Version 5

This version supports HomGates with lvl1 ciphertexts as the inputs.
The next release will come with supports for lvl2.

Version 4

I'll add the EvalKey support, so, a significant API change is expected after this release.

Version 3

l parallel NTT & fixed shared memory read before write bug by adding __syncthreads(). v3 API is the same as v2.

Version 2

Refactored & reduced shared memory usage (depends on TFHEpp for the parameter set select, runnable on old GPUs like GTX 1060Ti but slow)

Version 1

kvsp v29 compatible version. Run 2l NTT in parallel.