diff -ur openmsx-0.5.2-orig/src/sound/Y8950.cc openmsx-0.5.2/src/sound/Y8950.cc
--- openmsx-0.5.2-orig/src/sound/Y8950.cc	2005-09-03 22:39:10.000000000 -0400
+++ openmsx-0.5.2/src/sound/Y8950.cc	2005-09-03 22:31:27.000000000 -0400
@@ -18,6 +18,22 @@
 
 namespace openmsx {
 
+static const double PI = 3.14159265358979;
+static const double EG_STEP = 0.1875;
+static const double SL_STEP = 3.0;
+static const double TL_STEP = 0.75;
+static const double DB_STEP = 0.1875;
+
+// PM speed(Hz) and depth(cent)
+static const double PM_SPEED  = 6.4;
+static const double PM_DEPTH  = 13.75 / 2;
+static const double PM_DEPTH2 = 13.75;
+
+// AM speed(Hz) and depth(dB)
+static const double AM_SPEED  = 3.7;
+static const double AM_DEPTH  = 1.0;
+static const double AM_DEPTH2 = 4.8;
+
 short Y8950::dB2LinTab[(2*DB_MUTE)*2];
 int   Y8950::Slot::sintable[PG_WIDTH];
 int   Y8950::Slot::tllTable[16][8][1<<TL_BITS][4];
diff -ur openmsx-0.5.2-orig/src/sound/Y8950.hh openmsx-0.5.2/src/sound/Y8950.hh
--- openmsx-0.5.2-orig/src/sound/Y8950.hh	2005-09-03 22:39:10.000000000 -0400
+++ openmsx-0.5.2/src/sound/Y8950.hh	2005-09-03 22:31:27.000000000 -0400
@@ -115,11 +115,9 @@
 
 
 		// Dynamic range of envelope
-		static const double EG_STEP = 0.1875;
 		static const int EG_BITS = 9;
 		static const int EG_MUTE = 1<<EG_BITS;
 		// Dynamic range of sustine level
-		static const double SL_STEP = 3.0;
 		static const int SL_BITS = 4;
 		static const int SL_MUTE = 1<<SL_BITS;
 		// Size of Sintable ( 1 -- 18 can be used, but 7 -- 14 recommended.)
@@ -133,7 +131,6 @@
 		static const int EG_DP_BITS = 23;
 		static const int EG_DP_WIDTH = 1<<EG_DP_BITS;
 		// Dynamic range of total level
-		static const double TL_STEP = 0.75;
 		static const int TL_BITS = 6;
 		static const int TL_MUTE = 1<<TL_BITS;
 
@@ -193,7 +190,6 @@
 	// Definition of envelope mode
 	enum { ATTACK,DECAY,SUSHOLD,SUSTINE,RELEASE,FINISH };
 	// Dynamic range
-	static const double DB_STEP = 0.1875;
 	static const int DB_BITS = 9;
 	static const int DB_MUTE = 1<<DB_BITS;
 	// PM table is calcurated by PM_AMP * pow(2,PM_DEPTH*sin(x)/1200)
@@ -264,15 +260,6 @@
 	Slot *slot[18];
 
 	static const int CLK_FREQ = 3579545;
-	static const double PI = 3.14159265358979;
-	// PM speed(Hz) and depth(cent)
-	static const double PM_SPEED = 6.4;
-	static const double PM_DEPTH = (13.75/2);
-	static const double PM_DEPTH2 = 13.75;
-	// AM speed(Hz) and depth(dB)
-	static const double AM_SPEED = 3.7;
-	static const double AM_DEPTH = 1.0;
-	static const double AM_DEPTH2 = 4.8;
 	// Bits for liner value
 	static const int DB2LIN_AMP_BITS = 11;
 	static const int SLOT_AMP_BITS = DB2LIN_AMP_BITS;
diff -ur openmsx-0.5.2-orig/src/sound/YM2413.cc openmsx-0.5.2/src/sound/YM2413.cc
--- openmsx-0.5.2-orig/src/sound/YM2413.cc	2005-09-03 22:39:10.000000000 -0400
+++ openmsx-0.5.2/src/sound/YM2413.cc	2005-09-03 22:31:27.000000000 -0400
@@ -20,6 +20,18 @@
 
