Why do sites not review minimum frame rates? Isn't minimum frame rates the most important? I know they review average.

[WyteWatt]

?

 

[PliotronX]

I don't know why, man. But that's a good point, I recall a couple of reviews that showed the Kyro2 cards lagging behind GF2 cards in maximum framerates, but having a higher minimum framerate which would be much more important than having spikes of very high framerates.

One reason could be that very few programs or games include the ability to report minimum framerates.

 

[Cosmic_Horror]

How do you propose they measure the minimum frame rate? Admittedly i would want to see a graph showing actual frame rate against the total number of frames in the demo, so you can see where in a demo the lowest frame rate was achieved and also how the frame rate deviated over the demo. Some statisical analysis would be useful too.


The average is readily avalable, thus is used.

 

[andreasl]

I think you need to pay a visit to Aces Hardware

 

[WyteWatt]

andreasl ok i went there but don't see any reviews testing out the minimum frame rates on the GF3, GF4 ti 4600, and ATI 9700 pro?

 

[ElFenix]

bah, they should report average and then two standard deviations up and down.

 

[apoppin]

I don't think measuring "spikes" or "dips" in frame rates are that very helpful as long as the game doesn't stutter.

It would take more work to present graphs and "averages" are quite useful. Also something about "convention" - what we are accoustomed to - means it won't likely change.

When someone talks about their hardware, it usually isn't presented in the "negative".

 

[mchammer187]

I would like to see minimum frame rates as well

 

[andreasl]

Aces usually report minimum framerates in their reviews. But they have not done a pure video card review with the latest cards yet. I think it's in the pipeline however.

 

[Trevelyan]

Unreal Tournament had this feature... when you benchmarked it would show the highest, average, and lowest framerate of the entire demo...

Who cares if the average framerate is 70fps in SOF2 when all I see is 30fps during intense action...

You bring up such a good point... I would LOVE to see minimum framerates.

 

[Modus]

Minimum framerates definitely have a high visceral impact on the gameplay experience. That few seconds of low fluidity can really kill your suspension of disbelief. Some sites like Ace's Hardware make a point of including minimum FPS numbers in certain reviews, and it'd be nice to see them more often.

The problem with minimum FPS is that more often than not, it correlates fairly closely to average FPS, so that competitive rankings change very little. Also, it tends to "dilute" comparisons under many games, because those few highly-complex scenes often yield similarly poor FPS on many cards.

Modus

 

[BFG10K]

Minimum framerates are of course the most important figure but often they're not available in benchmark runs so the next best thing is used - average framerates. The higher the average, the more likely the minimum is going to be higher.

That's one of the reasons why the "you only need X FPS" arguments are so flawed in their reasoning.

 

[Markfw]

BTW intim83, this is one of your best posts.

Yes, I agree, that at least minimum, max, average, and maybe even a graph would be very informative. If you can't play with a video card for a while, these are the things that would iratate you after purchase, if the card had a good average, but was all over that chart constantly .vs. one that was more constant.

Anand ??? Could you try this one once ??

 

[kherman]

LISTEN UP
LISTEN UP
LISTEN UP

I used to work at a site doing bench marking. I actually set up the tests. I created a demo for Quake 3 by using bots. I put about 20-25 bots in a small arena with myself and started recording, always pointing myself at the action and firing weapons. Visitors would DL the demo and send me the results.

The site I worked at relied on user submitted benchmarks, but everyone was mostly honest. I could spot the liars after a while.

Anyways, that was the fairest way to do benchmarking in my honest opinion. Worst case scenerio! The problem is trying to get for-profit sites to do this. There are issues with advertising and complementary copy.

To get a true minimum framerate, a demo like this needs to be created that lasts about 1 minute on a high-end PC.

Karl