If we suppose to have 2 powerful computer, that are able to calculate all moves in some seconds, who will win, the black or the white? Or the game will always be draw? (let say we have a unlimited number of moves in this case) Does the winner in this case depend from the strategy? We now chess game is very complex having 8x8 board with 32 peices.
The answer to "who wins between two perfect computers" depends on whether White or Black wins (or draws) with perfect play from both sides.
No one knows the answer to this question. If we did, Chess would be considered to be solved. That is, we would be able to define "perfect play". However, Chess is not generally considered likely to be solved anytime soon: the number of possible positions is simply far too great to even represent on any current machinery. If you could somehow store the correct move for any given position in a single bit (which you can't), you'd still need (approximately)...
...PETABYTES of storage space just to hold the tablebase. Obviously, this is so far beyond our current technical capabilities that it is, for the moment, impossible.
See here for more (with further links to a small fraction of the huge amount of research that exists on the topic).
The two current strongest chess engines are Rybka 3 and Stockfish 5. An 8-game match was played between those two engines a month ago. Rybka 3 lost +0 =5 -3. Ofcourse, it all also heavily depends on how strong the chess engine is by itself, but I would hypothesise that if we were to take two absolutely equal engines in strengths, then the majority of games played would be draws, although wins could also be possible (specifically for the side playing white), since white has an extra tempo, which can be masterfully utilised by the engine.
You can find the match on YouTube, here's a link to the first game.
"Powerful computer" is a relative term. Different chess engines use different algorithms and strategies to weigh what would be considered the best moves in a given position, but no computer or chess program is able to calculate all the possible moves in a chess game in advance, except for when there's only a few pieces left on the board.
If you mean who would win when two perfect chess programs with unlimited computing power go head to head: noone knows the answer, since chess still hasn't been "solved". I'd assume the most likely outcome would be a draw though.