What are the ELO ratings for the main chess engines?

I am sorry if this has been asked many times before, but I don't feel like digging for it. What are the best engines available (price of no concern) and what is their strength relative to each other?

Also, what's so great about DEEP that it's over twice as much money as normal but the description, stats and hardware requirements are the same as normal? Fritz 13, Shredder 12, Rybka 4, Hiarcs 13, Junior 12, suggested at $65.95, sells for $49.95, DEEP versions suggested at $129.95, sell for $109.95.                                                

You may wonder why I know so much about engines but I am asking questions here; I researched House of Staunton and USCF Sales, I know prices and a few names but nothing more because of their horrible descriptions.

I apologize for asking so much. Please just throw in something, I'm not asking one person to answer it all. Thank you.

To get a rough overview of the relative strengths of the engines, use one of the many many chess engine tournament lists, like: http://www.computerchess.org.uk/ccrl/4040/

The ELO strength of engines is debated -- whoever sells the engine wants to brag with as high rating as possible. Also the engine ratings have not been calibrated against top human players (instead some weak engines have, and the rest are rated by how well the engines fare against each other). The rating lists put the best engines in 3100-3400 range, but some study (which i forgot the name of) suggests that the "true" rating (cf humans) is rather in the 2900-3000 range.

The "deep" in engines usually refers to utilization of at least one of two algorithmic techniques aimed at avoid mistakes that can bite you only in the long to very long term:

- PVS, principal variation search, which allows an engine to verify that a short-term good move is at least a decent move at a much higher search depth. The drawback is that they cannot guarantee the move is the best long term, but that the move is not "too bad".

- Deep can also refer to MCS, Monte Carlo Simulation techniques, where a tree of 'random' (but realistic) game continuations of the game is considered. If a short term move turns out to lead to many losing endgames, it might get discarded for the second best move short-term.

If anyone is interested about the deep algorithms google the technique names, it's hard to briefly describe them.  

Your question of engine ratings has no simple answer.  Calculating a rating based on the results of several engine tournaments may be somewhat accurate but I'm not sure how useful this estimation would be if you are concerned with its skill versus a human. 

Here's a neat list of engines:
http://www.computerchess.org.uk/ccrl/4040/

Not only is Stockfish free and open source but it's a hair from being the most powerful engine in the world as of five days ago.  The only difference between Stockfish and the engines you mention is marketing and branding.

Edit: I should mention that trivial differences in tournament conditions (ponder, opening book, time control) produces wildly different results.  So it's not uncommon for the top 5-10 engines in at any point in time to constantly bounce around in standings.

Well, this is a completely worthless forum, so I'm going to use it for advertising my new tournaments:

http://www.chess.com/tournament/dragon-showdown-o1700

http://www.chess.com/tournament/berlin-defense-shall-the-wall-be-destroyed-o1700

These are the kickoffs for my new tournament series.

The top computers-- stockfish, etc, are around 3200-3300 rating strength. Deep Blue was around 2900 in the 90's.

 As far as I know, deep was usually referring to the multi-processor version in contrast to the single processor (or core). In that respect deep means that the depth in ply that can be achieved is higher due to more ressources. Interestingly enough Fritz 15 and the Komodo 10 are available only as multi-processor version.

Concerning rating, on a relatively new PC almost every modern engine is strong enough to beat 99.9999% of the human players.

That is because modern computers rather have multiple cores than individual microprocessors.  So, Komodo 10 is multi-threaded. From its site:

For best performance it is very important to correctly set the Threads value. The default is 1. You should set Threads to the number of "real" cores on your machine. Consult your computer manufacturer to determine how many cpu cores your machine has (not to be confused with the number of cpu threads your machine has). We recommend running Komodo with Hyperthreading turned off on your computer.  You can often find this in one of the boot up BIOS settings.

This means that with even 1 processor, we have multiple cores and threads.

For example, on an i7 with hyperthreading, you typically have 4 "real" cores and 4 hyper-threaded or virtual cores. While the operating system may report 8 cores on this machine, there are only 4 cores and so you set Threads to 4.

I don't know what "true study" #2 is talking about, but in a recent time handicap match Komodo with 1 thread and 5 minutes had a 2900+ performance against a GM who had 90 minutes.

15 years ago the top engines of the day were at least as good as the top GMs in the world.  Modern engines crush those old engines the same way I would crush a 1000.

3200-3300 is a fair estimate for the best modern engines running on average hardware.