diff --git a/sm4/cipher_asm.go b/sm4/cipher_asm.go
index ca874445..0b65809c 100644
--- a/sm4/cipher_asm.go
+++ b/sm4/cipher_asm.go
@@ -71,6 +71,10 @@ func (c *sm4CipherAsm) Encrypt(dst, src []byte) {
 	if alias.InexactOverlap(dst[:BlockSize], src[:BlockSize]) {
 		panic("sm4: invalid buffer overlap")
 	}
+	c.encrypt(dst, src)
+}
+
+func (c *sm4CipherAsm) encrypt(dst, src []byte) {
 	if useAESNI4SingleBlock {
 		encryptBlockAsm(&c.enc[0], &dst[0], &src[0], INST_AES)
 	} else {
diff --git a/sm4/ctr_cipher_asm.go b/sm4/ctr_cipher_asm.go
index d316bbea..3e5dd35a 100644
--- a/sm4/ctr_cipher_asm.go
+++ b/sm4/ctr_cipher_asm.go
@@ -66,7 +66,8 @@ func (x *ctr) refill() {
 	copy(x.out, x.out[x.outUsed:])
 	x.out = x.out[:cap(x.out)]
 	for remain <= len(x.out)-x.b.blocksSize {
-		x.b.EncryptBlocks(x.out[remain:], x.ctr)
+		encryptBlocksAsm(&x.b.enc[0], x.out[remain:], x.ctr, INST_AES)
+
 		remain += x.b.blocksSize
 
 		// Generate complelte [x.b.batchBlocks] counters
diff --git a/sm4/gcm_cipher_asm.go b/sm4/gcm_cipher_asm.go
index f8106ccb..c5db35ce 100644
--- a/sm4/gcm_cipher_asm.go
+++ b/sm4/gcm_cipher_asm.go
@@ -93,7 +93,7 @@ func (g *gcm) Seal(dst, nonce, plaintext, data []byte) []byte {
 	var counter, tagMask [gcmBlockSize]byte
 	g.deriveCounter(&counter, nonce)
 
-	g.cipher.Encrypt(tagMask[:], counter[:])
+	g.cipher.encrypt(tagMask[:], counter[:])
 	gcmInc32(&counter)
 
 	g.counterCrypt(out, plaintext, &counter)
@@ -130,7 +130,7 @@ func (g *gcm) Open(dst, nonce, ciphertext, data []byte) ([]byte, error) {
 	var counter, tagMask [gcmBlockSize]byte
 	g.deriveCounter(&counter, nonce)
 
-	g.cipher.Encrypt(tagMask[:], counter[:])
+	g.cipher.encrypt(tagMask[:], counter[:])
 	gcmInc32(&counter)
 
 	var expectedTag [gcmTagSize]byte
diff --git a/sm4/sm4_gcm_asm.go b/sm4/sm4_gcm_asm.go
index 236dce4d..c8645808 100644
--- a/sm4/sm4_gcm_asm.go
+++ b/sm4/sm4_gcm_asm.go
@@ -80,7 +80,7 @@ func (g *gcmAsm) Seal(dst, nonce, plaintext, data []byte) []byte {
 		gcmSm4Finish(&g.bytesProductTable, &tagMask, &counter, uint64(len(nonce)), uint64(0))
 	}
 
-	g.cipher.Encrypt(tagMask[:], counter[:])
+	g.cipher.encrypt(tagMask[:], counter[:])
 
 	var tagOut [gcmTagSize]byte
 	gcmSm4Data(&g.bytesProductTable, data, &tagOut)
@@ -134,7 +134,7 @@ func (g *gcmAsm) Open(dst, nonce, ciphertext, data []byte) ([]byte, error) {
 		gcmSm4Finish(&g.bytesProductTable, &tagMask, &counter, uint64(len(nonce)), uint64(0))
 	}
 
-	g.cipher.Encrypt(tagMask[:], counter[:])
+	g.cipher.encrypt(tagMask[:], counter[:])
 
 	var expectedTag [gcmTagSize]byte
 	gcmSm4Data(&g.bytesProductTable, data, &expectedTag)
diff --git a/sm4/sm4ni_gcm_asm.go b/sm4/sm4ni_gcm_asm.go
index 0c50fe93..937e2e3d 100644
--- a/sm4/sm4ni_gcm_asm.go
+++ b/sm4/sm4ni_gcm_asm.go
@@ -73,7 +73,7 @@ func (g *gcmNI) Seal(dst, nonce, plaintext, data []byte) []byte {
 		gcmSm4Finish(&g.bytesProductTable, &tagMask, &counter, uint64(len(nonce)), uint64(0))
 	}
 
-	g.cipher.Encrypt(tagMask[:], counter[:])
+	encryptBlockAsm(&g.cipher.enc[0], &tagMask[0], &counter[0], INST_SM4)
 
 	var tagOut [gcmTagSize]byte
 	gcmSm4Data(&g.bytesProductTable, data, &tagOut)
@@ -127,7 +127,7 @@ func (g *gcmNI) Open(dst, nonce, ciphertext, data []byte) ([]byte, error) {
 		gcmSm4Finish(&g.bytesProductTable, &tagMask, &counter, uint64(len(nonce)), uint64(0))
 	}
 
-	g.cipher.Encrypt(tagMask[:], counter[:])
+	encryptBlockAsm(&g.cipher.enc[0], &tagMask[0], &counter[0], INST_SM4)
 
 	var expectedTag [gcmTagSize]byte
 	gcmSm4Data(&g.bytesProductTable, data, &expectedTag)