Crypto++ 5.6.0 Benchmarks

Here are speed benchmarks for some of the most commonly used cryptographic algorithms. All were coded in C++, compiled with Microsoft Visual C++ 2005 SP1 (whole program optimization, optimize for speed), and ran on an Intel Core 2 1.83 GHz processor under Windows Vista in 32-bit mode. x86/MMX/SSE2 assembly language routines were used for integer arithmetic, AES, VMAC, Sosemanuk, Panama, Salsa20, SHA-256, SHA-512, Tiger, and Whirlpool. (OpenMP support was disabled so only one core of the CPU was used for this benchmark.) The results from the same binary running on an Intel Pentium 4 (Prescott) CPU are available here.

Also available are benchmarks that ran on an AMD Opteron 8354 2.2 GHz processor under Linux. Those were compiled with GCC 4.1.2 using -O3 optimization, and x86-64/MMX/SSE2 assembly language routines were used for integer arithmetic, AES, VMAC, Sosemanuk, Panama, Salsa20, SHA-256, and Whirlpool.

Algorithm	MiB/Second	Cycles Per Byte	Microseconds to Setup Key and IV	Cycles to Setup Key and IV
AES/GCM (2K tables)	102	17.2	2.946	5391
AES/GCM (64K tables)	108	16.1	11.546	21130
AES/CCM	61	28.6	0.888	1625
AES/EAX	61	28.8	1.757	3216
GMAC(AES) (2K tables)	417	4.2	2.944	5388
GMAC(AES) (64K tables)	540	3.2	11.523	21088
VMAC(AES)-64	1519	1.1	3.133	5734
VMAC(AES)-128	779	2.2	3.738	6841
HMAC(SHA-1)	147	11.9	0.509	932
Two-Track-MAC	105	16.6	0.041	75
CMAC(AES)	109	16.1	0.600	1098
DMAC(AES)	109	16.0	1.761	3223
CRC32	253	6.9
Adler32	920	1.9
MD5	255	6.8
SHA-1	153	11.4
SHA-256	111	15.8
SHA-512	99	17.7
Tiger	214	8.1
Whirlpool	57	30.5
RIPEMD-160	106	16.5
RIPEMD-320	110	15.9
RIPEMD-128	153	11.4
RIPEMD-256	158	11.1
Panama-LE	843	2.1	1.695	3103
Panama-BE	397	4.4	2.703	4946
Salsa20	408	4.3	0.390	714
Salsa20/12	643	2.7	0.483	884
Salsa20/8	887	2.0	0.481	881
Sosemanuk	727	2.4	1.240	2269
MARC4	126	13.9	2.690	4923
SEAL-3.0-LE	375	4.7	63.697	116566
WAKE-OFB-LE	197	8.9	2.816	5153
AES/CTR (128-bit key)	139	12.6	0.698	1277
AES/CTR (192-bit key)	113	15.4	0.707	1293
AES/CTR (256-bit key)	96	18.2	0.756	1383
AES/CBC (128-bit key)	109	16.0	0.569	1041
AES/CBC (192-bit key)	92	18.9	0.572	1046
AES/CBC (256-bit key)	80	21.7	0.619	1133
AES/OFB (128-bit key)	103	16.9	0.702	1285
AES/CFB (128-bit key)	108	16.1	0.926	1695
AES/ECB (128-bit key)	109	16.0	0.253	462
Camellia/CTR (128-bit key)	48	36.3	0.632	1156
Camellia/CTR (256-bit key)	37	47.4	0.696	1273
Twofish/CTR	59	29.4	7.716	14121
Serpent/CTR	32	54.7	1.197	2191
CAST-256/CTR	37	47.1	2.659	4865
RC6/CTR	101	17.3	2.802	5128
MARS/CTR	47	37.2	3.516	6435
SHACAL-2/CTR (128-bit key)	53	32.9	0.896	1639
SHACAL-2/CTR (512-bit key)	54	32.6	0.919	1681
DES/CTR	32	54.7	8.372	15320
DES-XEX3/CTR	29	60.6	8.309	15206
DES-EDE3/CTR	13	134.5	27.317	49989
IDEA/CTR	35	49.9	0.698	1277
RC5 (r=16)	75	23.4	2.549	4665
Blowfish/CTR	58	30.0	62.683	114710
TEA/CTR	27	65.1	0.638	1167
XTEA/CTR	26	67.4	0.636	1165
CAST-128/CTR	55	31.9	1.008	1844
SKIPJACK/CTR	10	180.5	3.456	6324
SEED/CTR (1/2 K table)	29	59.2	0.762	1394