 static const int CLOCK_FREQ = 3579545;
 static const double PI = 3.14159265358979323846;
+static const double DB_STEP = 48.0 / (1 << 8); //   48 / (1 << DB_BITS)
+static const double EG_STEP = 0.375;
+static const double TL_STEP = 0.75;
+static const double SL_STEP = 3.0;
+
+// PM speed(Hz) and depth(cent)
+static const double PM_SPEED = 6.4;
+static const double PM_DEPTH = 13.75;
+
+// AM speed(Hz) and depth(dB)
+static const double AM_SPEED = 3.6413;
+static const double AM_DEPTH = 4.875;
 
 int YM2413::pmtable[PM_PG_WIDTH];
 int YM2413::amtable[AM_PG_WIDTH];
diff -ur openmsx-0.5.2-orig/src/sound/YM2413.hh openmsx-0.5.2/src/sound/YM2413.hh
--- openmsx-0.5.2-orig/src/sound/YM2413.hh	2005-09-03 22:39:10.000000000 -0400
+++ openmsx-0.5.2/src/sound/YM2413.hh	2005-09-03 22:31:27.000000000 -0400
@@ -11,13 +11,14 @@
 namespace openmsx {
 
 class EmuTime;
+class MSXMotherBoard;
 
 class YM2413 : public YM2413Core, private SoundDevice, private Debuggable
 {
 	struct Patch {
 		Patch();
 		Patch(int n, const byte* data);
-		
+
 		bool AM, PM, EG;
 		byte KR; // 0-1
 		byte ML; // 0-15
@@ -174,21 +175,17 @@
 
 	// Dynamic range (Accuracy of sin table)
 	static const int DB_BITS = 8;
-	static const double DB_STEP = 48.0 / (1 << DB_BITS);
 	static const int DB_MUTE = 1 << DB_BITS;
 
 	// Dynamic range of envelope
-	static const double EG_STEP = 0.375;
 	static const int EG_BITS = 7;
 	static const int EG_MUTE = 1 << EG_BITS;
 
 	// Dynamic range of total level
-	static const double TL_STEP = 0.75;
 	static const int TL_BITS = 6;
 	static const int TL_MUTE = 1 << TL_BITS;
 
 	// Dynamic range of sustine level
-	static const double SL_STEP = 3.0;
 	static const int SL_BITS = 4;
 	static const int SL_MUTE = 1 << SL_BITS;
 
@@ -210,18 +207,10 @@
 	static const int AM_DP_BITS = 16;
 	static const int AM_DP_WIDTH = 1 << AM_DP_BITS;
 
-	// PM table is calcurated by PM_AMP * pow(2,PM_DEPTH*sin(x)/1200) 
+	// PM table is calcurated by PM_AMP * pow(2,PM_DEPTH*sin(x)/1200)
 	static const int PM_AMP_BITS = 8;
 	static const int PM_AMP = 1 << PM_AMP_BITS;
 
-	// PM speed(Hz) and depth(cent) 
-	static const double PM_SPEED = 6.4;
-	static const double PM_DEPTH = 13.75;
-
-	// AM speed(Hz) and depth(dB)
-	static const double AM_SPEED = 3.6413;
-	static const double AM_DEPTH = 4.875;
-
 	int maxVolume;
 
