I was a bit dissapointed to see the benchmarks of the new geforce 6800 at ET. Seems no matter how good the graphics card and system (Athlon 64 3400+), ET has problems getting over 90 fps. How come? Compared to Call of Duty, which uses roughly the same engine and is about the same in graphics quality; COD easily doubles that.
Comments?
IW nearly completely rewrote the renderer for COD, afaik.
ET is pretty CPU bound for me, and these benchmarks seem to support that.
ET has a much more complicated scripting system, wich also takes a lot from the performance. Also, CoD has a upgraded q3 engine, wich is useable for better graphics with a better performance (do i sound smart now :p)
COD pushes so many polygons because IW did major surgery on the renderer code. Also, from what I understand, ET is too dependent on the CPU for more powerful graphics cards to have much of an effect.
Now I step aside for someone who knows what they’re talking about to reply. 
EDIT: wow, so many replies, so quickly!
OK maybe ET is CPU intensive, but that still does not explain why a Athlon 64 3400+ with say a GF5950 @ 1024x768 performs only about 25fps better than my athlon XP 2000+ with GF3. Sounds to me like a bug in the engine, fixable with the upcoming 1.04 patch :moo:
shame they didnt try it on lna trickjump map as I have reached 400+ in some spots with my 9800XT they would be able to get crazyyyy fps
I have an athalon64 3400+ with a radeon 9800 pro. i cap my fps at 125 and it never goes below. i can get 250+ on most maps with it uncapped.
My guess would be that it is even more dependant on memory and bus bandwidth than CPU power. While an Athlon 64 certainly has better bandwidth than an athlon XP, raw memory speed has not kept up with either CPU or GPU power.
That’s more like it 
I suppose that is not at maxed out settings?
Or maybe the demo in the benchmark is with 64 players all throwing airstrikes at each other or something 
my flatmate (before he got his 9800XT) had his athlon 64 with a geforce2 GTS… 200 fps all the time.
I think it is very likely that the reviewers forgot to enter “\com_maxfps 0” before running their benchmarks, thereby uncapping the maximum frame rate. This would yield a maximum FPS close to the 90 mark.
lol, so true, the default is 85, and it will go slightly higher so on average 90 is prolly right :D:D:D
Nah, look at the benchmarks, you can’t get an average of 93 if you cap at 90 
The capping is really spot on, I never have seen it go above it.
Just try it. Setting com_maxfps to the default value of “85”, like so:
…will give you a maximum frame rate of 90 FPS:
From http://ucguides.savagehelp.com/Quake3/FAQFPSJumps.html
Q3 measures frame time in an integer number of milliseconds. This means that there are only certain frame rates that it can hit… specifically, frame rates of 1000 / N, where N is an integer number of milliseconds. Given this, you can find viable frame rates by varying N. So, the possible frame rates are:
…
90-91
83-84
76-77
71-72
66-67
…
It seems that whatever com_maxfps you set the actual FPS can also hit the closest higher number on that list.
However…
Who would read the on-screen FPS counter when benchmarking when you have /timedemo, which isn’t affected by com_maxfps???
Yeah OK it’s true you can go above your cap if you use an unusual number. If you use a cap like 43,76,125, it’s steady.
Yup, it does indeed. My rambling above assumes that they used some kind of external tool like FRAPS to measure in-game frame rates. If they used timedemo, however, the whole hypothesis falls flat on its face and you’re invited to eat it with some gravy and pai. 