diff --git a/data/multibody.bullet b/data/multibody.bullet
index 028300994..32b9e7f37 100644
Binary files a/data/multibody.bullet and b/data/multibody.bullet differ
diff --git a/examples/Benchmarks/BenchmarkDemo.cpp b/examples/Benchmarks/BenchmarkDemo.cpp
index c0130dd86..3a54bbb74 100644
--- a/examples/Benchmarks/BenchmarkDemo.cpp
+++ b/examples/Benchmarks/BenchmarkDemo.cpp
@@ -1008,7 +1008,7 @@ void BenchmarkDemo::createTest4()
 	}
 
 	//this will enable polyhedral contact clipping, better quality, slightly slower
-	//convexHullShape->initializePolyhedralFeatures();
+	convexHullShape->initializePolyhedralFeatures();
 
 	btTransform trans;
 	trans.setIdentity();
diff --git a/examples/ExampleBrowser/OpenGLExampleBrowser.cpp b/examples/ExampleBrowser/OpenGLExampleBrowser.cpp
index 1ed8c7a37..d5c9ccac5 100644
--- a/examples/ExampleBrowser/OpenGLExampleBrowser.cpp
+++ b/examples/ExampleBrowser/OpenGLExampleBrowser.cpp
@@ -251,7 +251,6 @@ void MyKeyboardCallback(int key, int state)
 
 		if (key == 'p')
 		{
-#ifndef BT_NO_PROFILE
 			if (state)
 			{
 				b3ChromeUtilsStartTimings();
@@ -260,7 +259,6 @@ void MyKeyboardCallback(int key, int state)
 			{
 				b3ChromeUtilsStopTimingsAndWriteJsonFile("timings");
 			}
-#endif  //BT_NO_PROFILE
 		}
 
 #ifndef NO_OPENGL3
@@ -1129,6 +1127,7 @@ bool OpenGLExampleBrowser::init(int argc, char* argv[])
 
 		gui2->registerFileOpenCallback(fileOpenCallback);
 		gui2->registerQuitCallback(quitCallback);
+		
 	}
 
 	return true;
diff --git a/examples/SharedMemory/PhysicsServerExample.cpp b/examples/SharedMemory/PhysicsServerExample.cpp
index 93292ee89..36ec1f18b 100644
--- a/examples/SharedMemory/PhysicsServerExample.cpp
+++ b/examples/SharedMemory/PhysicsServerExample.cpp
@@ -1737,6 +1737,11 @@ void PhysicsServerExample::initPhysics()
 		m_args[w].m_cs2 = m_threadSupport->createCriticalSection();
 		m_args[w].m_cs3 = m_threadSupport->createCriticalSection();
 		m_args[w].m_csGUI = m_threadSupport->createCriticalSection();
+		m_multiThreadedHelper->setCriticalSection(m_args[w].m_cs);
+		m_multiThreadedHelper->setCriticalSection2(m_args[w].m_cs2);
+		m_multiThreadedHelper->setCriticalSection3(m_args[w].m_cs3);
+		m_multiThreadedHelper->setCriticalSectionGUI(m_args[w].m_csGUI);
+
 		m_args[w].m_cs->lock();
 		m_args[w].m_cs->setSharedParam(0, eMotionIsUnInitialized);
 		m_args[w].m_cs->unlock();
@@ -1759,12 +1764,9 @@ void PhysicsServerExample::initPhysics()
 		}
 	}
 
+	m_args[0].m_cs->lock();
 	m_args[0].m_cs->setSharedParam(1, eGUIHelperIdle);
-	m_multiThreadedHelper->setCriticalSection(m_args[0].m_cs);
-	m_multiThreadedHelper->setCriticalSection2(m_args[0].m_cs2);
-	m_multiThreadedHelper->setCriticalSection3(m_args[0].m_cs3);
-	m_multiThreadedHelper->setCriticalSectionGUI(m_args[0].m_csGUI);
-
+	m_args[0].m_cs->unlock();
 	m_args[0].m_cs2->lock();
 
