Request for help in research

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Request for help in research

Tanushree Banerjee

Hello,

I was looking for benchmarking the execution time of Elliptic Curve based Diffie Hellman of the OpenSSL implementation in C just for my own purpose of understanding. Is there any easy way to do that by coding in C ? Do you have any sample code for that ?


(I have used 

time openssl ecparam -name secp256k1 -genkey -noout -out secp256k1-key.pem

in command line. And it does give me some average timing. )


But using this I don't get timing for generation of shared secret key. Also I would like to know if affine coordinates are used or Jacobian projective coordinates are being used? Is there any way to know all these ?


I also tried this code -> 

The code over here gives me a bunch of error

https://wiki.openssl.org/index.php/Elliptic_Curve_Diffie_Hellman


Is there any other reliable source that has been used before ? If someone knows can you please forward me or let me know? That would be extremely helpful. Thank you for your patience.


Thanks and regards

Tanushree Banerjee

Graduate Student

University of Waterloo



--
openssl-users mailing list
To unsubscribe: https://mta.openssl.org/mailman/listinfo/openssl-users
Reply | Threaded
Open this post in threaded view
|

Re: Request for help in research

Dennis Clarke-2
On 24/03/18 06:10 PM, Tanushree Banerjee wrote:
> Hello,
>
> I was looking for benchmarking the execution time of Elliptic Curve
> based Diffie Hellman of the OpenSSL implementation in C just for my own
> purpose of understanding.

A few more specifics would be helpful here.

> |in command line. And it does give me some average timing. )|
>

Your results here would be portable across systems however one would
need to know the nature of the build and the architecture. I often
single step through functions and thus use non-optimized debug builds
which are vastly slower than a regular optimal build. Also the asm code
simply does not exist on some architectures like sparc ( non T4 ) or
some older powerpc types.


> ||I also tried this code ->
>
> The code over here gives me a bunch of error

That isn't too clear. How are you calling ecdh() and with what test data?

> Is there any other reliable source that has been used before ? If
> someone knows can you please forward me or let me know? That would be
> extremely helpful. Thank you for your patience.

Hey there UofW type, what rev of openssl are you using?  Should I assume
that you have "OpenSSL 1.1.0g  2 Nov 2017" on some reasonable
architecture?

Dennis Clarke
number cruncher math geek
--
openssl-users mailing list
To unsubscribe: https://mta.openssl.org/mailman/listinfo/openssl-users
Reply | Threaded
Open this post in threaded view
|

Re: Request for help in research

Tanushree Banerjee

Hey Hi,

I'm using 1.0.2n version of openSSL currently. 

I'm using it on ubuntu 16.04, processor is i7 - 6700 Skylake.

Actually I  understand that the code https://wiki.openssl.org/index.php/Elliptic_Curve_Diffie_Hellman  has omitted the details of how to obtain the other party's key (the peer key) . 

So is there any code available online which can be used to benchmark the execution time of the ecdh implementation of OpenSSL ? Is someone aware of that ?


Then please let me know.


Thanks and regards
Tanushree Banerjee

From: openssl-users <[hidden email]> on behalf of Dennis Clarke <[hidden email]>
Sent: Saturday, March 24, 2018 7:15:57 PM
To: [hidden email]
Subject: Re: [openssl-users] Request for help in research
 
On 24/03/18 06:10 PM, Tanushree Banerjee wrote:
> Hello,
>
> I was looking for benchmarking the execution time of Elliptic Curve
> based Diffie Hellman of the OpenSSL implementation in C just for my own
> purpose of understanding.

A few more specifics would be helpful here.

> |in command line. And it does give me some average timing. )|
>

Your results here would be portable across systems however one would
need to know the nature of the build and the architecture. I often
single step through functions and thus use non-optimized debug builds
which are vastly slower than a regular optimal build. Also the asm code
simply does not exist on some architectures like sparc ( non T4 ) or
some older powerpc types.


> ||I also tried this code ->
>
> The code over here gives me a bunch of error

That isn't too clear. How are you calling ecdh() and with what test data?

> Is there any other reliable source that has been used before ? If
> someone knows can you please forward me or let me know? That would be
> extremely helpful. Thank you for your patience.

Hey there UofW type, what rev of openssl are you using?  Should I assume
that you have "OpenSSL 1.1.0g  2 Nov 2017" on some reasonable
architecture?

Dennis Clarke
number cruncher math geek
--
openssl-users mailing list
To unsubscribe: https://mta.openssl.org/mailman/listinfo/openssl-users

--
openssl-users mailing list
To unsubscribe: https://mta.openssl.org/mailman/listinfo/openssl-users