possible mpi-pow improvement

[Werner Koch]

On Wed,  2 Oct 2013 02:41, gniibe at fsij.org said:

> Yes, at least on machine.

And on mine ;-)/

Before:

Algorithm         generate  100*sign  100*verify
------------------------------------------------
RSA 1024 bit          20ms     120ms        10ms
RSA 2048 bit         120ms     630ms        10ms
RSA 3072 bit         450ms    1780ms        10ms
RSA 4096 bit        4110ms    3700ms        30ms
DSA 1024/160             -      60ms        60ms
DSA 2048/224             -     210ms       200ms
DSA 3072/256             -     470ms       440ms

After:

Algorithm         generate  100*sign  100*verify
------------------------------------------------
RSA 1024 bit          10ms      80ms         0ms
RSA 2048 bit          60ms     370ms        10ms
RSA 3072 bit        1710ms    1060ms        10ms
RSA 4096 bit        1310ms    2220ms        20ms
DSA 1024/160             -      40ms        60ms
DSA 2048/224             -     140ms       200ms
DSA 3072/256             -     300ms       430ms


> With compilation option -DUSE_ALGORITHM_SIMPLE_EXPONENTIATION, we can
> use old implementation.
>
> Please test it out.

Please push your patch to master.


Salam-Shalom,

   Werner

-- 
Die Gedanken sind frei.  Ausnahmen regelt ein Bundesgesetz.