 	{
diff --git a/examples/SharedMemory/SharedMemoryPublic.h b/examples/SharedMemory/SharedMemoryPublic.h
index 166de101c..eb74d3f3c 100644
--- a/examples/SharedMemory/SharedMemoryPublic.h
+++ b/examples/SharedMemory/SharedMemoryPublic.h
@@ -928,7 +928,7 @@ enum eFileIOTypes
 };
 
 //limits for vertices/indices in PyBullet::createCollisionShape
-#define B3_MAX_NUM_VERTICES 1024
-#define B3_MAX_NUM_INDICES 1024
+#define B3_MAX_NUM_VERTICES 16
+#define B3_MAX_NUM_INDICES 16
 
 #endif  //SHARED_MEMORY_PUBLIC_H
diff --git a/examples/Utils/ChromeTraceUtil.cpp b/examples/Utils/ChromeTraceUtil.cpp
index 2a25678ed..78d478162 100644
--- a/examples/Utils/ChromeTraceUtil.cpp
+++ b/examples/Utils/ChromeTraceUtil.cpp
@@ -174,7 +174,7 @@ void MyEnterProfileZoneFunc(const char* msg)
 {
 	if (gProfileDisabled)
 		return;
-#ifndef BT_NO_PROFILE
+
 	int threadId = btQuickprofGetCurrentThreadIndex2();
 	if (threadId < 0 || threadId >= BT_QUICKPROF_MAX_THREAD_COUNT)
 		return;
@@ -191,13 +191,13 @@ void MyEnterProfileZoneFunc(const char* msg)
 		gStartTimes[threadId][gStackDepths[threadId]] = 1 + gStartTimes[threadId][gStackDepths[threadId] - 1];
 	}
 	gStackDepths[threadId]++;
-#endif
+
 }
 void MyLeaveProfileZoneFunc()
 {
 	if (gProfileDisabled)
 		return;
-#ifndef BT_NO_PROFILE
+
 	int threadId = btQuickprofGetCurrentThreadIndex2();
 	if (threadId < 0 || threadId >= BT_QUICKPROF_MAX_THREAD_COUNT)
 		return;
@@ -214,7 +214,7 @@ void MyLeaveProfileZoneFunc()
 
 	unsigned long long int endTime = clk.getTimeNanoseconds();
 	gTimings[threadId].addTiming(name, threadId, startTime, endTime);
-#endif  //BT_NO_PROFILE
+
 }
 
 void b3ChromeUtilsStartTimings()
diff --git a/examples/pybullet/examples/createObstacleCourse.py b/examples/pybullet/examples/createObstacleCourse.py
index 1194a213a..ef145c72c 100644
--- a/examples/pybullet/examples/createObstacleCourse.py
+++ b/examples/pybullet/examples/createObstacleCourse.py
@@ -21,28 +21,6 @@ colSphereId = p.createCollisionShape(p.GEOM_SPHERE,radius=sphereRadius)
 #convex mesh from obj
 stoneId = p.createCollisionShape(p.GEOM_MESH,fileName="stone.obj")
 
-#concave mesh from obj
-stoneId = p.createCollisionShape(p.GEOM_MESH,fileName="stone.obj", flags=p.GEOM_FORCE_CONCAVE_TRIMESH)
-
-
-verts=[[-0.246350, -0.246483, -0.000624],
-	[ -0.151407, -0.176325, 0.172867],
-	[ -0.246350, 0.249205, -0.000624],
-	[ -0.151407, 0.129477, 0.172867],
-	[ 0.249338, -0.246483, -0.000624],
-	[ 0.154395, -0.176325, 0.172867],
-	[ 0.249338, 0.249205, -0.000624],
-	[ 0.154395, 0.129477, 0.172867]]
-#convex mesh from vertices
-stoneConvexId = p.createCollisionShape(p.GEOM_MESH,vertices=verts)
-
-indices=[0,3,2,3,6,2,7,4,6,5,0,4,6,0,2,3,5,7,0,1,3,3,7,6,7,5,4,5,1,0,6,4,0,3,1,5]
-
-#concave mesh from vertices+indices
-stoneConcaveId = p.createCollisionShape(p.GEOM_MESH,vertices=verts, indices=indices)
-
-stoneId = stoneConvexId
-#stoneId = stoneConcaveId
 
