diff --git a/SM4-with-AESENCLAST.md b/SM4-with-AESENCLAST.md
index e306482..78ab4b1 100644
--- a/SM4-with-AESENCLAST.md
+++ b/SM4-with-AESENCLAST.md
@@ -21,6 +21,75 @@ My question is how to generate all (M1, C1, M2, C2)?
 {(M1, C1, M2, C2) | SM4-S(x) = A2(AES-S(A1(x)), A1(x) = M1*x + C1, A2(x) = M2*x + C2}
 ```
 
+**Evolution path**
+
+sm4_box_aesenclast <-> sm4_box_aesbox_1 <-> sm4_box_aesbox_2 <-> sm4_box_aesbox_3 <-> sm4_box_aesbox_4
+```
+sm4_box_aesenclast
+	y := mm_and_si128(x, const_0f)
+	y = mm_shuffle_epi8(a1l, y)
+	x = mm_srli_epi64(x, 4)
+	x = mm_and_si128(x, const_0f)
+	x = xor(mm_shuffle_epi8(a1h, x), y)
+
+	x = mm_shuffle_epi8(x, shift_row_inv)
+	x = mm_aesenclast_si128(x, const_0f)
+
+	y = mm_andnot_si128(x, const_0f)
+	y = mm_shuffle_epi8(a2l, y)
+	x = mm_srli_epi64(x, 4)
+	x = mm_and_si128(x, const_0f)
+	x = xor(mm_shuffle_epi8(a2h, x), y)
+
+sm4_box_aesbox_1 
+	var y __m128i
+	for i := 0; i < 16; i++ {
+		y.bytes[i] = a1l.bytes[x.bytes[i]&0xf] ^ a1h.bytes[x.bytes[i]>>4]
+	}
+	x = y
+
+	for i := 0; i < 16; i++ {
+		x.bytes[i] = aes_sbox[x.bytes[i]] ^ 0xf
+	}
+
+	for i := 0; i < 16; i++ {
+		y.bytes[i] = a2l.bytes[(^x.bytes[i])&0xf] ^ a2h.bytes[x.bytes[i]>>4]
+	}
+
+sm4_box_aesbox_2 
+	for i := 0; i < 16; i++ {
+		v := x.bytes[i]
+		v = a1l.bytes[v&0xf] ^ a1h.bytes[v>>4]
+		v = aes_sbox[v] ^ 0xf
+		v = a2l.bytes[^v&0xf] ^ a2h.bytes[v>>4]
+		x.bytes[i] = v
+	}
+
+sm4_box_aesbox_3 
+	for i := 0; i < 16; i++ {
+		v := x.bytes[i]
+		v = a1l.bytes[v&0xf] ^ a1h.bytes[v>>4]
+
+		v = aes_sbox[v]
+
+		v = a2l.bytes[v&0xf] ^ a2h.bytes[v>>4]
+
+		x.bytes[i] = v
+	}
+
+sm4_box_aesbox_4
+	for i := 0; i < 16; i++ {
+		v := x.bytes[i]
+		v = a1_table[v]
+
+		v = aes_sbox[v]
+
+		v = a2_table[v]
+
+		x.bytes[i] = v
+	} 
+```
+
 **How to calculate M, C from lookup table?**
 
     1.The first element of the table, T[0] should be the C.