Reimplementing the cubic sieve faster

Bonjour,

as you know, the cubic sieve already exist in Pari‑ɢᴘ, but as far I 
understand, it lacks the improvement of 
https://www.sciencedirect.com/science/article/pii/S0747717113001703 for 
computing the initial cubic sieve congruence far more faster. A part of 
the problem, is I fail to understand how to implement the main algorithm 
in any language (which of course include pari/ɢᴘ).

Would it be possible to get help for understanding how exactly to 
perform the sieving and relation collection steps ? My use case would be 
on prime fields…

Sincerely,