 
 boxHalfLength = 0.5
diff --git a/examples/pybullet/gym/pybullet_envs/ARS/ars.py b/examples/pybullet/gym/pybullet_envs/ARS/ars.py
new file mode 100644
index 000000000..b4c0c3127
--- /dev/null
+++ b/examples/pybullet/gym/pybullet_envs/ARS/ars.py
@@ -0,0 +1,142 @@
+# AI 2018
+
+# Importing the libraries
+import os
+import numpy as np
+import gym
+from gym import wrappers
+import pybullet_envs
+
+# Setting the Hyper Parameters
+
+class Hp():
+    
+    def __init__(self):
+        self.nb_steps = 1000
+        self.episode_length = 1000
+        self.learning_rate = 0.02
+        self.nb_directions = 16
+        self.nb_best_directions = 16
+        assert self.nb_best_directions <= self.nb_directions
+        self.noise = 0.03
+        self.seed = 1
+        self.env_name = 'HalfCheetahBulletEnv-v0'
+
+# Normalizing the states
+
+class Normalizer():
+    
+    def __init__(self, nb_inputs):
+        self.n = np.zeros(nb_inputs)
+        self.mean = np.zeros(nb_inputs)
+        self.mean_diff = np.zeros(nb_inputs)
+        self.var = np.zeros(nb_inputs)
+    
+    def observe(self, x):
+        self.n += 1.
+        last_mean = self.mean.copy()
+        self.mean += (x - self.mean) / self.n
+        self.mean_diff += (x - last_mean) * (x - self.mean)
+        self.var = (self.mean_diff / self.n).clip(min = 1e-2)
+    
+    def normalize(self, inputs):
+        obs_mean = self.mean
+        obs_std = np.sqrt(self.var)
+        return (inputs - obs_mean) / obs_std
+
+# Building the AI
+
+class Policy():
+    
+    def __init__(self, input_size, output_size):
+        self.theta = np.zeros((output_size, input_size))
+        print("self.theta=",self.theta)    
+    def evaluate(self, input, delta = None, direction = None):
+        if direction is None:
+            return np.clip(self.theta.dot(input), -1.0, 1.0)
+        elif direction == "positive":
+            return np.clip((self.theta + hp.noise*delta).dot(input), -1.0, 1.0)
+        else:
+            return np.clip((self.theta - hp.noise*delta).dot(input), -1.0, 1.0)
+    
+    def sample_deltas(self):
+        return [np.random.randn(*self.theta.shape) for _ in range(hp.nb_directions)]
+    
+    def update(self, rollouts, sigma_r):
+        step = np.zeros(self.theta.shape)
+        for r_pos, r_neg, d in rollouts:
+            step += (r_pos - r_neg) * d
+        self.theta += hp.learning_rate / (hp.nb_best_directions * sigma_r) * step
+
+# Exploring the policy on one specific direction and over one episode
+
+def explore(env, normalizer, policy, direction = None, delta = None):
+    state = env.reset()
+    done = False
+    num_plays = 0.
+    sum_rewards = 0
+    while not done and num_plays < hp.episode_length:
+        normalizer.observe(state)
+        state = normalizer.normalize(state)
+        action = policy.evaluate(state, delta, direction)
