I enjoyed UT2004 for LAN parties working coop on some of the maps against bots. I liked the maps where you had to capture the individual points before you could destroy the enemy's reactor. I can't remember the name for that type of map.
Don't all the games now need some sort of internet check? You have either the install check or the check on each load or save. GTA4 to even save games you had to be logged in to something, you could play but never save without the internet.
City of Heroes just did a major revamp to its graphics engine, to support more world detail. My ATI 5870 went from being able to deliver 60fps@1920x1200 100% of the time to only displaying 12 - 25fps.
And now I want a faster graphics card.
I'm still impatiently tapping my foot and waiting for Mechwarrior 5.
What is the latest news there?
I just found a torrent of the various LucasArts SCUMM-engine games. Apparently they've been modified to support more modern graphics settings so for example Loom can be played with a full VGA palette now.
I'm looking forward to playing Grim Fandango.
It ran smoother than the 32-bit version. A lot of people said that the 64-bit loaded levels faster, which doesn't make a lot of sense to me. Personally, I prefer to run 64-bit native when there's no apparent penalty.I wasn't even aware that there was a 64-bit version or even any need for it.
You may not think that a 64-bit application will perform any better than a 32-bit, but you are wrong.
x86_64 has twice as many general purpose registers as x86, this alone should give a nice performance boost to most applications. An application compiled for x86_64 will have to swap far fewer variables from the stack to registers than a comparable x86 compiled application.
There are numerous other architectural improvements as well.
The flip side is that your variables can double in size if you're not careful, and your code base will be significantly larger no matter what you do thereby effectively halving the amount of cache compared to 32-bit and slowing down memory accesses as your code base is much larger, so you have to read and write more data.
On a 3 or 4 Meg L2 Core 2 or later, it's not so bad. On a Celeron or Pentium E (early 1MB L2 version)--:errr::errr: