The time spent in Paillier encryption/decryption is proportional to the modulus size.
The Paillier test takes a while due to the scalability test in 'testPaillierWithKeyGenerationBitSetOption' in which large moduli sizes are tested. Unless we take out this method (which we shouldn't do), we just have to wait. When working with the code, we are using surefire so excluding the test is an option. We have done some benchmarking on the encryption/decryption operations -- I put a JIRA issue in to quantify/qualify directly and plan to complete it soon. The two most expensive operations that we currently perform are the modular exponentiation and modular multiplication. We are using JNAGMP to help with modular exponentiation and I just put in a JIRA to take a closer look at modular multiplication (this was already on the radar, just formalizing it). https://issues.apache.org/jira/browse/PIRK-40?jql=project%20%3D%20PIRK On Sun, Jul 31, 2016 at 8:55 PM, Suneel Marthi <[email protected]> wrote: > Any ideas on how to speed up PaillierTest? > > Is the bottleneck in BigInter.multiply(BigInteger) or is the KeyGeneration > ? >