+        state, reward, done, _ = env.step(action)
+        reward = max(min(reward, 1), -1)
+        sum_rewards += reward
+        num_plays += 1
+    return sum_rewards
+
+# Training the AI
+
+def train(env, policy, normalizer, hp):
+    
+    for step in range(hp.nb_steps):
+        
+        # Initializing the perturbations deltas and the positive/negative rewards
+        deltas = policy.sample_deltas()
+        positive_rewards = [0] * hp.nb_directions
+        negative_rewards = [0] * hp.nb_directions
+        
+        # Getting the positive rewards in the positive directions
+        for k in range(hp.nb_directions):
+            positive_rewards[k] = explore(env, normalizer, policy, direction = "positive", delta = deltas[k])
+        
+        # Getting the negative rewards in the negative/opposite directions
+        for k in range(hp.nb_directions):
+            negative_rewards[k] = explore(env, normalizer, policy, direction = "negative", delta = deltas[k])
+        
+        # Gathering all the positive/negative rewards to compute the standard deviation of these rewards
+        all_rewards = np.array(positive_rewards + negative_rewards)
+        sigma_r = all_rewards.std()
+        
+        # Sorting the rollouts by the max(r_pos, r_neg) and selecting the best directions
+        scores = {k:max(r_pos, r_neg) for k,(r_pos,r_neg) in enumerate(zip(positive_rewards, negative_rewards))}
+        order = sorted(scores.keys(), key = lambda x:scores[x])[:hp.nb_best_directions]
+        rollouts = [(positive_rewards[k], negative_rewards[k], deltas[k]) for k in order]
+        
+        # Updating our policy
+        policy.update(rollouts, sigma_r)
+        
+        # Printing the final reward of the policy after the update
+        reward_evaluation = explore(env, normalizer, policy)
+        print('Step:', step, 'Reward:', reward_evaluation)
+
+# Running the main code
+
+def mkdir(base, name):
+    path = os.path.join(base, name)
+    if not os.path.exists(path):
+        os.makedirs(path)
+    return path
+work_dir = mkdir('exp', 'brs')
+monitor_dir = mkdir(work_dir, 'monitor')
+
+hp = Hp()
+np.random.seed(hp.seed)
+env = gym.make(hp.env_name)
+# env.render(mode = "human")
+#env = wrappers.Monitor(env, monitor_dir, force = True)
+nb_inputs = env.observation_space.shape[0]
+nb_outputs = env.action_space.shape[0]
+policy = Policy(nb_inputs, nb_outputs)
+normalizer = Normalizer(nb_inputs)
+train(env, policy, normalizer, hp)
diff --git a/examples/pybullet/gym/pybullet_utils/examples/mjcf2urdf.py b/examples/pybullet/gym/pybullet_utils/examples/mjcf2urdf.py
index a4fc7ca0e..c1e39745d 100644
--- a/examples/pybullet/gym/pybullet_utils/examples/mjcf2urdf.py
+++ b/examples/pybullet/gym/pybullet_utils/examples/mjcf2urdf.py
@@ -14,7 +14,7 @@ p.setAdditionalSearchPath(pd.getDataPath())
 objs = p.loadMJCF(args.mjcf, flags=p.URDF_USE_IMPLICIT_CYLINDER)
 
