CEGT, CCRL and most other popular testing organizations are uniform hardware (at least as calibrated as a function of CPU time).

There are many questions we can ask about a chess program and associated data:
1.  How strong is it on platform X at time control Y using pondering setting Z?
2.  How strong is it using SMP with N cores?
3.  How strong is it using MPICH2 with M cores?
4.  How strong is it if we allow learning?
5.  How strong is it with opening book X?
6.  How strong is it with endgame bitbase Y?
7.  How strong is it with EGTB system Z?
8.  How strong is it with EGTB system Z on RAMDISK Q?
etc.
We can combine these in any way we choose.  All of the questions are interesting.

I also don't like tournaments with unequal hardware but rybka would still be the strongest

the stronger engine silly..

There are several ratings sites.  At first only standard time controls were used: SSDF, then as hardware improved other time controls became interesting and of course more threads. Then all the playing sites became popular and fast time controls became easily preferred...probably because of disconnects.  But those naturally had ponder on, as a machine was only running one engine so it might as well be on.  So the current trend is to include ponder but that means the engines run on fewer threads and arguably play worse.  2-threads in a known position beats 1-thread in known + 1-thread a ply away.  Though depending on programming, they may be less prone to outright blunders when they get in time trouble and only have a tenth of a second to move.

CCRL and CEGT are by far the best for number of engines tested, variety in time controls, threads, and being regularly updated, but occasionally one has to look elsewhere when an engine is very new or ponder is a concern.

http://ssdf.bosjo.net/list.htm
http://computerchess.org.uk/ccrl/
http://www.husvankempen.de/nunn/
http://wbec-ridderkerk.nl/html/lastupdates.htm
http://sedatchess.110mb.com/index.php?p=1_59
http://www.playchess.com/stats/engineranking.htm (you can copy and paste to Excel and reorder by rating)
http://www.elhchess.demon.co.uk/ehss.htm
http://www.open-aurec.com/chesswar/
http://rwbc-chess.de/rapid.htm
http://amateurschach.de/

odds are, the CCRL/etc highest rated engine.  of course there would also be more randomness, a much lower level of chess, etc, & the only real point of interest would probably be the opening preparation of the teams. don't get the complaining about unequal h/w in tournaments about chess machines when we have thousands upon thousands of games played under fair conditions on equal hardware between all (or virtually all) publically available engines.

How about: which engine wins on a 2000$ harware?