Hi there,

On Thu, 11 Jan 2001, adrien mistretta wrote:

> Thanks, it works fine, but now i got another problem, some errors on the 
> speed test, and the time test seems to be false 

if you read the errors you'll see what the problem is;

> Doing 4096 bit private rsa's for 10s: 64 4096 bit private RSA's in 0.10s
> RSA verify failure.  No RSA verify will be done.
> 13839:error:2606607A:engine routines:CSWIFT_MOD_EXP:size too large or too 
> small:hw_cswift.c:391:

CryptoSwift support doesn't deal with 4096-bit RSA keys (or larger). I
think 2048-bit is the maximum. If you call "speed rsa1024" then you only
get tests for that size, for example. You called it for "rsa" which means
it tried every size and thus flunked on the 4096-bit case. That's why we
have string-based error stacks, sometimes they contain important clues :-)

As for the time tests ...

>                   sign    verify    sign/s verify/s
> rsa 1024 bits   0.0000s   0.0000s  20208.5  33839.1
> rsa 2048 bits   0.0000s   0.0000s  28773.8  23351.1
> rsa 4096 bits   0.0001s   0.0059s  10163.6    168.2

That's because it's measuring CPU time ... as most of the work is
happening inside dedicated hardware the program itself only occupies the
CPU for a few milliseconds during the 10 seconds or so that the test runs,
hence the inflated numbers. If you use the switch "-elapsed" on the
command line you will see it just measure elapsed time which is more
meaningful in this case. (And it will display a little message about the
benchmarking that may give you a hint as to why normally measuring CPU
usage rather than elapsed time is a better idea, at least in the
software-only world).

As for the DSA problem ... not sure off the top of my head. Could you
please first check that "speed dsa1024" works? If that does, then it's
something subtle about the code/API, if it doesn't work it's probably
something more basic. Maybe someone else try this (ie. someone who has a
CryptoSwift card? I don't have one so can't really second guess what's
going on here without more information).

Cheers,
Geoff

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [EMAIL PROTECTED]
Automated List Manager                           [EMAIL PROTECTED]

Reply via email to