 for o in objs:
-	print("o=",o, p.getBodyInfo(o), p.getNumJoints(o))
+	#print("o=",o, p.getBodyInfo(o), p.getNumJoints(o))
 	humanoid = objs[o]
 	ed0 = ed.UrdfEditor()
 	ed0.initializeFromBulletBody(humanoid, p._client)
diff --git a/src/Bullet3Common/b3AlignedAllocator.cpp b/src/Bullet3Common/b3AlignedAllocator.cpp
index 6e54c5fb7..d546d5e06 100644
--- a/src/Bullet3Common/b3AlignedAllocator.cpp
+++ b/src/Bullet3Common/b3AlignedAllocator.cpp
@@ -15,9 +15,11 @@ subject to the following restrictions:
 
 #include "b3AlignedAllocator.h"
 
+#ifdef B3_ALLOCATOR_STATISTICS
 int b3g_numAlignedAllocs = 0;
 int b3g_numAlignedFree = 0;
 int b3g_totalBytesAlignedAllocs = 0;  //detect memory leaks
+#endif
 
 static void *b3AllocDefault(size_t size)
 {
@@ -109,10 +111,10 @@ void *b3AlignedAllocInternal(size_t size, int alignment, int line, char *filenam
 {
 	void *ret;
 	char *real;
-
+#ifdef B3_ALLOCATOR_STATISTICS
 	b3g_totalBytesAlignedAllocs += size;
 	b3g_numAlignedAllocs++;
-
+#endif
 	real = (char *)b3s_allocFunc(size + 2 * sizeof(void *) + (alignment - 1));
 	if (real)
 	{
@@ -135,14 +137,16 @@ void *b3AlignedAllocInternal(size_t size, int alignment, int line, char *filenam
 void b3AlignedFreeInternal(void *ptr, int line, char *filename)
 {
 	void *real;
+#ifdef B3_ALLOCATOR_STATISTICS
 	b3g_numAlignedFree++;
-
+#endif
 	if (ptr)
 	{
 		real = *((void **)(ptr)-1);
 		int size = *((int *)(ptr)-2);
+#ifdef B3_ALLOCATOR_STATISTICS
 		b3g_totalBytesAlignedAllocs -= size;
-
+#endif
 		b3Printf("free #%d at address %x, from %s,line %d, size %d\n", b3g_numAlignedFree, real, filename, line, size);
 
 		b3s_freeFunc(real);
@@ -157,7 +161,9 @@ void b3AlignedFreeInternal(void *ptr, int line, char *filename)
 
 void *b3AlignedAllocInternal(size_t size, int alignment)
 {
+#ifdef B3_ALLOCATOR_STATISTICS
 	b3g_numAlignedAllocs++;
+#endif
 	void *ptr;
 	ptr = b3s_alignedAllocFunc(size, alignment);
 	//	b3Printf("b3AlignedAllocInternal %d, %x\n",size,ptr);
@@ -170,8 +176,9 @@ void b3AlignedFreeInternal(void *ptr)
 	{
 		return;
 	}
-
+#ifdef B3_ALLOCATOR_STATISTICS
 	b3g_numAlignedFree++;
+#endif
 	//	b3Printf("b3AlignedFreeInternal %x\n",ptr);
 	b3s_alignedFreeFunc(ptr);
 }
diff --git a/src/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp b/src/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp
index 936130753..803f6e067 100644
--- a/src/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp
+++ b/src/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp
@@ -36,9 +36,6 @@ btScalar gGjkEpaPenetrationTolerance = 1.0e-12;
 btScalar gGjkEpaPenetrationTolerance = 0.001;
 #endif
 
-//temp globals, to improve GJK/EPA/penetration calculations
-int gNumDeepPenetrationChecks = 0;
-int gNumGjkChecks = 0;
 
 btGjkPairDetector::btGjkPairDetector(const btConvexShape *objectA, const btConvexShape *objectB, btSimplexSolverInterface *simplexSolver, btConvexPenetrationDepthSolver *penetrationDepthSolver)
 	: m_cachedSeparatingAxis(btScalar(0.), btScalar(1.), btScalar(0.)),
@@ -708,7 +705,6 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput &inpu
 	btScalar marginA = m_marginA;
 	btScalar marginB = m_marginB;
 
-	gNumGjkChecks++;
 
 	//for CCD we don't use margins
 	if (m_ignoreMargin)
@@ -1021,7 +1017,6 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput &inpu
 				// Penetration depth case.
 				btVector3 tmpPointOnA, tmpPointOnB;
 
