f4 is best? Really?

No your depth is somewhere between 20-30. i can't read it

The point is web assembly stockfish (java script script ) is significantly slow, 60% of original speed at max.

But in chess.com, i guess they get only 15% of original speed of stockfish. ( i guess the speed is approx 800knps, where I get 4.5 Mnps in my  4- cores. Or 6 times slower, which is massive. 

This is a screenshot where I get the best depth 28 in  10 seconds. ( I use single PV)

And also depth 40 is too vague. Depth 40 in middle game, opening, and endgame phase, and type of branching positions are factors influencing the depth of stockfish.  In this position, my stockfish hit depth 50 in 2 seconds,

24/47 at 5 seconds, is the last position SF searched before I stop it. (look at the top evaluation ,there are three evluation tree at 5 seconds) 

The first letter 24 means, stockfish is currently searching the brach/tree at depth 24. 

47 means the depth stockfish already reached in his principal variation. 

So the depth is 47.  

When we are talking about depth, we don't mean the depth of a single variation, but the depth of the tree which is 24 here. chess.com also shows this number as depth.

I am pretty sure that the depth means=47, (current variation of the node is additional info that Arena GUI display).

You should ask a programmer first, before telling your assumption like "everything you think is right".

It is very unlikely that chess.com use the wrong info.

You only have to look at the numbers to know you are wrong. You simply can't reach depth 47 with many pieces on the board that fast on a normal computer.

Then it is your another assumption. I wont argue again second assumption (ask stockfish programmers in fishcooking)

This is the group of stockfish programmers.

https://groups.google.com/forum/?nomobile=true#!forum/fishcooking

And again, if chess.com display the current tree as depth, then the depth would be back and forth like, depth 2, depth 22, depth 31, depth 3,  in chaos numbers in every second, cos it is the way how stockfish search, where she has the priority/heuristic which tree she need to search first.

In chess.com depth, it is constant and display the highest depth.(so it should be the highest depth SF reached)

I know what a tree is. If you add more branches, that does not change the depths of the existing nodes. So there is no reason why the depth would go down.

Also the depth which was that number 24 in arena doesn't go down in arena either. 

I'm not convinced that this is due to the search depth.

If you look at the screenshot the OP first produced, below where it says that best move is 17 f4, it gives the best replies for Black, which are:

17...Qxg2 (mate)

17...Qxf4 (-1.87)

17...Qg4 (-1.75)

So the engine is seeing the mate in 1, but the analysis assumes that the 3rd best response 17...Qg4 will be played - or something. So it looks like it's a bug in the code that interprets what the engine is saying, not in the engine itself.

That said, it works fine for me in my browser

Yes, that is odd, that the mate is seen as a possibility, but it marked something else as "best." Thanks drmrboss for posting your stockfish analysis. I've downloaded stockfish but never actually used it yet. Your screenshot gives me a better idea of what's going on in terms of "depth" and what the computer is calculating. (that f4 move is shown as one of the early considerations, 3 times, with the queen just moving out of the attack as the response, instead of going for the checkmate!) I wonder if there's a rhyme or reason as to which piece moves are considered first? In any case, this discussion on the computer analysis and depth has been enlightening as it's turned out to involve more than I initially assumed. Thanks all!

1000 IQ play for Chess.com