• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

FPS R.I.P.

S

Spook

Platinum Member
Pulled this article from Overclockers.


"FPS R.I.P.?"
Ed Stroligo - 6/5/00

--------------------------------------------------------------------------------


For years, we have measured video cards by how many frames per second they could put out in various games.

It was a good measurement. You want to play a game, not Keystone Kops, especially when the cops are trying to blow you away. No one wants to get killed because of a slow or missed frame.

It's simple. It's easily quantifiable. You can look at two numbers and say one video card is better than another.

However, it seems to me we are approaching the point of uselessness when it comes to FPS.

I'm not saying we're there yet, or will be tomorrow. I realize that features like FSAA increase quality at the cost of frame rate, which could keep FPS as a measurement in business for a while.

At some point in time, though, it's going to get ridiculous.

Twenty years from now, will we be boasting, "My video card does 5000 fps?"

There comes a point where quantity becomes useless. Measuring quantity is good when you don't have enough. Once you do, though, then you should start working on other things, like quality.

There was a country that focussed solely on quantity rather than quality. It was called the Soviet Union. Towards the end, the Soviet Union made a lot more steel and lots of other things than America or Japan; after decades of effort, it finally "beat" them.

Did the Soviet Union win the war? No. America and Japan made enough steel, then moved on to making other things. Even things that didn't need steel. Steel is very good for certain things, but no one wants a 10-pound portable CD player.

There's no point making more of something when more becomes useless.

At some point, we'll reach that stage with video cards, where more FPS becomes visually undetectable.

Then what?

All the other measurements we have measure quantity. We don't even have another measurement that can measure quality.

People say the Matrox cards have better "image quality." How so? No measurement, people just say it looks better. I'm not saying this isn't so, but looking at something is pretty subjective.

People would much rather measure quantity rather than quality because it's a lot easier. It's very easy to say 12 cans of paint are more than 10 cans. However, if the 12 cans of paint are lousy paint, and the 10 are very good, which is "better?"

Fortunately, with video cards, there is a relatively simple way to measure quality: accuracy. How perfectly does the video card translate its instructions into output. Does it get the color right? Does it put the pixel in the right place? Can it handle every pixel correctly? Can it do all this fast enough for human perception?

That's a good deal harder than FPS, and involves a little more calculation, but it's doable. Some of the measuring tests for monitors point the way to this kind of measurement, but we will need more.

This isn't meant to provide an answer, just to ask a question that some software developer out there might want to answer.

--------------------------------------------------------------------------------

The reason I posted this is because, lately I have been seeing many articles on Image quality... It seems that even though Nvidia holds the FPS crown, they are lacking in image quality. I have a Prophet 64Meg GF2, and yes between different games I see many visual anomalies. I've actually considered getting an ATI Radeon, or a V5-xx, just because the GF line of cards, even though it records the highest frame rates, it also seems to have the greatest swing in low and high FPS... Its nice that I can get 120FPS, but those lo swings of 15-20FPS kill game play... Sure it averages higher, but what's the point, If I'm in a fire fight, I need as many FPS as I can get... I got my hands on a V5, and noticed in UT(I know Voodoo is UT friendly) in the flyby that there is a point in the very begining where the V5 dips to a high 30 FPS, and the GF2 dips to a low 20's... Thats where it count's.... maybe we need to start gearing our tests toward averaging on a heavy fire fight... Don't start walking down some hall, start in the middle of a ferocious fire fight in Q3 or UT, and end the bench at the end of the Fire fight... that way we don't get cards that can pump out 120FPS walking down the hall distorting their Average FPS.... Just an Idea...

 
Yeah, with todays software this may be true, but how about tommarows? Will it run at 5000fps? As hardware grows software does. I beleive that as long as hardware keeps getting better, there will always we software availble that will put a stress on it.

my oppinion
 
Good lord, I can't wait until the day we're all going to be running 16x12 res at 5,000fps with RAA (regional anti-aliasing). So Nvidia is like the Soviet Union, Matrox is the US, and ATI is Japan? ehehe..
 
