Perenyi Attack

is this not the keres attack?

Not quite, it's a delayed keres attack because Be3 was played first. Games on chessgames.com call this the delayed keres attack perenyi gambit.

All I know is that Qf3 is the move I'd never play. It looks like a computer move that I don't see the logic of... Houdini says it likes it, but I can't figure it... why intentionally put the Q right there where it'll be lined up against the LSB in a few moves? The line the compter gives looks interesting and unique, but unnatural to my style... 

I also, personally, don't like the look of f4 after e5... although it's a popular move

Bg2, h4, and h3 are the ones I've tried in blitz... They lead to all kinds of stuff that seems pretty even... as they say "it's a game"

Rg1 is interesting, too...

But I don't know the extensive theory on all these lines, the branches here are just sooo diverse.

I'd have to agree 8.Qf3 is crap due to 8...e5 9.Nf5 g6 10.O-O-O gxf5 11.exf5 Bd7! (coming to c6)

And I think the problem with moves like h3, Rg1 is that Black gets to play ..b5 and ..Bb7 in peace.  e4 comes under pressure and any time White is forced to defend against ..b5-b4 with a3 it can be said Black is doing well.

8.Bg2 is probably the move you want if you want to sacrifice a N on f5.  For instance 8...e5 9.Nf5 g6 10.Qe2 intending exf5 and O-O-O.

Qf3 may not be "crap"... Houdini still liked it after a 20 minute think. But if I PLAYED IT WOULD BE "MEGA CRAP" because I don't understand the ideas of it. Here's an alternate line to the knight sac...

"The game continues" indeed but with White having nothing after 11...Nc6.

Also, the 9.Nde2 retreat drops the g4 pawn for no compensation.  It is safe to say 8.Qf3 is a bad move, at the very least it's inconsistent with the aggressive idea of the Perenyi which is sacrificing a knight on f5 to get a pawn roller.  If you are going to be forced to do an equalizing trick to stay in the game, it is better for White to choose a different line altogether.

8.Qf3 is fine as long as white does not mind a level endgame, which is the case in #6 after 11.g5. Actually this is somewhat easier to play as white, but with some care Black should not lose this, theoretically speaking. In practice it's different: Black lost this ending both times it was played, and in one case the man playing Black was a Grandmaster.

8.Qf3 is about equal, but certainly not "crap", or even harmless- far from that.

Good to know.  What do you think about 11...Nc6, though? I'm not sure if there is a later point where Black can improve in that above game.  His position collapses with remarkable speed.

Sure, 11...Nc6 looks like a better try, but since this move means a (temporary) pawn sacrifice, Black players were rather unwilling to play it. 8.Qf3 is not white's best, but not a bad move, either.

Thank you all for your input.