diff --git a/gen.go b/gen.go
index f07721b..a5b7510 100644
--- a/gen.go
+++ b/gen.go
@@ -53,11 +53,11 @@ func mul(a, b int) int { return a * b }
 func mod(a, b int) int { return a % b }
 
 func afunc(x, y int) string {
-	return fmt.Sprintf("a[%d][%d]", y%5, x%5)
+	return fmt.Sprintf("a[%d]", y%5*5 + x%5)
 }
 
 func bfunc(x, y int) string {
-	return fmt.Sprintf("a%d%d", x%5, y%5)
+	return fmt.Sprintf("b%d%d", x%5, y%5)
 }
 
 var funcs = template.FuncMap{
@@ -77,9 +77,9 @@ var tmpl = template.Must(template.New("keccak").Funcs(funcs).Parse(`
 package keccak
 
 // round implements one round of the keccak-f[1600] permutation.
-func roundGo(a *[5][5]uint64) {
+func roundGo(a *[25]uint64) {
 	{{ range $x := count 5 }}
-		var a{{$x}}0, a{{$x}}1, a{{$x}}2, a{{$x}}3, a{{$x}}4 uint64
+		var b{{$x}}0, b{{$x}}1, b{{$x}}2, b{{$x}}3, b{{$x}}4 uint64
 	{{ end }}
 
 	// Theta
diff --git a/keccak.go b/keccak.go
index ee18b14..15eaf11 100644
--- a/keccak.go
+++ b/keccak.go
@@ -1,38 +1,37 @@
+
 package keccak
 
 // roundGeneric implements one round of the keccak-f[1600] permutation.
-func roundGeneric(a *[5][5]uint64) {
+func roundGeneric(a *[25]uint64) {
 	// Theta
 	var c [5]uint64
-	for x := range *a {
-		c[x] = a[0][x] ^ a[1][x] ^ a[2][x] ^ a[3][x] ^ a[4][x]
+	for x := range c {
+		c[x] = a[0+x] ^ a[5+x] ^ a[10+x] ^ a[15+x] ^ a[20+x]
 	}
-	for x := range a[0] {
-		x0, x1 := (x+4)%5, (x+1)%5
-		a[0][x] ^= c[x0] ^ rotl(c[x1], 1)
-		a[1][x] ^= c[x0] ^ rotl(c[x1], 1)
-		a[2][x] ^= c[x0] ^ rotl(c[x1], 1)
-		a[3][x] ^= c[x0] ^ rotl(c[x1], 1)
-		a[4][x] ^= c[x0] ^ rotl(c[x1], 1)
+	for x := 0; x < 5; x++ {
+		d := c[(x+4)%5] ^ rotl(c[(x+1)%5], 1)
+		for y := 0; y < 5; y++ {
+			a[y*5+x] ^= d
+		}
 	}
 
 	// Rho and pi
 	var b [5][5]uint64
-	for y := range *a {
-		for x := range a[0] {
+	for y := 0; y < 5; y++ {
+		for x := 0; x < 5; x++ {
 			x0 := y
 			y0 := (x*2 + y*3) % 5
-			b[y0][x0] = rotl(a[y][x], rotc[y][x])
+			b[y0][x0] = rotl(a[y*5+x], rotc[y][x])
 		}
 	}
 
 	// Chi
-	for y := range a[0] {
-		a[y][0] = b[y][0] ^ ^b[y][1]&b[y][2]
-		a[y][1] = b[y][1] ^ ^b[y][2]&b[y][3]
-		a[y][2] = b[y][2] ^ ^b[y][3]&b[y][4]
-		a[y][3] = b[y][3] ^ ^b[y][4]&b[y][0]
-		a[y][4] = b[y][4] ^ ^b[y][0]&b[y][1]
+	for y := 0; y < 5; y++ {
+		a[y*5+0] = b[y][0] ^ ^b[y][1]&b[y][2]
+		a[y*5+1] = b[y][1] ^ ^b[y][2]&b[y][3]
+		a[y*5+2] = b[y][2] ^ ^b[y][3]&b[y][4]
+		a[y*5+3] = b[y][3] ^ ^b[y][4]&b[y][0]
+		a[y*5+4] = b[y][4] ^ ^b[y][0]&b[y][1]
 	}
 }
 
diff --git a/keccak_gen.go b/keccak_gen.go
index 9a1fd9b..a6169d6 100644
--- a/keccak_gen.go
+++ b/keccak_gen.go
@@ -4,235 +4,235 @@
 package keccak
 