 	// Register
Only in openmsx-0.5.2-orig/src/sound: YM2413.hh.orig
diff -ur openmsx-0.5.2-orig/src/video/Scale2xScaler.cc openmsx-0.5.2/src/video/Scale2xScaler.cc
--- openmsx-0.5.2-orig/src/video/Scale2xScaler.cc	2005-09-03 22:39:10.000000000 -0400
+++ openmsx-0.5.2/src/video/Scale2xScaler.cc	2005-09-03 22:31:27.000000000 -0400
@@ -132,8 +132,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
@@ -254,8 +256,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
@@ -364,8 +368,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
@@ -570,8 +576,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
@@ -702,8 +710,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
@@ -805,8 +815,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
@@ -922,8 +934,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
@@ -1037,8 +1051,10 @@
 			, "r" (src2) // 2
 			, "r" (dst)  // 3
 			: "eax"
+			#ifdef __MMX__
 			, "mm0", "mm1", "mm2", "mm3"
 			, "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	};
diff -ur openmsx-0.5.2-orig/src/video/Scaler.cc openmsx-0.5.2/src/video/Scaler.cc
--- openmsx-0.5.2-orig/src/video/Scaler.cc	2005-09-03 22:39:10.000000000 -0400
+++ openmsx-0.5.2/src/video/Scaler.cc	2005-09-03 22:40:18.000000000 -0400
@@ -86,8 +86,11 @@
 			, "r" (pOut) // 1
 			, "r" (nBytes) // 2
 			, "r" (0) // 3
-			: "mm0", "mm1", "mm2", "mm3"
-			, "mm4", "mm5", "mm6", "mm7"
+			#ifdef __MMX__
+			:
+			"mm0", "mm1", "mm2", "mm3" ,
+			"mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -126,8 +129,11 @@
 			, "r" (pOut) // 1
 			, "r" (nBytes) // 2
 			, "r" (0) // 3
-			: "mm0", "mm1", "mm2", "mm3"
-			, "mm4", "mm5", "mm6", "mm7"
+			#ifdef __MMX__
+			:
+			"mm0", "mm1", "mm2", "mm3",
+			"mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -195,8 +201,11 @@
 			, "r" (pOut) // 1
 			, "r" (width) // 2
 			, "r" (0) // 3
-			: "mm0", "mm1", "mm2", "mm3"
-			, "mm4", "mm5", "mm6", "mm7"
+			#ifdef __MMX__
+			:
+			"mm0", "mm1", "mm2", "mm3",
+			"mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -244,8 +253,11 @@
 			, "r" (pOut) // 1
 			, "r" (width) // 2
 			, "r" (0) // 3
-			: "mm0", "mm1", "mm2", "mm3"
-			, "mm4", "mm5", "mm6", "mm7"
+			#ifdef __MMX__
+			:
+			"mm0", "mm1", "mm2", "mm3",
+			"mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -293,8 +305,11 @@
 			, "r" (pOut) // 1
 			, "r" (width) // 2
 			, "r" (0) // 3
-			: "mm0", "mm1", "mm2", "mm3"
-			, "mm4", "mm5", "mm6", "mm7"
+			#ifdef __MMX__
+			:
+			"mm0", "mm1", "mm2", "mm3",
+			"mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -342,8 +357,11 @@
 			, "r" (pOut) // 1
 			, "r" (width) // 2
 			, "r" (0) // 3
-			: "mm0", "mm1", "mm2", "mm3"
-			, "mm4", "mm5", "mm6", "mm7"
+			#ifdef __MMX__
+			:
+			"mm0", "mm1", "mm2", "mm3",
+			"mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -396,7 +414,10 @@
 			, "rm" (col32) // 1
 			, "r" (width * sizeof(Pixel)) // 2
 			, "r" (0) // 3
-			: "mm0"
+			#ifdef __MMX__
+			:
+			"mm0"
+			#endif
 		);
 		return;
 	}
@@ -429,7 +450,10 @@
 			, "rm" (col32) // 1
 			, "r" (width * sizeof(Pixel)) // 2
 			, "r" (0) // 3
-			: "mm0"
+			#ifdef __MMX__
+			:
+			"mm0"
+			#endif
 		);
 		return;
 	}
diff -ur openmsx-0.5.2-orig/src/video/SimpleScaler.cc openmsx-0.5.2/src/video/SimpleScaler.cc
--- openmsx-0.5.2-orig/src/video/SimpleScaler.cc	2005-09-03 22:39:10.000000000 -0400
+++ openmsx-0.5.2/src/video/SimpleScaler.cc	2005-09-03 22:31:27.000000000 -0400
@@ -260,13 +260,13 @@
 			"punpcklwd %%mm6, %%mm6;"
 			"punpckldq %%mm6, %%mm6;"	// mm6 = c2
 			"pxor	%%mm7, %%mm7;"
-			
-			"movd	(%0,%%eax,4), %%mm0;"
+
+			"movd	(%0,%%eax), %%mm0;"
 			"punpcklbw %%mm7, %%mm0;"	// p0 = pIn[0]
 			"movq	%%mm0, %%mm2;"
 			"pmullw	%%mm5, %%mm2;"		// f0 = multiply(p0, c1)
 			"movq	%%mm2, %%mm3;"		// f1 = f0
-			
+
 			".p2align 4,,15;"
 		"1:"
 			"pmullw	%%mm6, %%mm0;"
@@ -274,31 +274,31 @@
 			"paddw	%%mm3, %%mm0;"
 			"psrlw	$8, %%mm0;"		// f1 + tmp
 