Operation	Milliseconds/Operation	Megacycles/Operation
RSA 1024 Encryption	0.08	0.14
RSA 1024 Decryption	1.46	2.68
LUC 1024 Encryption	0.08	0.14
LUC 1024 Decryption	2.49	4.55
DLIES 1024 Encryption	0.85	1.56
DLIES 1024 Encryption with precomputation	1.49	2.72
DLIES 1024 Decryption	1.18	2.17
LUCELG 512 Encryption	0.58	1.05
LUCELG 512 Encryption with precomputation	0.58	1.05
LUCELG 512 Decryption	0.65	1.18
RSA 2048 Encryption	0.16	0.29
RSA 2048 Decryption	6.08	11.12
LUC 2048 Encryption	0.18	0.33
LUC 2048 Decryption	9.89	18.10
DLIES 2048 Encryption	4.11	7.52
DLIES 2048 Encryption with precomputation	4.54	8.30
DLIES 2048 Decryption	3.86	7.07
LUCELG 1024 Encryption	1.89	3.45
LUCELG 1024 Encryption with precomputation	1.88	3.45
LUCELG 1024 Decryption	1.73	3.17
RSA 1024 Signature	1.48	2.71
RSA 1024 Verification	0.07	0.13
RW 1024 Signature	2.03	3.72
RW 1024 Verification	0.05	0.08
LUC 1024 Signature	2.49	4.56
LUC 1024 Verification	0.08	0.14
NR 1024 Signature	0.45	0.82
NR 1024 Signature with precomputation	0.42	0.76
NR 1024 Verification	0.51	0.92
NR 1024 Verification with precomputation	0.70	1.29
DSA 1024 Signature	0.45	0.83
DSA 1024 Signature with precomputation	0.42	0.78
DSA 1024 Verification	0.52	0.94
DSA 1024 Verification with precomputation	0.67	1.23
LUC-HMP 512 Signature	0.58	1.06
LUC-HMP 512 Signature with precomputation	0.58	1.06
LUC-HMP 512 Verification	0.60	1.09
LUC-HMP 512 Verification with precomputation	0.60	1.09
ESIGN 1023 Signature	0.22	0.40
ESIGN 1023 Verification	0.07	0.12
ESIGN 1536 Signature	0.38	0.70
ESIGN 1536 Verification	0.14	0.26
RSA 2048 Signature	6.05	11.06
RSA 2048 Verification	0.16	0.29
RW 2048 Signature	7.37	13.49
RW 2048 Verification	0.10	0.18
LUC 2048 Signature	9.92	18.16
LUC 2048 Verification	0.18	0.32
NR 2048 Signature	2.09	3.83
NR 2048 Signature with precomputation	0.93	1.71
NR 2048 Verification	2.39	4.37
NR 2048 Verification with precomputation	1.57	2.88
LUC-HMP 1024 Signature	1.91	3.50
LUC-HMP 1024 Signature with precomputation	1.92	3.51
LUC-HMP 1024 Verification	2.02	3.70
LUC-HMP 1024 Verification with precomputation	1.97	3.60
ESIGN 2046 Signature	0.48	0.88
ESIGN 2046 Verification	0.15	0.28
XTR-DH 171 Key-Pair Generation	0.78	1.43
XTR-DH 171 Key Agreement	1.55	2.83
XTR-DH 342 Key-Pair Generation	1.87	3.43
XTR-DH 342 Key Agreement	3.74	6.85
DH 1024 Key-Pair Generation	0.45	0.82
DH 1024 Key-Pair Generation with precomputation	0.83	1.51
DH 1024 Key Agreement	1.18	2.16
DH 2048 Key-Pair Generation	2.14	3.92
DH 2048 Key-Pair Generation with precomputation	2.40	4.39
DH 2048 Key Agreement	3.84	7.03
LUCDIF 512 Key-Pair Generation	0.30	0.55
LUCDIF 512 Key-Pair Generation with precomputation	0.30	0.55
LUCDIF 512 Key Agreement	0.64	1.18
LUCDIF 1024 Key-Pair Generation	0.98	1.79
LUCDIF 1024 Key-Pair Generation with precomputation	0.98	1.79
LUCDIF 1024 Key Agreement	1.73	3.17
MQV 1024 Key-Pair Generation	0.43	0.78
MQV 1024 Key-Pair Generation with precomputation	0.40	0.73
MQV 1024 Key Agreement	0.91	1.66
MQV 2048 Key-Pair Generation	2.09	3.83
MQV 2048 Key-Pair Generation with precomputation	0.92	1.68
MQV 2048 Key Agreement	3.97	7.27
ECIES over GF(p) 256 Encryption	5.65	10.34
ECIES over GF(p) 256 Encryption with precomputation	4.21	7.70
ECIES over GF(p) 256 Decryption	3.98	7.29
ECDSA over GF(p) 256 Signature	2.88	5.27
ECDSA over GF(p) 256 Signature with precomputation	2.14	3.92
ECDSA over GF(p) 256 Verification	8.53	15.61
ECDSA over GF(p) 256 Verification with precomputation	3.58	6.56
ECDHC over GF(p) 256 Key-Pair Generation	2.87	5.25
ECDHC over GF(p) 256 Key-Pair Generation with precomputation	2.11	3.87
ECDHC over GF(p) 256 Key Agreement	2.82	5.17
ECMQVC over GF(p) 256 Key-Pair Generation	2.84	5.20
ECMQVC over GF(p) 256 Key-Pair Generation with precomputation	2.19	4.01
ECMQVC over GF(p) 256 Key Agreement	8.31	15.21
ECIES over GF(2^n) 233 Encryption	21.17	38.74
ECIES over GF(2^n) 233 Encryption with precomputation	6.08	11.13
ECIES over GF(2^n) 233 Decryption	12.15	22.23
ECDSA over GF(2^n) 233 Signature	10.62	19.43
ECDSA over GF(2^n) 233 Signature with precomputation	3.08	5.64
ECDSA over GF(2^n) 233 Verification	12.80	23.43
ECDSA over GF(2^n) 233 Verification with precomputation	5.30	9.69
ECDHC over GF(2^n) 233 Key-Pair Generation	10.51	19.23
ECDHC over GF(2^n) 233 Key-Pair Generation with precomputation	3.03	5.55
ECDHC over GF(2^n) 233 Key Agreement	10.72	19.61
ECMQVC over GF(2^n) 233 Key-Pair Generation	10.49	19.20
ECMQVC over GF(2^n) 233 Key-Pair Generation with precomputation	3.04	5.57
ECMQVC over GF(2^n) 233 Key Agreement	13.20	24.16

