/pyCryptoBenchmarking

benchmarking the python pyCrypto and cryptography moduls

Primary LanguagePythonMIT LicenseMIT

results

symmetric crypto

In iterations per second.

pycrypt cryptography factor algorithm
756499.3 379890.2 0.50 AES
424332.1 259299.5 0.61 3DES.64
229043.5 259282.9 1.13 3DES.128
226057.2 255484.5 1.13 3DES.192
487464.4 302878.6 0.62 Blowfish
351429.9 317410.7 0.90 CAST5
797340.5 361095.3 0.45 RC4

Most symmetric ciphers are significantly faster in pyCrypto.

hash algorithms

message size = 64 kbyte, in Mbyte/sec

pycrypt cryptography factor algorithm
57 343 5.98 sha224
53 333 6.19 sha256
83 487 5.85 sha384
81 487 6.00 sha512
910 781 0.86 sha1
678 419 0.62 md5
40 199 4.89 ripemd160

message size = 32 bytes, in Mbyte/sec

pycrypt cryptography factor algorithm
16.2 2.8 0.17 sha224
15.7 2.9 0.18 sha256
13.0 2.8 0.21 sha384
12.4 2.8 0.22 sha512
15.3 2.8 0.19 sha1
16.2 2.8 0.18 md5
14.6 1.5 0.10 ripemd160

Both pycrypto and cryptography have some call overhead. for pycrypto the byterate stabilizes for messages over 1K, while for `cryptography this happens for messages over 16K.

That said, cryptography is generally faster for very large messages, while pycrypto is fast for smaller messages. For the SHAxxx algorithms, the cross over point is around 1k mesage size. for SHA1 and MD5 pycrypto is always faster.

asymmetric crypto

pow() pycrypt cryptography factor algorithm
17267.1 12137.1 33917.4 2.8 rsa.1024
4920.6 4217.5 18242.9 4.3 rsa.2048
1440.8 1467.3 6448.8 4.4 rsa.4096

So for RSA the cryptography library is generally faster. The pycrypt performance is roughly equal to using the pow() function.

random numbers

in Mbyte/sec

random sysrand pycrypti r/s s/p msgsize
478.0 55.9 2.0 8.6 27.4 32
726.4 70.9 3.1 10.2 22.6 64
1148.9 81.0 3.8 14.2 21.1 128
1476.4 88.1 4.4 16.8 20.1 256
1757.5 93.4 3.9 18.8 24.2 512
2141.6 94.3 2.6 22.7 36.6 1024
2339.3 100.2 1.9 23.3 53.4 2048
2405.1 94.8 1.1 25.4 83.1 4096
2361.0 101.9 0.6 23.2 158.7 8192
2568.1 97.7 0.3 26.3 286.8 16384

The cryptography library does not have a builtin PRNG. Conclusion: secure random numbers are expensive.