 // round implements one round of the keccak-f[1600] permutation.
-func roundGo(a *[5][5]uint64) {
+func roundGo(a *[25]uint64) {
 
-	var a00, a01, a02, a03, a04 uint64
+	var b00, b01, b02, b03, b04 uint64
 
-	var a10, a11, a12, a13, a14 uint64
+	var b10, b11, b12, b13, b14 uint64
 
-	var a20, a21, a22, a23, a24 uint64
+	var b20, b21, b22, b23, b24 uint64
 
-	var a30, a31, a32, a33, a34 uint64
+	var b30, b31, b32, b33, b34 uint64
 
-	var a40, a41, a42, a43, a44 uint64
+	var b40, b41, b42, b43, b44 uint64
 
 	// Theta
 	var c0, c1, c2, c3, c4 uint64
 
-	c0 = a[0][0]
+	c0 = a[0]
 
-	c1 = a[0][1]
+	c1 = a[1]
 
-	c2 = a[0][2]
+	c2 = a[2]
 
-	c3 = a[0][3]
+	c3 = a[3]
 
-	c4 = a[0][4]
+	c4 = a[4]
 
-	c0 ^= a[1][0]
+	c0 ^= a[5]
 
-	c1 ^= a[1][1]
+	c1 ^= a[6]
 
-	c2 ^= a[1][2]
+	c2 ^= a[7]
 
-	c3 ^= a[1][3]
+	c3 ^= a[8]
 
-	c4 ^= a[1][4]
+	c4 ^= a[9]
 
-	c0 ^= a[2][0]
+	c0 ^= a[10]
 
-	c1 ^= a[2][1]
+	c1 ^= a[11]
 
-	c2 ^= a[2][2]
+	c2 ^= a[12]
 
-	c3 ^= a[2][3]
+	c3 ^= a[13]
 
-	c4 ^= a[2][4]
+	c4 ^= a[14]
 
-	c0 ^= a[3][0]
+	c0 ^= a[15]
 
-	c1 ^= a[3][1]
+	c1 ^= a[16]
 
-	c2 ^= a[3][2]
+	c2 ^= a[17]
 
-	c3 ^= a[3][3]
+	c3 ^= a[18]
 
-	c4 ^= a[3][4]
+	c4 ^= a[19]
 
-	c0 ^= a[4][0]
+	c0 ^= a[20]
 
-	c1 ^= a[4][1]
+	c1 ^= a[21]
 
-	c2 ^= a[4][2]
+	c2 ^= a[22]
 
-	c3 ^= a[4][3]
+	c3 ^= a[23]
 
-	c4 ^= a[4][4]
+	c4 ^= a[24]
 
 	var d uint64
 
 	d = c4 ^ (c1<<1 | c1>>63)
 
-	a00 = a[0][0] ^ d
+	b00 = a[0] ^ d
 
-	a01 = a[1][0] ^ d
+	b01 = a[5] ^ d
 
-	a02 = a[2][0] ^ d
+	b02 = a[10] ^ d
 
-	a03 = a[3][0] ^ d
+	b03 = a[15] ^ d
 
-	a04 = a[4][0] ^ d
+	b04 = a[20] ^ d
 
 	d = c0 ^ (c2<<1 | c2>>63)
 
-	a10 = a[0][1] ^ d
+	b10 = a[1] ^ d
 
-	a11 = a[1][1] ^ d
+	b11 = a[6] ^ d
 
-	a12 = a[2][1] ^ d
+	b12 = a[11] ^ d
 
-	a13 = a[3][1] ^ d
+	b13 = a[16] ^ d
 
-	a14 = a[4][1] ^ d
+	b14 = a[21] ^ d
 
 	d = c1 ^ (c3<<1 | c3>>63)
 
-	a20 = a[0][2] ^ d
+	b20 = a[2] ^ d
 
-	a21 = a[1][2] ^ d
+	b21 = a[7] ^ d
 
-	a22 = a[2][2] ^ d
+	b22 = a[12] ^ d
 
-	a23 = a[3][2] ^ d
+	b23 = a[17] ^ d
 
-	a24 = a[4][2] ^ d
+	b24 = a[22] ^ d
 
 	d = c2 ^ (c4<<1 | c4>>63)
 