-			"movd	4(%0,%%eax,4), %%mm1;"
+			"movd	4(%0,%%eax), %%mm1;"
 			"punpcklbw %%mm7, %%mm1;"	// p1 = pIn[x + 1]
 			"movq	%%mm1, %%mm3;"
 			"pmullw	%%mm5, %%mm3;"		// f1 = multiply(p1, c1)
 			"paddw	%%mm3, %%mm4;"
 			"psrlw	$8, %%mm4;"		// f1 + tmp
 			"packuswb %%mm4, %%mm0;"
-			"movq	%%mm0, (%1,%%eax,8);"	// pOut[2*x+0] = ..  pOut[2*x+1] = ..
+			"movq	%%mm0, (%1,%%eax,2);"	// pOut[2*x+0] = ..  pOut[2*x+1] = ..
 
 			"pmullw	%%mm6, %%mm1;"
 			"movq	%%mm1, %%mm4;"		// tmp = multiply(p1, c2)
 			"paddw	%%mm2, %%mm1;"
 			"psrlw	$8, %%mm1;"		// f0 + tmp
 
-			"movd	8(%0,%%eax,4), %%mm0;"
+			"movd	8(%0,%%eax), %%mm0;"
 			"punpcklbw %%mm7, %%mm0;"	// p0 = pIn[x + 2]
 			"movq	%%mm0, %%mm2;"
 			"pmullw %%mm5, %%mm2;"		// f0 = multiply(p0, c1)
 			"paddw	%%mm2, %%mm4;"
 			"psrlw	$8, %%mm4;"		// f0 + tmp
 			"packuswb %%mm4, %%mm1;"
-			"movq	%%mm1, 8(%1,%%eax,8);"	// pOut[2*x+2] = ..  pOut[2*x+3] = ..
+			"movq	%%mm1, 8(%1,%%eax,2);"	// pOut[2*x+2] = ..  pOut[2*x+3] = ..
 
-			"addl	$2, %%eax;"
-			"cmpl	$318, %%eax;"
+			"addl	$8, %%eax;"
+			"cmpl	$1272, %%eax;"
 			"jl	1b;"
 
 			"pmullw	%%mm6, %%mm0;"
@@ -306,23 +306,23 @@
 			"paddw	%%mm3, %%mm0;"
 			"psrlw	$8, %%mm0;"		// f1 + tmp
 
-			"movd	4(%0,%%eax,4), %%mm1;"
+			"movd	4(%0,%%eax), %%mm1;"
 			"punpcklbw %%mm7, %%mm1;"	// p1 = pIn[x + 1]
 			"movq	%%mm1, %%mm3;"
 			"pmullw	%%mm5, %%mm3;"		// f1 = multiply(p1, c1)
 			"paddw	%%mm3, %%mm4;"
 			"psrlw	$8, %%mm4;"		// f1 + tmp
 			"packuswb %%mm4, %%mm0;"
-			"movq	%%mm0, (%1,%%eax,8);"	// pOut[2*x+0] = ..  pOut[2*x+1] = ..
-			
+			"movq	%%mm0, (%1,%%eax,2);"	// pOut[2*x+0] = ..  pOut[2*x+1] = ..
+
 			"movq	%%mm1, %%mm4;"
 			"pmullw	%%mm6, %%mm1;" 		// tmp = multiply(p1, c2)
 			"paddw	%%mm2, %%mm1;"
 			"psrlw	$8, %%mm1;"		// f0 + tmp
 
 			"packuswb %%mm4, %%mm1;"
-			"movq	%%mm1, 8(%1,%%eax,8);"	// pOut[2*x+0] = ..  pOut[2*x+1] = ..
-			
+			"movq	%%mm1, 8(%1,%%eax,2);"	// pOut[2*x+0] = ..  pOut[2*x+1] = ..
+
 			"emms;"
 
