Slight cleanup

benchmark/Makefile

@@ -7,6 +7,7 @@ default:
 	@echo "make [ Linux | MacOSX | Win32 | clean | clean-Win32]"
 
 clean:
+	cd common ; make clean
 	cd Linux ; make clean
 	cd MacOSX ; make clean
 
@@ -23,4 +24,5 @@ Win32:
 	cd $@ && $(WINMAKE) && $(WINMAKE)
 
 clean-Win32:
+	cd common && $(WINMAKE) clean
 	cd Win32 && $(WINMAKE) clean

benchmark/Win32/Makefile

@@ -9,31 +9,28 @@ LINKOBJ = noisebench.o
 LIBS = -L$(MINGW32)/lib -mwindows -lglfw -lopengl32 -lglu32 -mconsole -g3
 INCS = -I. -I$(MINGW32)/include
 CFLAGS = $(INCS) -Wall -O3 -ffast-math -g3
-RM = del
-CP = copy
-CAT = type
 EXECNAME = noisebench.exe
 OUTPUTFILE = ashimanoise.log
 
 all: $(EXECNAME)
 
 clean:
-	${RM} $(OBJ) $(EXECNAME) $(SHADERS) $(OUTPUTFILE) $(SRC)
+	del $(OBJ) $(EXECNAME) $(SHADERS) $(OUTPUTFILE) $(SRC)
 
 GLSL-ashimanoise.vert:
-	${CP} ..\common\GLSL-ashimanoise.vert .
+	copy ..\common\GLSL-ashimanoise.vert .
 
 GLSL-ashimanoise2D.frag:
-	${CP} ..\common\GLSL-ashimanoise2D.frag .
+	copy ..\common\GLSL-ashimanoise2D.frag .
 
 GLSL-ashimanoise3D.frag:
-	${CP} ..\common\GLSL-ashimanoise3D.frag .
+	copy ..\common\GLSL-ashimanoise3D.frag .
 
 GLSL-ashimanoise4D.frag:
-	${CP} ..\common\GLSL-ashimanoise4D.frag .
+	copy ..\common\GLSL-ashimanoise4D.frag .
 
 $(SRC):
-	${CP} ..\common\$(SRC) .
+	copy ..\common\$(SRC) .
 
 $(OBJ): $(SRC)
 	$(CC) -c $(SRC) -o $(OBJ) $(CFLAGS)
@@ -43,4 +40,4 @@ $(EXECNAME): $(OBJ) $(SHADERS)
 
 run: $(EXECNAME)
 	./$(EXECNAME)
-	$(CAT) $(OUTPUTFILE)
+	type $(OUTPUTFILE)

src/noise4D.glsl

@@ -41,18 +41,17 @@ float simplexNoise(vec4 v)
   i0.x = dot( isX, vec3( 1.0 ) );
   i0.yzw = 1.0 - isX;
 
-  vec2 isY = step( x0.zw, x0.yy );
-  i0.y += dot( isY, vec2( 1.0 ) );
-  i0.zw += 1.0 - isY;
+  vec3 isYZ = step( x0.zww, x0.yyz );
+  i0.y += dot( isYZ.xy, vec2( 1.0 ) );
+  i0.zw += 1.0 - isYZ.xy;
 
-  float isZ = step( x0.w, x0.z ); // Could be folded into the vec2 step() above
-  i0.z += isZ;
-  i0.w += 1.0 - isZ;
+  i0.z += isYZ.z;
+  i0.w += 1.0 - isYZ.z;
 
   // i0 now contains the unique values 0,1,2,3 in each channel
-  vec4 i1 = clamp(   i0, 0.0, 1.0 );
-  vec4 i2 = clamp( --i0, 0.0, 1.0 );
-  vec4 i3 = clamp( --i0, 0.0, 1.0 );
+  vec4 i3 = clamp( i0, 0.0, 1.0 );
+  vec4 i2 = clamp( i0-1.0, 0.0, 1.0 );
+  vec4 i1 = clamp( i0-2.0, 0.0, 1.0 );
 #else
 // Force existance of strict total ordering in sort.
   vec4 q0 = floor(x0 * 1024.0) + vec4( 0.0, 1.0/4.0, 2.0/4.0 , 3.0/4.0);