-				gNumDeepPenetrationChecks++;
 				m_cachedSeparatingAxis.setZero();
 
 				bool isValid2 = m_penetrationDepthSolver->calcPenDepth(
diff --git a/src/LinearMath/TaskScheduler/btThreadSupportPosix.cpp b/src/LinearMath/TaskScheduler/btThreadSupportPosix.cpp
index 47f57943b..25bfda0bf 100644
--- a/src/LinearMath/TaskScheduler/btThreadSupportPosix.cpp
+++ b/src/LinearMath/TaskScheduler/btThreadSupportPosix.cpp
@@ -1,3 +1,4 @@
+
 /*
 Bullet Continuous Collision Detection and Physics Library
 Copyright (c) 2003-2018 Erwin Coumans  http://bulletphysics.com
@@ -72,7 +73,7 @@ public:
 		pthread_t thread;
 		//each tread will wait until this signal to start its work
 		sem_t* startSemaphore;
-
+		btCriticalSection* m_cs;
 		// this is a copy of m_mainSemaphore,
 		//each tread will signal once it is finished with its work
 		sem_t* m_mainSemaphore;
@@ -90,7 +91,7 @@ private:
 	void startThreads(const ConstructionInfo& threadInfo);
 	void stopThreads();
 	int waitForResponse();
-
+	btCriticalSection* m_cs;
 public:
 	btThreadSupportPosix(const ConstructionInfo& threadConstructionInfo);
 	virtual ~btThreadSupportPosix();
@@ -119,6 +120,7 @@ public:
 
 btThreadSupportPosix::btThreadSupportPosix(const ConstructionInfo& threadConstructionInfo)
 {
+	m_cs = createCriticalSection();
 	startThreads(threadConstructionInfo);
 }
 
@@ -126,6 +128,8 @@ btThreadSupportPosix::btThreadSupportPosix(const ConstructionInfo& threadConstru
 btThreadSupportPosix::~btThreadSupportPosix()
 {
 	stopThreads();
+	deleteCriticalSection(m_cs);
+	m_cs=0;
 }
 
 #if (defined(__APPLE__))
@@ -181,14 +185,18 @@ static void* threadFunction(void* argument)
 		{
 			btAssert(status->m_status);
 			status->m_userThreadFunc(userPtr);
+			status->m_cs->lock();
 			status->m_status = 2;
+			status->m_cs->unlock();
 			checkPThreadFunction(sem_post(status->m_mainSemaphore));
 			status->threadUsed++;
 		}
 		else
 		{
 			//exit Thread
+			status->m_cs->lock();
 			status->m_status = 3;
+			status->m_cs->unlock();
 			checkPThreadFunction(sem_post(status->m_mainSemaphore));
 			printf("Thread with taskId %i exiting\n", status->m_taskId);
 			break;
@@ -206,7 +214,7 @@ void btThreadSupportPosix::runTask(int threadIndex, void* userData)
 	btThreadStatus& threadStatus = m_activeThreadStatus[threadIndex];
 	btAssert(threadIndex >= 0);
 	btAssert(threadIndex < m_activeThreadStatus.size());
-
+	threadStatus.m_cs = m_cs;
 	threadStatus.m_commandId = 1;
 	threadStatus.m_status = 1;
 	threadStatus.m_userPtr = userData;
@@ -231,7 +239,10 @@ int btThreadSupportPosix::waitForResponse()
 
