Janmajayamall

Implement Error trait for GetRecordError (Kad)

1

I am not sure why Error trait hasn't been implemented for GetRecordError ([ref](https://github.com/libp2p/rust-libp2p/blob/a1f3547b43bfac17ef50248e4a8ce1523d3aaa76/protocols/kad/src/behaviour.rs#L2558)).

LWE params n=2^12, q=2^790 outputs 1179 bit security (uSVP)

1

Running: `LWE.primal_usvp(LWE.Parameters(n=2^11, q=2^790, Xs=ND.DiscreteGaussian(2.0), Xe=ND.DiscreteGaussian(2.0)))` outputs: `rop: ≈2^44.2, red: ≈2^44.2, δ: 1.012950, β: 40, d: 2166, tag: usvp` This matches my expectations. However running: `LWE.primal_usvp(LWE.Parameters(n=2^12, q=2^790, Xs=ND.DiscreteGaussian(2.0), Xe=ND.DiscreteGaussian(2.0)))` outputs: `rop:...

(Rust binding) Repeated invocation of EltwiseFMAModAVX512 (with different data) in loop has unexpected performance regression

5

I am weiting rust bindings for hexl [here](https://github.com/Janmajayamall/hexl-rs). I have added support for NTT operations and some elwise operations. However, I am running into issues with elwise operations with `prime`...

I have `a, b, c` of type `Array1`. `a` is mutable reference, whereas `b & c` are just reference. I want to set each element in `a` as product of...

Add support for pk on G2 and sig G1 and make it configurable

1

I think swapping between g1 & g2 for pk & signature isn't supported. Is there any plan to support it & make it configurable?

range_fn improves with more threads, but the improvement isn't exactly linear. This is directly related to #1 since most of what range_fn does is calculate powers and iterate m loop...

fma_poly_scale_slice_hexl performance times does not increase linearly with mod_size

2

`fma_poly_scale_slice_hexl` performance times does not increase linearly with mod_size (ie moduli count in poly). Instead as mod_size increases time blows up. Poly stores its coefficients in row major form and...

Performance of single iteration for m loop regresses as no. of threads increase

1

Performance of single iteration for `m` loop regresses as no. of threads increase. This casues loop iterations in range_fn to not scale linearly with no. of threads. To be more...

Add Karatsuba's trick for tensoring ciphertexts in mul

5

Add Karatsuba's trick for ciphertext multiplication and benchmark it against normal tensoring for different polynomial degrees.