@@ -40,11 +40,11 @@
 {
 	float result[4][4];
 
 	for (uint_fast8_t i = 0; i < 4; i++) {
 		for (uint_fast8_t j = 0; j < 4; j++) {
-			__asm__ (
+			__asm__ __volatile__ (
 			    "movd	(%2), %%mm0\n\t"
 			    "punpckldq	16(%2), %%mm0\n\t"
 			    "pfmul	(%1), %%mm0\n\t"
 			    "movd	32(%2), %%mm1\n\t"
 			    "punpckldq	48(%2), %%mm1\n\t"
@@ -58,11 +58,11 @@
 			    : "mm0", "mm1", "memory"
 			);
 		}
 	}
 
-	__asm__ ("femms");
+	__asm__ __volatile__ ("femms");
 
 	memcpy(self->_values, result, sizeof(result));
 }
 static void
 multiplyWithMatrix_3DNow(OFMatrix4x4 *self, SEL _cmd, OFMatrix4x4 *matrix)
@@ -69,11 +69,11 @@
 {
 	float result[4][4];
 
 	for (uint_fast8_t i = 0; i < 4; i++) {
 		for (uint_fast8_t j = 0; j < 4; j++) {
-			__asm__ (
+			__asm__ __volatile__ (
 			    "movd	(%2), %%mm0\n\t"
 			    "punpckldq	16(%2), %%mm0\n\t"
 			    "pfmul	(%1), %%mm0\n\t"
 			    "movd	32(%2), %%mm1\n\t"
 			    "punpckldq	48(%2), %%mm1\n\t"
@@ -88,20 +88,20 @@
 			    : "mm0", "mm1", "memory"
 			);
 		}
 	}
 
-	__asm__ ("femms");
+	__asm__ __volatile__ ("femms");
 
 	memcpy(self->_values, result, sizeof(result));
 }
 
 static void
 transformVectors_enhanced3DNow(OFMatrix4x4 *self, SEL _cmd, OFVector4D *vectors,
     size_t count)
 {
-	__asm__ (
+	__asm__ __volatile__ (
 	    "0:\n\t"
 	    "test	%0, %0\n\t"
 	    "jz		0f\n"
 	    "\n\t"
 	    "movq	(%1), %%mm0\n\t"
@@ -157,11 +157,11 @@
 
 static void
 transformVectors_3DNow(OFMatrix4x4 *self, SEL _cmd, OFVector4D *vectors,
     size_t count)
 {
-	__asm__ (
+	__asm__ __volatile__ (
 	    "0:\n\t"
 	    "test	%0, %0\n\t"
 	    "jz		0f\n"
 	    "\n\t"
 	    "movq	(%1), %%mm0\n\t"