-	a30 = a[0][3] ^ d
+	b30 = a[3] ^ d
 
-	a31 = a[1][3] ^ d
+	b31 = a[8] ^ d
 
-	a32 = a[2][3] ^ d
+	b32 = a[13] ^ d
 
-	a33 = a[3][3] ^ d
+	b33 = a[18] ^ d
 
-	a34 = a[4][3] ^ d
+	b34 = a[23] ^ d
 
 	d = c3 ^ (c0<<1 | c0>>63)
 
-	a40 = a[0][4] ^ d
+	b40 = a[4] ^ d
 
-	a41 = a[1][4] ^ d
+	b41 = a[9] ^ d
 
-	a42 = a[2][4] ^ d
+	b42 = a[14] ^ d
 
-	a43 = a[3][4] ^ d
+	b43 = a[19] ^ d
 
-	a44 = a[4][4] ^ d
+	b44 = a[24] ^ d
 
 	// Rho
 
-	a00 = a00<<0 | a00>>64
+	b00 = b00<<0 | b00>>64
 
-	a10 = a10<<1 | a10>>63
+	b10 = b10<<1 | b10>>63
 
-	a20 = a20<<62 | a20>>2
+	b20 = b20<<62 | b20>>2
 
-	a30 = a30<<28 | a30>>36
+	b30 = b30<<28 | b30>>36
 
-	a40 = a40<<27 | a40>>37
+	b40 = b40<<27 | b40>>37
 
-	a01 = a01<<36 | a01>>28
+	b01 = b01<<36 | b01>>28
 
-	a11 = a11<<44 | a11>>20
+	b11 = b11<<44 | b11>>20
 
-	a21 = a21<<6 | a21>>58
+	b21 = b21<<6 | b21>>58
 
-	a31 = a31<<55 | a31>>9
+	b31 = b31<<55 | b31>>9
 
-	a41 = a41<<20 | a41>>44
+	b41 = b41<<20 | b41>>44
 
-	a02 = a02<<3 | a02>>61
+	b02 = b02<<3 | b02>>61
 
-	a12 = a12<<10 | a12>>54
+	b12 = b12<<10 | b12>>54
 
-	a22 = a22<<43 | a22>>21
+	b22 = b22<<43 | b22>>21
 
-	a32 = a32<<25 | a32>>39
+	b32 = b32<<25 | b32>>39
 
-	a42 = a42<<39 | a42>>25
+	b42 = b42<<39 | b42>>25
 
-	a03 = a03<<41 | a03>>23
+	b03 = b03<<41 | b03>>23
 
-	a13 = a13<<45 | a13>>19
+	b13 = b13<<45 | b13>>19
 
-	a23 = a23<<15 | a23>>49
+	b23 = b23<<15 | b23>>49
 
-	a33 = a33<<21 | a33>>43
+	b33 = b33<<21 | b33>>43
 
-	a43 = a43<<8 | a43>>56
+	b43 = b43<<8 | b43>>56
 
-	a04 = a04<<18 | a04>>46
+	b04 = b04<<18 | b04>>46
 
-	a14 = a14<<2 | a14>>62
+	b14 = b14<<2 | b14>>62
 
-	a24 = a24<<61 | a24>>3
+	b24 = b24<<61 | b24>>3
 
-	a34 = a34<<56 | a34>>8
+	b34 = b34<<56 | b34>>8
 
-	a44 = a44<<14 | a44>>50
+	b44 = b44<<14 | b44>>50
 
 	// Pi / Chi / output
 
-	a[0][0] = a00 ^ (a22 &^ a11)
+	a[0] = b00 ^ (b22 &^ b11)
 
-	a[0][1] = a11 ^ (a33 &^ a22)
+	a[1] = b11 ^ (b33 &^ b22)
 
-	a[0][2] = a22 ^ (a44 &^ a33)
+	a[2] = b22 ^ (b44 &^ b33)
 
-	a[0][3] = a33 ^ (a00 &^ a44)
+	a[3] = b33 ^ (b00 &^ b44)
 
-	a[0][4] = a44 ^ (a11 &^ a00)
+	a[4] = b44 ^ (b11 &^ b00)
 
-	a[1][0] = a30 ^ (a02 &^ a41)
+	a[5] = b30 ^ (b02 &^ b41)
 
