Help (kind testers wanted) :)

Hi everybody,
we have developped with a friend a CUDA based test program for a client-server protocol called Private Information Retrieval (see [1] for a large reference on the subject).

These protocols are a great tool for privacy protection as they allow a user to retrieve a file from a server while avoiding that the server may learn which file has been retrieved. They are in fact used as a tool for many other privacy enhancing protocols.

The server side is accelerated with CUDA, but we were just able to test it on my friends laptop with a 8600M GT. We wondered if someone would be kind enough to test it on more efficient hardware to see how much can we seed-up the process…

The sources (which are only a few dozen lines long) can be found attached to this message.

thx for any help!!!

Carlos

[1] http://hal.archives-ouvertes.fr/docs/00/13…/PDF/tdsc06.pdf
PIR.tar (130 KB)

I’ll give it a try

I get some build problems. While building the server:

orion@sulphur:/tmp/PIR/server$ make

mkdir: cannot create directory `/home/stan': Permission denied

When building the client:

g++ -o RIPClient -I../Common/include src/mainClient.cpp -lntl -lgmp

src/mainClient.cpp:41:26: error: NTL/mat_ZZ_p.h: No such file or directory

src/mainClient.cpp:45: error: ‘NTL_CLIENT’ does not name a type

src/mainClient.cpp:65: error: ‘ZZ_p’ was not declared in this scope

This is because you should not directly execute make yourself, the configure script does it for you creating the right makefile.

Keeping your notations you should do:

orion@sulphur:/tmp/PIR$ ./configure

The /home/stan directory comes from the original CUDA makefile that should be replaced during the configure script (modifying server/Makefile.conf)

:blink: Ooooops that one is our mistake. The script supposes that you have installed NTL, the number theory library. You can get it here:

http://www.shoup.net/ntl/ntl-5.4.1.tar.gz

In fact GMP is needed too but you seem to have it already!

If you prefer I can post a tarball with the PIR project and NTL inside…

thx for trying!!!

We are two small budget researchers trying to improve privacy preserving protocols and we appreciate much your help!! :thumbup: :thumbup: