Closed haochenuw closed 6 years ago
Hi, it seems bootstrapping in this library is faster than the original HEAANBOOT:
TestScheme::testBootstrap(15, 23, 29, 620, 5, 2);
Returns
So, the EvalExp is about 2 times faster, and similar for the linear parts. Can you explain what makes this speedup?
Hi, we use CRT + NTT in this version. Previously we just use ZZX multiplication and reduce the result, and this was quite in efficieny. This change makes overall performance much better :)
Hi, it seems bootstrapping in this library is faster than the original HEAANBOOT:
TestScheme::testBootstrap(15, 23, 29, 620, 5, 2);
Returns
Start SubSum SubSum time = 1982.59 ms
Start CoeffToSlot CoeffToSlot time = 6505.46 ms
Start EvalExp EvalExp time = 5553.62 ms
Start SlotToCoeff SlotToCoeff time = 3986.2 ms
So, the EvalExp is about 2 times faster, and similar for the linear parts. Can you explain what makes this speedup?