OT: Website suggestion

[Quakindude]

Something I've noticed lately is way too few websites, when reviewing video cards, focus on the given review samples ability to maintain a high minimum FPS. This, to me, is much more crucial than maximum FPS. I would much rather have a card that can maintain 28FPS at a given resolution and detail setting.

For example.

If you take card A and card B and compare them, they may both get 60FPS at 1920x1200 and all settings maxed. Let's say that card B actually gets 72FPS in this setting.

But card B may only be able to maintain 15FPS when the action gets real heavy while card A might maintain 25. To me, card A would be the better performer.

I know FEAR has a built in benchmark that shows this performance level on video cards and so does 3DMark. Would it be worth focusing on that aspect of video card benchmarks seeing as how no one else really mentions it or places much focus on it? Everyone out there is all worried about maximum FPS. It would be cool to see HL start an Internet revolution and focus on where the Frames really matter.

[polobunny]

You're likely not looking at the good benches then, because there's more than a handful of reviews that list/chart the minimum FPS (though it's sparse when compared to those who list maximum fps). It's not something bad to consider having in video card reviews, though. :)

[Stormcrow]

Hrm, did you intend this to go in the Joke Room?

[drew and not u]

I agree. The minimum FPS is the most important part. No one can tell a difference between ~30FPS and 60 or 120, mostly because of limited refresh rates. The minimum FPS is the most important factor. Max FPS is a good way to judge raw rendering muscle, but min FPS is where gamers are going to feel it the most.

[polobunny]

While we're talking about "gripes" on most reviews... Too many benchmarks are ran with CRAZY ASS COMPUTERS. No I'm dead serious.
I don't have the money to buy myself a QX9650. I don't have 4GB of DDR3. I don't have a SSD RAID0. I don't have a $300 motherboard.
I really wish I could see how some scores with more mainstream parts sometimes.

[One4yu2c]

Originally Posted by Quakindude

If you take card A and card B and compare them, they may both get 60FPS at 1920x1200 and all settings maxed. Let's say that card B actually gets 72FPS in this setting.

But card B may only be able to maintain 15FPS when the action gets real heavy while card A might maintain 25. To me, card A would be the better performer.

The problem with this is that a minimum frame rate count doesn't necessarily mean the videocard maintained that lower benchmark during a heavily rendered scene, it just means it temporarily dipped, perhaps for a fraction of a second as the scene was first rendered. Over-focusing on this number could/would give a very wrong impression of a videocard's capabilities.

And the better benchmarks, like SupCom, keep the action consistently heavily during the run, so when you look at the average, you're essentially seeing a worst case scenario, or close to it.

[Capper]

Originally Posted by polobunny

While we're talking about "gripes" on most reviews... Too many benchmarks are ran with CRAZY ASS COMPUTERS. No I'm dead serious.
I don't have the money to buy myself a QX9650. I don't have 4GB of DDR3. I don't have a SSD RAID0. I don't have a $300 motherboard.
I really wish I could see how some scores with more mainstream parts sometimes.

The problem with that is that to review many components, you can't/don't want to have a single component bottlenecking the system, or part being tested.

[polobunny]

Originally Posted by Capper

The problem with that is that to review many components, you can't/don't want to have a single component bottlenecking the system, or part being tested.

Which is why 2 test systems could be used.

[Quakindude]

I put the topic here so it wouldn't be out in the general area for all to see just in case someone were to take it as website criticism, which it certainly isn't.

Paul, I hear you on the minimum frame rate. I just think if you could somehow track the minimum frame rate, it would be a better indicator of the cards overall performance. I know some websites do touch on minimum sustained frame rates, but no one really explains the benefits and why you should be concerned about them.

[Capper]

Originally Posted by polobunny

Which is why 2 test systems could be used.

you completely miss my point....I have three test systems, and am working usually on 2-3 things at a time.......

I've also never seen anyone question/criticize our testbenches or testing methodology.

As to the frame rate issue, I think framerates are way overblown, 50-60FPS is all you need, anything over that is just penis envy.