 			: // no output
@@ -330,17 +330,19 @@
 			, "r" (pOut)  // 1
 			, "r" (c1)    // 2
 			, "r" (c2)    // 3
-			: "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
-			, "eax"
+			: "eax"
+			#ifdef __MMX__
+			, "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
 	#endif
-	
+
 	// non-MMX routine, both 16bpp and 32bpp
 	mult1.setFactor(c1);
 	mult2.setFactor(c2);
-	 
+
 	Pixel p0 = pIn[0];
 	Pixel p1;
 	unsigned f0 = mult1.mul32(p0);
@@ -428,16 +430,16 @@
 			"punpcklwd %%mm6, %%mm6;"
 			"punpckldq %%mm6, %%mm6;"	// mm6 = c2
 			"pxor	%%mm7, %%mm7;"
-		
-			"movd	(%0,%%eax,4), %%mm0;"
+
+			"movd	(%0,%%eax), %%mm0;"
 			"punpcklbw %%mm7, %%mm0;"	// p0 = pIn[0]
 			"movq	%%mm0, %%mm2;"
 			"pmullw	%%mm5, %%mm2;"		// f0 = multiply(p0, c1)
 			"movq	%%mm2, %%mm3;"		// f1 = f0
-		
+
 			".p2align 4,,15;"
 		"1:"
-			"movd	4(%0,%%eax,4), %%mm1;"
+			"movd	4(%0,%%eax), %%mm1;"
 			"pxor	%%mm7, %%mm7;"
 			"punpcklbw %%mm7, %%mm1;"	// p1 = pIn[x + 1]
 			"movq	%%mm0, %%mm4;"
@@ -449,7 +451,7 @@
 			"psrlw	$8, %%mm4;"		// f0 + t + t0
 			"movq	%%mm0, %%mm2;"		// f0 = t0
 
-			"movd	8(%0,%%eax,4), %%mm0;"
+			"movd	8(%0,%%eax), %%mm0;"
 			"punpcklbw %%mm7, %%mm0;"
 			"movq	%%mm1, %%mm7;"
 			"pmullw	%%mm6, %%mm7;"		// t = multiply(p1, c2)
@@ -460,13 +462,13 @@
 			"psrlw	$8, %%mm7;"		// f1 + t + t1
 			"movq	%%mm1, %%mm3;"		// f1 = t1
 			"packuswb %%mm7, %%mm4;"
-			"movq	%%mm4, (%1,%%eax,4);"	// pOut[x] = ..  pOut[x+1] = ..
-			
-			"addl	$2, %%eax;"
-			"cmpl	$638, %%eax;"
+			"movq	%%mm4, (%1,%%eax);"	// pOut[x] = ..  pOut[x+1] = ..
+
+			"addl	$8, %%eax;"
+			"cmpl	$2552, %%eax;"
 			"jl	1b;"
 
-			"movd	4(%0,%%eax,4), %%mm1;"
+			"movd	4(%0,%%eax), %%mm1;"
 			"pxor	%%mm7, %%mm7;"
 			"punpcklbw %%mm7, %%mm1;"	// p1 = pIn[x + 1]
 			"movq	%%mm0, %%mm4;"
@@ -482,8 +484,8 @@
 			"paddw	%%mm0, %%mm1;"
 			"psrlw	$8, %%mm1;"		// f1 + t + t1
 			"packuswb %%mm1, %%mm4;"
-			"movq	%%mm4, (%1,%%eax,4);"	// pOut[x] = ..  pOut[x+1] = ..
-			
+			"movq	%%mm4, (%1,%%eax);"	// pOut[x] = ..  pOut[x+1] = ..
+
 			"emms;"
 
