);
+      std::unique_lock<LockableBase(std::mutex)> lock(physics.state.data_mtx);
+      physics.state.data_ready_cnd.wait(
+          lock, [&] { return physics.state.data_ready; });
+
       masses = physics.state.masses;
       state_masses = physics.state.state_masses;
       springs = physics.state.springs;
       state_springs = physics.state.state_springs;
+
+      physics.state.data_ready = false;
+      physics.state.data_consumed = true;
     }
+    physics.state.data_consumed_cnd.notify_one();
 
     // Update the camera after the physics, so target lock is smooth
     if (state_masses.contains(state.current_state)) {
diff --git a/src/physics.cpp b/src/physics.cpp
index bbac526..db590cb 100644
--- a/src/physics.cpp
+++ b/src/physics.cpp
@@ -349,7 +349,7 @@ auto ThreadedPhysics::PhysicsThread(ThreadedPhysics::PhysicsState &state)
       [&](const struct ClearGraph &cg) { mass_springs.Clear(); },
   };
 
-  while (state.running) {
+  while (state.running.load()) {
     // Handle queued commands
     {
       std::lock_guard<LockableBase(std::mutex)> lock(state.command_mtx);
@@ -366,7 +366,6 @@ auto ThreadedPhysics::PhysicsThread(ThreadedPhysics::PhysicsState &state)
     }
 
     // Physics update
-    // TODO: I need this thread to run at a constant rate
     mass_springs.ClearForces();
     mass_springs.CalculateSpringForces();
     mass_springs.CalculateRepulsionForces();
@@ -380,12 +379,20 @@ auto ThreadedPhysics::PhysicsThread(ThreadedPhysics::PhysicsState &state)
 
     // Publish the positions for the renderer (copy)
     {
-      std::lock_guard<LockableBase(std::mutex)> lock(state.pos_mtx);
+      std::unique_lock<LockableBase(std::mutex)> lock(state.data_mtx);
+      state.data_consumed_cnd.wait(lock, [&] { return state.data_consumed; });
+
       state.masses = mass_springs.masses;
       state.state_masses = mass_springs.state_masses;
       state.springs = mass_springs.springs;
       state.state_springs = mass_springs.state_springs;
+
+      state.data_ready = true;
+      state.data_consumed = false;
     }
+
+    // Notify the rendering thread that new data is available
+    state.data_ready_cnd.notify_one();
   }
 }
 
diff --git a/src/renderer.cpp b/src/renderer.cpp
index 725395e..21b0508 100644
--- a/src/renderer.cpp
+++ b/src/renderer.cpp
@@ -33,8 +33,7 @@ auto Renderer::UpdateTextureSizes() -> void {
   menu_target = LoadRenderTexture(width * 2, MENU_HEIGHT);
 }
 
-auto Renderer::AllocateGraphInstancing(const std::vector<Mass> &masses)
-    -> void {
+auto Renderer::AllocateGraphInstancing(std::size_t size) -> void {
   cube_instance = GenMeshCube(VERTEX_SIZE, VERTEX_SIZE, VERTEX_SIZE);
 
   instancing_shader = LoadShader("shader/instancing_vertex.glsl",
@@ -48,16 +47,14 @@ auto Renderer::AllocateGraphInstancing(const std::vector<Mass> &masses)
   vertex_mat.maps[MATERIAL_MAP_DIFFUSE].color = VERTEX_COLOR;
   vertex_mat.shader = instancing_shader;
 
-  transforms = (Matrix *)MemAlloc(masses.size() * sizeof(Matrix));
-  transforms_size = masses.size();
+  transforms = (Matrix *)MemAlloc(size * sizeof(Matrix));
+  transforms_size = size;
 }
 
-auto Renderer::ReallocateGraphInstancingIfNecessary(
-    const std::vector<Mass> &masses) -> void {
-  if (transforms_size != masses.size()) {
-    transforms =
-        (Matrix *)MemRealloc(transforms, masses.size() * sizeof(Matrix));
-    transforms_size = masses.size();
+auto Renderer::ReallocateGraphInstancingIfNecessary(std::size_t size) -> void {
+  if (transforms_size != size) {
+    transforms = (Matrix *)MemRealloc(transforms, size * sizeof(Matrix));
+    transforms_size = size;
   }
 }
 
@@ -76,9 +73,9 @@ auto Renderer::DrawMassSprings(
     ZoneNamedN(prepare_masses, "PrepareMasses", true);
     if (masses.size() < DRAW_VERTICES_LIMIT) {
       if (transforms == nullptr) {
-        AllocateGraphInstancing(masses);
+        AllocateGraphInstancing(masses.size());
       }
-      ReallocateGraphInstancingIfNecessary(masses);
+      ReallocateGraphInstancingIfNecessary(masses.size());
 
       int i = 0;
       for (const auto &mass : masses) {
diff --git a/src/state.cpp b/src/state.cpp
index 59f3d7a..f54628e 100644
--- a/src/state.cpp
+++ b/src/state.cpp
@@ -43,7 +43,7 @@ auto StateManager::FillGraph() -> void {
   // TODO: We have only dispatched the commands, the states won't be downloaded
   //       when calling this... Make FindWinningStates() another command?
   //       Or recalculate whenever masses.size() changes?
-  FindWinningStates();
+  // FindWinningStates();
 }
 
 auto StateManager::UpdateGraph() -> void {
@@ -74,7 +74,7 @@ auto StateManager::FindWinningStates() -> void {
   winning_states.clear();
   std::unordered_map<State, int> state_masses;
   {
-    std::lock_guard<LockableBase(std::mutex)> lock(physics.state.pos_mtx);
+    std::lock_guard<LockableBase(std::mutex)> lock(physics.state.data_mtx);
     state_masses = physics.state.state_masses;
   }
   for (const auto &[state, mass] : state_masses) {