I'm missing a gigabyte of memory.

[Jazzmail]

1. What? Quake 3 runs at like 500 fps on modern pcs.
2. Sure.
3. Min fps on old games are going to be way over 60 with decent video cards. Left 4 Dead has a min fps of over 60 with the GTX 260 at 1920x1200 with 4x AA and maximum settings otherwise.
4. Answered in #2.
5. Differences in fps over 60 are totally irrelavent, as LCDs only show 60 frames per second. If your min fps is over 60, you should be running vsync to lock the fps at 60 and eliminate tearing anyway.
6. Supreme Commander, GTAIV, The Witcher, Stalker all run better on Vista or 64-bit, off the top of my head.
7. Thanks, I'll get on that.
8. Perhaps you could provide a detailed explanation.

Quake 3? Couldn't you find anything more ancient? I meant older games, not hopelessly outdated games.

L4D is a fairer comparison I suppose. You should be able to stay above 60 with a good rig. Then again, L4D looked like the previous generation even back on the day of it's first release.

Now try these old games at the same rez, full detail and 4xAA: call of duty 4 (2007), crysis (2007), lost planet (2006) and world in conflict (2007).

As for why going above 60 fps makes a difference: it makes the game feel more responsive when you play a fast game.
Even when the monitors refresh rate is 60hz and you have V-sync on, enable triple buffering and the game will keep churning out more frames above 60 fps. The latency between the player's input and the result shown on screen will be lower the higher the framerate.

Hardcore FPS players know that framerates in the 200 range make a shooter even more responsive. It's something you would need to experience for yourself.
Take out Quake3 again, ignore the square character models and play for a bit and then play again with v-sync + double buffering (capping the frames at 60). The second run will feel more sluggish.

 

[Zelig]

As for why going above 60 fps makes a difference: it makes the game feel more responsive when you play a fast game.
Even when the monitors refresh rate is 60hz and you have V-sync on, enable triple buffering and the game will keep churning out more frames above 60 fps. The latency between the player's input and the result shown on screen will be lower the higher the framerate.

Hardcore FPS players know that framerates in the 200 range make a shooter even more responsive. It's something you would need to experience for yourself.
Take out Quake3 again, ignore the square character models and play for a bit and then play again with v-sync + double buffering (capping the frames at 60). The second run will feel more sluggish.

Not really, it's the method that vsync uses that increases input lag, not the fact that fps gets capped at 60.

In short, with triple buffering you'll get optimal performance/visuals and minimal input lag if the buffers are all getting filled as quickly as they can be at 60 fps. Measuring fps of a triple buffered game isn't even useful, since you're only measuring the front buffer.

Anandtech explains it better than me, in any case.

 

[Jazzmail]

Not really, it's the method that vsync uses that increases input lag, not the fact that fps gets capped at 60.

Only if the fps are too low. When the fps are more than twice the refresh rate (like with Quake3), enabling v-sync in double buffer will not result in any additional lag beyond the capping.

In short, with triple buffering you'll get optimal performance/visuals and minimal input lag if the buffers are all getting filled as quickly as they can be at 60 fps. Measuring fps of a triple buffered game isn't even useful, since you're only measuring the front buffer.

You can benchmark all the frames drawn, including the frames that get discarded from the back buffer.
Not that you actually need to benchmark, because you should already know the calculated frames will be in the hundreds with Q3.