pycryptopp benchmarks on ARM CPU

Okay, folks, pycryptopp is now working on one of our ARM CPU
buildbots, and I've run some measurements of the time to do different
crypto operations.

This machine has an ARM926EJ-S rev 0 (v5l) "Feroceon" at 266 MHz, with
32 KiB icache and 32 KiB dcache.

Here are a few crypto algorithm measurements:

Algorithm    MiB/Sec Cycle/Byte
---------    ------- ----------
MD5               24       10.7
SHA-1              9       28.5
SHA-256            6       42.5

Salsa20/12        13       19.0
AES/CTR (128-bit)  2      131.9

(By the way, the SHA-3 project looks like it will produce a hash
algorithm at least as fast as SHA-256 on this sort of CPU.  The
candidate which has the best performance on such a CPU so far is
"edonr256" which takes 17.5 cycles/byte on a similar ARM CPU:
http://bench.cr.yp.to/results-hash.html#arm-apollo .)

Algorithm          millisec/Op megacycle/Op
---------          ----------- ------------
RSA 2056 Sig            950.00       252.70
RSA 2056 Verify          12.82         3.41
ECDSA secp256k1 Sig      49.05        13.05
ECDSA secp256k1 Sig/pre  31.56         8.40
ECDSA secp256k1 Ver     128.75        34.25
ECDSA secp256k1 Ver/pre  51.00        13.57
ECDSA secp192r1 Sig      32.90         8.75
ECDSA secp192r1 Sig/pre  16.95         4.51
ECDSA secp192r1 Ver      58.82        15.65
ECDSA secp192r1 Ver/pre  28.86         7.68

And here is the big issue -- the main reason why we're switching to
ECDSA in the next release of Tahoe-LAFS (not the imminent v1.5
release).  It is the time it takes to generate a new key pair.  (I
suspect that the RSA keypair generation in Crypto++ is not optimized.)

Algorithm       millisec/new key
---------       ---------------
RSA 2056                31,670    (best)
RSA 2056                57,860    (mean)
RSA 2056                98,180    (worst)
ECDSA secp192r1              2.75 (best)
ECDSA secp192r1              2.79 (mean)
ECDSA secp192r1              2.89 (worst)

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the "Crypto++ Users" Google Group.
To unsubscribe, send an email to cryptopp-users-unsubscribe@....
More information about Crypto++ and this group is available at http://www.cryptopp.com.
-~----------~----~----~----~------~----~------~--~---