 			: // no output
@@ -491,8 +493,10 @@
 			, "r" (pOut)  // 1
 			, "r" (c1)    // 2
 			, "r" (c2)    // 3
-			: "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
-			, "eax"
+			: "eax"
+			#ifdef __MMX__
+			, "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -547,8 +551,8 @@
 			"pshufw $0, %%mm6, %%mm6;"
 			".p2align 4,,15;"
 		"1:"
-			"movq	(%0,%%eax,4), %%mm0;"
-			"pavgb	(%1,%%eax,4), %%mm0;"
+			"movq	(%0,%%eax), %%mm0;"
+			"pavgb	(%1,%%eax), %%mm0;"
 			"movq	%%mm0, %%mm4;"
 			"punpcklbw %%mm7, %%mm0;"
 			"punpckhbw %%mm7, %%mm4;"
@@ -556,8 +560,8 @@
 			"pmulhuw %%mm6, %%mm4;"
 			"packuswb %%mm4, %%mm0;"
 
-			"movq	8(%0,%%eax,4), %%mm1;"
-			"pavgb	8(%1,%%eax,4), %%mm1;"
+			"movq	8(%0,%%eax), %%mm1;"
+			"pavgb	8(%1,%%eax), %%mm1;"
 			"movq	%%mm1, %%mm5;"
 			"punpcklbw %%mm7, %%mm1;"
 			"punpckhbw %%mm7, %%mm5;"
@@ -565,8 +569,8 @@
 			"pmulhuw %%mm6, %%mm5;"
 			"packuswb %%mm5, %%mm1;"
 
-			"movq	16(%0,%%eax,4), %%mm2;"
-			"pavgb	16(%1,%%eax,4), %%mm2;"
+			"movq	16(%0,%%eax), %%mm2;"
+			"pavgb	16(%1,%%eax), %%mm2;"
 			"movq	%%mm2, %%mm4;"
 			"punpcklbw %%mm7, %%mm2;"
 			"punpckhbw %%mm7, %%mm4;"
@@ -574,8 +578,8 @@
 			"pmulhuw %%mm6, %%mm4;"
 			"packuswb %%mm4, %%mm2;"
 
-			"movq	24(%0,%%eax,4), %%mm3;"
-			"pavgb	24(%1,%%eax,4), %%mm3;"
+			"movq	24(%0,%%eax), %%mm3;"
+			"pavgb	24(%1,%%eax), %%mm3;"
 			"movq	%%mm3, %%mm5;"
 			"punpcklbw %%mm7, %%mm3;"
 			"punpckhbw %%mm7, %%mm5;"
@@ -583,24 +587,26 @@
 			"pmulhuw %%mm6, %%mm5;"
 			"packuswb %%mm5, %%mm3;"
 
-			"movntq %%mm0,   (%2,%%eax,4);"
-			"movntq %%mm1,  8(%2,%%eax,4);"
-			"movntq %%mm2, 16(%2,%%eax,4);"
-			"movntq %%mm3, 24(%2,%%eax,4);"
-			
-			"addl	$8, %%eax;"
-			"cmpl	$640, %%eax;"
+			"movntq %%mm0,   (%2,%%eax);"
+			"movntq %%mm1,  8(%2,%%eax);"
+			"movntq %%mm2, 16(%2,%%eax);"
+			"movntq %%mm3, 24(%2,%%eax);"
+
+			"addl	$32, %%eax;"
+			"cmpl	$2560, %%eax;"
 			"jl	1b;"
-			
+
 			"emms;"
-			
+
 			: // no output
 			: "r" (src1)  // 0
 			, "r" (src2)  // 1
 			, "r" (dst)   // 2
 			, "r" (alpha << 8) // 3
-			: "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
-			, "eax"
+			: "eax"
+			#ifdef __MMX__
+			, "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
+			#endif
 		);
 		return;
 
