diff --git a/bip32.c b/bip32.c
index 8ecfc6f73..d9bda71af 100644
--- a/bip32.c
+++ b/bip32.c
@@ -264,6 +264,45 @@ int hdnode_public_ckd(HDNode *inout, uint32_t i)
 	return 1;
 }
 
+int hdnode_public_ckd_address_optimized(const HDNode *in, const curve_point *pub, uint32_t i, uint8_t version, char *addr, int addrsize)
+{
+	uint8_t data[1 + 32 + 4];
+	uint8_t I[32 + 32];
+	uint8_t public_key[33];
+	curve_point b;
+	bignum256 c;
+
+	if (i & 0x80000000) { // private derivation
+		return 0;
+	}
+	memcpy(data, in->public_key, 33);
+	write_be(data + 33, i);
+
+	while (true) {
+		bool failed = false;
+		hmac_sha512(in->chain_code, 32, data, sizeof(data), I);
+		bn_read_be(I, &c);
+		if (!bn_is_less(&c, &in->curve->params->order)) { // >= order
+			failed = true;
+		} else {
+			scalar_multiply(in->curve->params, &c, &b); // b = c * G
+			point_add(in->curve->params, pub, &b);       // b = a + b
+			if (point_is_infinity(&b)) {
+				failed = true;
+			}
+		}
+		if (!failed) {
+			public_key[0] = 0x02 | (b.y.val[0] & 0x01);
+			bn_write_be(&b.x, public_key + 1);
+			break;
+		}
+		data[0] = 1;
+		memcpy(data + 1, I + 32, 32);
+	}
+	ecdsa_get_address(public_key, version, addr, addrsize);
+	return 1;
+}
+
 #if USE_BIP32_CACHE
 
 static bool private_ckd_cache_root_set = false;
diff --git a/bip32.h b/bip32.h
index 2f7dca3e1..ed45498bb 100644
--- a/bip32.h
+++ b/bip32.h
@@ -56,6 +56,8 @@ int hdnode_private_ckd(HDNode *inout, uint32_t i);
 
 int hdnode_public_ckd(HDNode *inout, uint32_t i);
 
+int hdnode_public_ckd_address_optimized(const HDNode *in, const curve_point *pub, uint32_t i, uint8_t version, char *addr, int addrsize);
+
 #if USE_BIP32_CACHE
 
 int hdnode_private_ckd_cached(HDNode *inout, const uint32_t *i, size_t i_count);
diff --git a/test_speed.c b/test_speed.c
index 0541ef246..75ccb4d8b 100644
--- a/test_speed.c
+++ b/test_speed.c
@@ -6,6 +6,7 @@
 #include "curves.h"
 #include "ecdsa.h"
 #include "secp256k1.h"
+#include "nist256p1.h"
 #include "ed25519.h"
 
 uint8_t msg[32];
@@ -34,6 +35,23 @@ void bench_secp256k1(void) {
 	printf("SECP256k1 verifying speed: %0.2f sig/s\n", 500.0f / ((float)(clock() - t) / CLOCKS_PER_SEC));
 }
 
+void bench_nist256p1(void) {
+	uint8_t sig[64], pub[33], priv[32], pby;
+
+	const ecdsa_curve *curve = &nist256p1;
+
+	memcpy(priv, "\xc5\x5e\xce\x85\x8b\x0d\xdd\x52\x63\xf9\x68\x10\xfe\x14\x43\x7c\xd3\xb5\xe1\xfb\xd7\xc6\xa2\xec\x1e\x03\x1f\x05\xe8\x6d\x8b\xd5", 32);
+	ecdsa_get_public_key33(curve, priv, pub);
+	ecdsa_sign(curve, priv, msg, sizeof(msg), sig, &pby);
+
+	clock_t t = clock();
+	for (int i = 0 ; i < 500; i++) {
+		int res = ecdsa_verify(curve, pub, sig, msg, sizeof(msg));
+		assert(res == 0);
+	}
+	printf("NIST256p1 verifying speed: %0.2f sig/s\n", 500.0f / ((float)(clock() - t) / CLOCKS_PER_SEC));
+}
+
 void bench_ed25519(void) {
 	ed25519_public_key pk;
 	ed25519_secret_key sk;
@@ -49,12 +67,58 @@ void bench_ed25519(void) {
 		assert(res == 0);
 	}
 	printf("Ed25519 verifying speed: %0.2f sig/s\n", 500.0f / ((float)(clock() - t) / CLOCKS_PER_SEC));
-
 }
 
-int main(void) {
+void test_verify_speed(void) {
 	prepare_msg();
 	bench_secp256k1();
+	bench_nist256p1();
 	bench_ed25519();
+}
+
+HDNode root;
+
+void prepare_node(void)
+{
+	hdnode_from_seed((uint8_t *)"NothingToSeeHere", 16, SECP256K1_NAME, &root);
+}
+
+void bench_ckd_normal(void) {
+	char addr[40];
+	clock_t t = clock();
+	for (int i = 0; i < 1000; i++) {
+		HDNode node = root;
+		hdnode_public_ckd(&node, i);
+		ecdsa_get_address(node.public_key, 0, addr, 40);
+		if (i == 0) {
+			printf("address = %s\n", addr);
+		}
+	}
+	printf("CKD normal speed: %0.2f iter/s\n", 1000.0f / ((float)(clock() - t) / CLOCKS_PER_SEC));
+}
+
+void bench_ckd_optimized(void) {
+	char addr[40];
+	curve_point pub;
+	ecdsa_read_pubkey(root.curve->params, root.public_key, &pub);
+	clock_t t = clock();
+	for (int i = 0; i < 1000; i++) {
+		hdnode_public_ckd_address_optimized(&root, &pub, i, 0, addr, 40);
+		if (i == 0) {
+			printf("address = %s\n", addr);
+		}
+	}
+	printf("CKD optim speed: %0.2f iter/s\n", 1000.0f / ((float)(clock() - t) / CLOCKS_PER_SEC));
+}
+
+void test_ckd_speed(void) {
+	prepare_node();
+	bench_ckd_normal();
+	bench_ckd_optimized();
+}
+
+int main(void) {
+	test_verify_speed();
+	test_ckd_speed();
 	return 0;
 }