How did Fritz analyze this game?

Can someone give some professional insight on this game?

Thanks

Tryvex; 

I think you probably understand this now, but here's what happened: 

*Computers have a depth limit; they can find tactics in the current position, but not extremely far ahead. 

This means that, if you get a powerful attack with a sacrifice, but you don't have an immediate tactical killer (but you will have a forced tactical kill 10 moves down the road) the computer will likely evaluate you, in the current sacrifice position, as worse, as computers love material. 

However, when you get closer to the forced tactic, given best play, the computer will "suddenly" change its evaluation of the position, as it now sees the impending tactic. 

You generally don't see this much with the most powerful computers, however, including Komodo. Fritz is strong, but I find it still suffers somewhat in this area (positional sacrifices that lead to a forced win). These positional sacs are a little too long to be classified as "tactics", however.

The only huge mistake is the 20th move of white, it goes from 0.00 to -10 eval. It considers many other moves like Rd1, Re1 etc as keeping the draw. At move 22 is recommends Rd1 or taking the rook with the queen. It considers your move a blunder which results in a forced mate in 8.

Anyway, until the move 20 mistake it considers the game a draw, even after the sacrifice. But maybe it can't see far enough ahead with the time I gave it for analysis and my laptop's weak CPU. I think the position needs hours of analysis.

Komodo 9.42 after 5 minutes prefers g4 though, instead of the bishop sacrifice.