 	for (size_t t = 0; t < size_t(m_activeThreadStatus.size()); ++t)
 	{
-		if (2 == m_activeThreadStatus[t].m_status)
+		m_cs->lock();
+		bool hasFinished = (2 == m_activeThreadStatus[t].m_status);
+		m_cs->unlock(); 
+		if (hasFinished)
 		{
 			last = t;
 			break;
@@ -273,15 +284,15 @@ void btThreadSupportPosix::startThreads(const ConstructionInfo& threadConstructi
 		printf("starting thread %d\n", i);
 		btThreadStatus& threadStatus = m_activeThreadStatus[i];
 		threadStatus.startSemaphore = createSem("threadLocal");
-		checkPThreadFunction(pthread_create(&threadStatus.thread, NULL, &threadFunction, (void*)&threadStatus));
-
 		threadStatus.m_userPtr = 0;
+		threadStatus.m_cs = m_cs;
 		threadStatus.m_taskId = i;
 		threadStatus.m_commandId = 0;
 		threadStatus.m_status = 0;
 		threadStatus.m_mainSemaphore = m_mainSemaphore;
 		threadStatus.m_userThreadFunc = threadConstructionInfo.m_userThreadFunc;
 		threadStatus.threadUsed = 0;
+		checkPThreadFunction(pthread_create(&threadStatus.thread, NULL, &threadFunction, (void*)&threadStatus));
 
 		printf("started thread %d \n", i);
 	}
diff --git a/src/LinearMath/btQuickprof.cpp b/src/LinearMath/btQuickprof.cpp
index 8d3310981..86fd1d781 100644
--- a/src/LinearMath/btQuickprof.cpp
+++ b/src/LinearMath/btQuickprof.cpp
@@ -694,6 +694,24 @@ void CProfileManager::dumpAll()
 	CProfileManager::Release_Iterator(profileIterator);
 }
 
+
+void btEnterProfileZoneDefault(const char* name)
+{
+}
+void btLeaveProfileZoneDefault()
+{
+}
+
+#else
+void btEnterProfileZoneDefault(const char* name)
+{
+}
+void btLeaveProfileZoneDefault()
+{
+}
+#endif  //BT_NO_PROFILE
+
+
 // clang-format off
 #if defined(_WIN32) && (defined(__MINGW32__) || defined(__MINGW64__))
   #define BT_HAVE_TLS 1
@@ -743,22 +761,6 @@ unsigned int btQuickprofGetCurrentThreadIndex2()
 #endif  //BT_THREADSAFE
 }
 
-void btEnterProfileZoneDefault(const char* name)
-{
-}
-void btLeaveProfileZoneDefault()
-{
-}
-
-#else
-void btEnterProfileZoneDefault(const char* name)
-{
-}
-void btLeaveProfileZoneDefault()
-{
-}
-#endif  //BT_NO_PROFILE
-
 static btEnterProfileZoneFunc* bts_enterFunc = btEnterProfileZoneDefault;
 static btLeaveProfileZoneFunc* bts_leaveFunc = btLeaveProfileZoneDefault;
 
diff --git a/src/LinearMath/btQuickprof.h b/src/LinearMath/btQuickprof.h
index 18ba4d5fb..990d401d5 100644
--- a/src/LinearMath/btQuickprof.h
+++ b/src/LinearMath/btQuickprof.h
@@ -61,18 +61,19 @@ btLeaveProfileZoneFunc* btGetCurrentLeaveProfileZoneFunc();
 void btSetCustomEnterProfileZoneFunc(btEnterProfileZoneFunc* enterFunc);
 void btSetCustomLeaveProfileZoneFunc(btLeaveProfileZoneFunc* leaveFunc);
 
-#ifndef BT_NO_PROFILE  // FIX redefinition
-//To disable built-in profiling, please comment out next line
-//#define BT_NO_PROFILE 1
+#ifndef BT_ENABLE_PROFILE
+#define BT_NO_PROFILE 1
 #endif  //BT_NO_PROFILE
 
 const unsigned int BT_QUICKPROF_MAX_THREAD_COUNT = 64;
 
-#ifndef BT_NO_PROFILE
 //btQuickprofGetCurrentThreadIndex will return -1 if thread index cannot be determined,
 //otherwise returns thread index in range [0..maxThreads]
 unsigned int btQuickprofGetCurrentThreadIndex2();
 
+#ifndef BT_NO_PROFILE
+
+
 #include <stdio.h>  //@todo remove this, backwards compatibility
 
 #include "btAlignedAllocator.h"