-	a[1][1] = a41 ^ (a13 &^ a02)
+	a[6] = b41 ^ (b13 &^ b02)
 
-	a[1][2] = a02 ^ (a24 &^ a13)
+	a[7] = b02 ^ (b24 &^ b13)
 
-	a[1][3] = a13 ^ (a30 &^ a24)
+	a[8] = b13 ^ (b30 &^ b24)
 
-	a[1][4] = a24 ^ (a41 &^ a30)
+	a[9] = b24 ^ (b41 &^ b30)
 
-	a[2][0] = a10 ^ (a32 &^ a21)
+	a[10] = b10 ^ (b32 &^ b21)
 
-	a[2][1] = a21 ^ (a43 &^ a32)
+	a[11] = b21 ^ (b43 &^ b32)
 
-	a[2][2] = a32 ^ (a04 &^ a43)
+	a[12] = b32 ^ (b04 &^ b43)
 
-	a[2][3] = a43 ^ (a10 &^ a04)
+	a[13] = b43 ^ (b10 &^ b04)
 
-	a[2][4] = a04 ^ (a21 &^ a10)
+	a[14] = b04 ^ (b21 &^ b10)
 
-	a[3][0] = a40 ^ (a12 &^ a01)
+	a[15] = b40 ^ (b12 &^ b01)
 
-	a[3][1] = a01 ^ (a23 &^ a12)
+	a[16] = b01 ^ (b23 &^ b12)
 
-	a[3][2] = a12 ^ (a34 &^ a23)
+	a[17] = b12 ^ (b34 &^ b23)
 
-	a[3][3] = a23 ^ (a40 &^ a34)
+	a[18] = b23 ^ (b40 &^ b34)
 
-	a[3][4] = a34 ^ (a01 &^ a40)
+	a[19] = b34 ^ (b01 &^ b40)
 
-	a[4][0] = a20 ^ (a42 &^ a31)
+	a[20] = b20 ^ (b42 &^ b31)
 
-	a[4][1] = a31 ^ (a03 &^ a42)
+	a[21] = b31 ^ (b03 &^ b42)
 
-	a[4][2] = a42 ^ (a14 &^ a03)
+	a[22] = b42 ^ (b14 &^ b03)
 
-	a[4][3] = a03 ^ (a20 &^ a14)
+	a[23] = b03 ^ (b20 &^ b14)
 
-	a[4][4] = a14 ^ (a31 &^ a20)
+	a[24] = b14 ^ (b31 &^ b20)
 
 }
diff --git a/sponge.go b/sponge.go
index 0b12c86..e1ededb 100644
--- a/sponge.go
+++ b/sponge.go
@@ -10,7 +10,7 @@ var round = roundGo
 
 // digest implements hash.Hash
 type digest struct {
-	a      [5][5]uint64 // a[y][x][z]
+	a      [25]uint64 // a[y][x][z]
 	buf    [200]byte
 	dsbyte byte
 	len    int
@@ -28,7 +28,7 @@ func (d *digest) BlockSize() int { return 200 - d.size*2 }
 
 func (d *digest) Reset() {
 	//fmt.Println("resetting")
-	d.a = [5][5]uint64{}
+	d.a = [25]uint64{}
 	d.buf = [200]byte{}
 	d.len = 0
 }
@@ -51,23 +51,21 @@ func (d *digest) flush() {
 	//fmt.Printf("Flushing with %d bytes\n", d.len)
 	b := d.buf[:d.len]
 loop:
-	for y := range d.a {
-		for x := range d.a[0] {
+	for i := range d.a {
 			if len(b) == 0 {
 				break loop
 			}
-			d.a[y][x] ^= le64dec(b)
+			d.a[i] ^= le64dec(b)
 			b = b[8:]
-		}
 	}
 	keccakf(&d.a)
 	d.len = 0
 }
 
-func keccakf(a *[5][5]uint64) {
+func keccakf(a *[25]uint64) {
 	for i := 0; i < 24; i++ {
 		round(a)
-		a[0][0] ^= roundc[i]
+		a[0] ^= roundc[i]
 	}
 }
 
@@ -83,7 +81,7 @@ func (d0 *digest) Sum(b []byte) []byte {
 	d.flush()
 
 	for i := 0; i < d.size/8; i++ {
-		b = le64enc(b, d.a[i/5][i%5])
+		b = le64enc(b, d.a[i])
 	}
 	return b
 }