Notes

Crypto++ 5.6 Benchmark Average for this platform: 377.01. This number is computed by taking the geometric mean of the number of MB/second of each cipher, hash function, and MAC, and operations/second of each asymmetric operation listed above.
RSA and LUC use 17 as the public exponent.
DH and ElGamal encryption and decryption use short exponents to save time. The size of the secret exponents were chosen so that a meet-in-the-middle attack would be slower than the general discrete log algorithm (NFS). The sizes used were:

modulus exponent
1024 164
2048 226
EC means elliptic curve. Operations in GF(2^n) are implemented using trinomial basis. Note that compared to other algorithms listed here, Crypto++'s implementation of EC over GF(2^n) is less optimized.
All tests were done by repeating the crypto operations over small blocks of random data. In practice you will likely see slower speeds because time is needed to transfer data to and from memory.
For the ciphers that specify big endian byte order, the timing data listed include time needed to convert to and from little endian order.
The RSA, RW, DH, MQV, and elliptic curve schemes come from the IEEE P1363 standard. For more info see http://grouper.ieee.org/groups/1363/index.html.
Precomputation means using a table of 16 precomputed powers of each fixed base to speed up exponentiation.
Source code for these benchmarks is available as part of the Crypto++ library.

modulus	exponent
1024	164
2048	226

Written by: Wei Dai Last modified: 3/31/2009