diff --git a/core/command_queue_mt.cpp b/core/command_queue_mt.cpp
index 36d4b0a32f..2bdf02295c 100644
--- a/core/command_queue_mt.cpp
+++ b/core/command_queue_mt.cpp
@@ -97,7 +97,7 @@ tryagain:
 		return false;
 	}
 
-	dealloc_ptr += (size >> 1) + sizeof(uint32_t);
+	dealloc_ptr += (size >> 1) + 8;
 	return true;
 }
 
@@ -107,6 +107,7 @@ CommandQueueMT::CommandQueueMT(bool p_sync) {
 	write_ptr = 0;
 	dealloc_ptr = 0;
 	mutex = Mutex::create();
+	command_mem = (uint8_t *)memalloc(COMMAND_MEM_SIZE);
 
 	for (int i = 0; i < SYNC_SEMAPHORES; i++) {
 
@@ -128,4 +129,5 @@ CommandQueueMT::~CommandQueueMT() {
 
 		memdelete(sync_sems[i].sem);
 	}
+	memfree(command_mem);
 }
diff --git a/core/command_queue_mt.h b/core/command_queue_mt.h
index 3fd660a3db..59eabd8786 100644
--- a/core/command_queue_mt.h
+++ b/core/command_queue_mt.h
@@ -36,6 +36,7 @@
 #include "core/os/semaphore.h"
 #include "core/simple_type.h"
 #include "core/typedefs.h"
+
 /**
 	@author Juan Linietsky <reduzio@gmail.com>
 */
@@ -254,6 +255,7 @@
 		unlock();                                                                              \
 		if (sync) sync->post();                                                                \
 		ss->sem->wait();                                                                       \
+		ss->in_use = false;                                                                    \
 	}
 
 #define CMD_SYNC_TYPE(N) CommandSync##N<T, M COMMA(N) COMMA_SEP_LIST(TYPE_ARG, N)>
@@ -270,6 +272,7 @@
 		unlock();                                                                     \
 		if (sync) sync->post();                                                       \
 		ss->sem->wait();                                                              \
+		ss->in_use = false;                                                           \
 	}
 
 #define MAX_CMD_PARAMS 13
@@ -295,7 +298,6 @@ class CommandQueueMT {
 
 		virtual void post() {
 			sync_sem->sem->post();
-			sync_sem->in_use = false;
 		}
 	};
 
@@ -318,7 +320,7 @@ class CommandQueueMT {
 		SYNC_SEMAPHORES = 8
 	};
 
-	uint8_t command_mem[COMMAND_MEM_SIZE];
+	uint8_t *command_mem;
 	uint32_t read_ptr;
 	uint32_t write_ptr;
 	uint32_t dealloc_ptr;
@@ -330,7 +332,7 @@ class CommandQueueMT {
 	T *allocate() {
 
 		// alloc size is size+T+safeguard
-		uint32_t alloc_size = sizeof(T) + sizeof(uint32_t);
+		uint32_t alloc_size = ((sizeof(T) + 8 - 1) & ~(8 - 1)) + 8;
 
 	tryagain:
 
@@ -360,7 +362,7 @@ class CommandQueueMT {
 				}
 
 				// if this happens, it's a bug
-				ERR_FAIL_COND_V((COMMAND_MEM_SIZE - write_ptr) < sizeof(uint32_t), NULL);
+				ERR_FAIL_COND_V((COMMAND_MEM_SIZE - write_ptr) < 8, NULL);
 				// zero means, wrap to beginning
 
 				uint32_t *p = (uint32_t *)&command_mem[write_ptr];
@@ -372,12 +374,13 @@ class CommandQueueMT {
 		// Allocate the size and the 'in use' bit.
 		// First bit used to mark if command is still in use (1)
 		// or if it has been destroyed and can be deallocated (0).
+		uint32_t size = (sizeof(T) + 8 - 1) & ~(8 - 1);
 		uint32_t *p = (uint32_t *)&command_mem[write_ptr];
-		*p = (sizeof(T) << 1) | 1;
-		write_ptr += sizeof(uint32_t);
+		*p = (size << 1) | 1;
+		write_ptr += 8;
 		// allocate the command
 		T *cmd = memnew_placement(&command_mem[write_ptr], T);
-		write_ptr += sizeof(T);
+		write_ptr += size;
 		return cmd;
 	}
 
@@ -415,7 +418,7 @@ class CommandQueueMT {
 			goto tryagain;
 		}
 
-		read_ptr += sizeof(uint32_t);
+		read_ptr += 8;
 
 		CommandBase *cmd = reinterpret_cast<CommandBase *>(&command_mem[read_ptr]);