On the same cabinet, same drop cable, I have one line that is unstable on Lantiq due to the G.INP trial (also went interleaved when I tried to force 3dB SNR on Lantiq before G.INP) so went Broadcom, and another that I can set the SNRm down to 3dB thanks to Lantiq and is completely stable.
Its really quite eye opening to have two completely different quality lines taking identical routes back to the same cabinet.
When its the specific line conditions that determine what works best, its not just difficult to predict what works best, its downright impossible.