@@ -610,15 +616,14 @@
 			"movd	%3, %%mm6;"
 			"pxor	%%mm7, %%mm7;"
 			"punpcklwd %%mm6, %%mm6;"
-			"punpckldq %%mm6, %%mm6;"
-
 			"xorl	%%eax, %%eax;"
+			"punpckldq %%mm6, %%mm6;"
 			".p2align 4,,15;"
 		"1:"
 			// load
-			"movq	(%0,%%eax,4), %%mm0;"
+			"movq	(%0,%%eax), %%mm0;"
 			"movq	%%mm0, %%mm1;"
-			"movq	(%1,%%eax,4), %%mm2;"
+			"movq	(%1,%%eax), %%mm2;"
 			"movq	%%mm2, %%mm3;"
 			// unpack
 			"punpcklbw %%mm7, %%mm0;"
@@ -634,21 +639,23 @@
 			// pack
 			"packuswb %%mm1, %%mm0;"
 			// store
-			"movq %%mm0, (%2,%%eax,4);"
-			
-			"addl	$2, %%eax;"
-			"cmpl	$640, %%eax;"
+			"movq %%mm0, (%2,%%eax);"
+
+			"addl	$8, %%eax;"
+			"cmpl	$2560, %%eax;"
 			"jl	1b;"
-			
+
 			"emms;"
-			
+
 			: // no output
 			: "r" (src1)  // 0
 			, "r" (src2)  // 1
 			, "r" (dst)   // 2
 			, "r" (alpha << 7) // 3
-			: "mm0", "mm1", "mm2", "mm3", "mm6", "mm7"
-			, "eax"
+			: "eax"
+			#ifdef __MMX__
+			, "mm0", "mm1", "mm2", "mm3", "mm6", "mm7"
+			#endif
 		);
 		return;
 	}
@@ -659,17 +666,17 @@
 		darkener.setFactor(alpha);
 		word* table = darkener.getTable();
 		Pixel mask = ~blender.getMask();
-		
+
 		asm (
 			"movd	%4, %%mm7;"
 			"xorl	%%ecx, %%ecx;"
 			"pshufw	$0, %%mm7, %%mm7;"
-			
+
 			".p2align 4,,15;"
-		"1:"	"movq	 (%0,%%ecx,2), %%mm0;"
-			"movq	8(%0,%%ecx,2), %%mm1;"
-			"movq	 (%1,%%ecx,2), %%mm2;"
-			"movq	8(%1,%%ecx,2), %%mm3;"
+		"1:"	"movq	 (%0,%%ecx), %%mm0;"
+			"movq	8(%0,%%ecx), %%mm1;"
+			"movq	 (%1,%%ecx), %%mm2;"
+			"movq	8(%1,%%ecx), %%mm3;"
 
 			"movq	%%mm7, %%mm4;"
 			"movq	%%mm7, %%mm5;"
@@ -683,14 +690,14 @@
 			"pavgw	%%mm3, %%mm1;"
 			"paddw	%%mm4, %%mm0;"
 			"paddw	%%mm5, %%mm1;"
-			
+
 			"pextrw	$0, %%mm0, %%eax;"
 			"movw	(%2,%%eax,2), %%ax;"
 			"pinsrw	$0, %%eax, %%mm0;"
 			"pextrw	$0, %%mm1, %%eax;"
 			"movw	(%2,%%eax,2), %%ax;"
 			"pinsrw	$0, %%eax, %%mm1;"
-			
+
 			"pextrw	$1, %%mm0, %%eax;"
 			"movw	(%2,%%eax,2), %%ax;"
 			"pinsrw	$1, %%eax, %%mm0;"
@@ -711,12 +718,12 @@
 			"pextrw	$3, %%mm1, %%eax;"
 			"movw	(%2,%%eax,2), %%ax;"
 			"pinsrw	$3, %%eax, %%mm1;"
-			
-			"movntq	%%mm0,   (%3,%%ecx,2);"
-			"movntq	%%mm1,  8(%3,%%ecx,2);"
 
-			"addl	$8, %%ecx;"
-			"cmpl	$640, %%ecx;"
+			"movntq	%%mm0,   (%3,%%ecx);"
+			"movntq	%%mm1,  8(%3,%%ecx);"
+
+			"addl	$16, %%ecx;"
+			"cmpl	$1280, %%ecx;"
 			"jl	1b;"
 			"emms;"
 			: // no output
@@ -725,16 +732,18 @@
 			, "r" (table) // 2
 			, "r" (dst)   // 3
 			, "m" (mask)   // 4
-			: "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm7"
-			, "eax", "ecx"
+			: "eax", "ecx"
+			#ifdef __MMX__
+			, "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm7"
+			#endif
 		);
 		return;
 	}
 	// MMX routine 16bpp is missing, but it's difficult to write because
 	// of the missing "pextrw" and "pinsrw" instructions
-	
+
 	#endif
-		
+
 	// non-MMX routine, both 16bpp and 32bpp
 	for (unsigned x = 0; x < 640; ++x) {
 		dst[x] = mult1.multiply(