FPS higher at max settings than min :confused:

That's right, somehow I'm getting more FPS at max settings than min settings. I'm not sure why. Here's the specs:

Game: Battlefield 1942
Video Card: nVIDIA GeForce MX 440 (way outdated, I know...getting a new rig this month)
Benchmark: FRAPS, the latest free version

Benchmarks:
Maximum Settings
1280 x 960 x 16 @ 60 Hz, High graphics settings, 100% texture detail, shadows, all optional settings enabled
Min FPS: 3
Avg FPS: 10.53
Max FPS: 26

Minimum Settings
640 x 480 x 16 @ 60 Hz, Low graphics, 20% texture detail, no shadows, all optional settings disabled, view distance left at 100% (only non-minimum)
Min FPS: 3
Avg FPS: 9.627
Max FPS: 31

So my question is, why would I be getting better benchmarks at maximum settings than minimum? I'd lowered the settings before I tried FRAPS and didn't notice it being quicker - that's why I benchmarked it. The game does look much better on maximum settings than minimum, with better FPS, so I'm pretty sure the minimum settings aren't entirely maxing out my card.

Note the benchmarks are just for one map (Omaha Beach), and is with all 63 bots playing. Could I be maxing out my CPU instead of my graphics card? Here's the other major specs of my machine:

CPU: 2.66 GHz Pentium 4
Memory: 1 GB DDR RAM @ 266 MHz FSB

If you have any idea why this might be happening, I'd like to hear them - doesn't seem to make much sense to me.

 

Hmm somthing must be giving way since the card is way under spec for the game. Do you visually see much better graphics with max settings?

It may be so high that the card flat out doesnt render it and thus a higher fps, that or it does a frame skip.

 

I think your maxing out the CPU with 64 players. I use to play 1942 on a celeron 1.8ghz, 512mb of ram, and a mx 420 and it played just fine.

 

ahh yeah 63 bots lol I dont know why i didnt see that part... thats a super heavy load a cpu depending on the games AI coding.

Try some lower bot numbers and rebench.

Usually video benchmarks are done with a benchmark untility or a time demo, not actully playing the game. To many unknowns.

I was under the impression that it was a time demo wich means the exact same events happened every time.

I cant play a fps game with atleast 30fps average anyways :/

 

Haha, yeah, I had a bit of a feeling when I was writing it maybe it was the CPU, that's why I added that near the end. Too bad; I like to play with 63 bots. But at least it makes me feel better about shelling out the extra $ for a T7500 on my new laptop.

The graphics are indeed much better with maximum settings; the difference of the top of the factory in Battle of Britain was night and day when I switched it. So I'll half the number of bots and try again!

 

Hmm...still getting equal results for min and max graphics with 31 bots. How much of a bottleneck is BF 1942? I've got the AI skill on the highest settings, yes, but I didn't think my P4 would be the bottleneck on 31 bots. Oh well, so long as the C2D T7500 is amazing - the results are sufficient with 31 bots for the next month or so.

Updated results (w/31 bots, everything else the same)

Minimum Settings
Min FPS: 4
Max FPS: 71
Avg FPS: 21.296

Maximum Settings
Min FPS: 6
Max FPS: 49
Avg FPS: 20.299

Granted, these aren't the most scientific of benchmarks - only 1 or 2 minute bits of gameplay, but I think they're adequate to establish the trend.

And nice signature, ViciousXUSMC.

 

OK...finally broke the CPU bottleneck! This time I had 18 bots playing - 110% of the default. I guess that's why they made 15 the default number - makes sense now. *note: I know 110% of 15 isn't 18 - that's just how BF1942 calculated it.

Here's the updated results (w/18 bots, everything else the same)

Minimum Settings
Min FPS: 21
Avg FPS: 56.779
Max FPS: 88

Maximum Settings
Min FPS: 12
Avg FPS: 32.931
Max FPS: 50

A few other observations: Task Manager still claims 100% of the CPU was being used, but obviously it wasn't the bottleneck anymore. But I've seen it say Autoplay is using 100% of CPU, so I'm not sure it's the most accurate measure all the time.

As to the GPU being the bottleneck for most games...this doesn't seem to be the case for me. The CPU is certainly the bottleneck in my favorite game, Civilization III. I'd expected otherwise in BF1942, since it is an FPS, but again the CPU was the bottleneck. It wasn't like the MX440 was top of the line when I got it in 2003 either, though the FX series wasn't doing so great. Interesting. Do most people prefer to play with fewer bots? I've always liked more bots, or bigger maps and more rivals in strategy games, so maybe that's why it always seems like the CPU is the bottleneck for me.

Or perhaps benchmarking is usually done with the default number of bots/map size? I wonder if GPUs would still be the bottleneck if the maximum bots/map size were used?

 

well the game was released in 2002. and in BF2 and 2142, which was released several years later, the bots can be only be maxed at 31 in a co-op game. its the fact that with a lot of bots the CPU is having to do all the AI and path finding work for each bot.

 

Well thats sorted out, I have had alot of post talking to people about the cpu bottlneck and this just backs up everything I have been saying so I'll use this as fuel for the fire and link too it if I need too.

 

Go ahead...it's the second game I've had run into a significant CPU bottleneck. The other is Civilization III, which although a 2001 game, it does much better than BF1942 if you configure it right.

Right now I'm playing with 26 bots, and it's just a bit below the bottleneck. Max settings gives 27 FPS, minimum settings gives 36 FPS. So with my 2.66 GHz P4, just above that the CPU bottleneck will start hampering gameplay.

It'll be nice when games start being configured for dual-core CPUs...I just wish some of the older ones had that capability.