Hey Sunner, there's a Crusher.dm3 for Q3. Click here, scroll down to Locki's config, and it says download crusher.dm3 in yellow 🙂 LOL I get a higher framerate with my TnT2U than that dude with the GF on Crusher 😀
 
5000 fps will never happen unless you run yesterday's software.

JC has been quoted as saying that doom3 will not be able to push more than 60fps on the fastest card avaialble (read NV20) when released.

Seb
 
This is a little rediculous, frames per second was merely a measurement of performance. There are many sides to a video card including image quality and features, but if you don't know how well the card runs when the rubber meets the road its worthless. You could say FPS is useless because you can play GLQuake or JetMoto 1 at 300FPS on a Geforce GTS. But what use is it if your comparing video cards using CURRENT games that run no where near 300FPS. Sure the i740 might have had better image quality then a VooDoo2, but if I was trying to play Q3 you can be sure I'd pick the V2.

I think this idea that FPS is no longer valid is a result of a few things. First that originally in the V1-V2 generation the hardware was very young still compared with non-realtime 3D or professional cards. Developers knew what they could do with faster cards and had to cut the performance down to run on that generation of cards. Also the framerate was generally low, 20-30fps average was somewhat common with cutting edge games.

Now we have cards that have surpassed the software in capability, mainly because too many of the older cards, line Rage128's and TNT1's are still in the market and must be supported to sell well. So your new Radeon or GTS can run many many games at well over your refresh rate, only some games even at high resolutions such at 1024x768 will go below 60FPS. I think this will change once the XBox hits the market, you'll see more and more PC games shooting for the Geforce/Radeon as the baseline performance level. With non-T&L cards running the games, but very ugly and slowly(like Q3 on a V2/Riva 128).

BTW, I think the Geforce series has gotten a bad rap for image quality. Generally for 3D quality the Geforce series tops the list. Do you honestly think 3D artists would be using them if it has such horrible quality? Did you know that ID software is using Geforce/Quadro cards for their developer/rendering boxes?

There are a few issues such as the realtime compressed textures in Q3. And for FSAA 3dfx has the quality crown, with ATI, nVidia and VideoLogic somewhat behind. Otherwise the Radeon has slightly better Anistropic filter then the Geforce series. OTOH the V5 doesn't even have Anistropic filtering.

The other problem is 2D image quality, which can vary from Excellent to Poor depending on the manufacturer and model. This is generally out of nVidia's control, they just sell the chips which have the ABILITY to have very good quality 2D.

PS, the Geforce has lower "lows" in UT simply because its poorly coded for non-Glide cards. A Radeon will have similiar drops, OTOH with the s3tc textures and OpenGL with the 4.28 patch the performance of the Geforce & Radeon is much improved even over D3D. Probably matching or exceeding the V5 in some cases. Someone should do a comparison review, Anand?
 
Is this guy assuming that today's games are as good as it's going to get?

Personally I think this guys logic is all wrong. FPS will always be a determining factor, because games will constantly push the envelope.
 
I don't see this happening. Software in the future is going to keep pushing the video cards hard and FPS will still be important. Although if I am getting over 100fps i am generally happy with that. Unless it really slow down when there is a lot of action which you always have to watch out for.
 
I think the main point of the article is flawed. Even today people test time demos called Crusher or what have you. The demos have all the details turned on in the graphics engine.
The stronger cards will be able to run the more detailed games at a better speed. The weaker ones won't. Yes people test FPS, but they dont just aim for FPS. The fact they are running all the details would make it more visually appealing by default!
If you had a choice between a FSAA 75 FPS full detail or
a 160 FPS basic detail which would you pick? .. I would hope the FSAA one!
The change is already happening, they just seem to have missed it.

 
I agree with the article, note that he didn't say we were to that point yet. If nothing else, we do need a quality test now, because sometimes, Geforce2 vs Radeon for eg. , there is little performance difference between the cards. OTOH, there does seem to be a significant quality difference, AFA some people claim. It would be beneficial to have some way to bench picture quality.
 
You must log in or register to reply here.
Back
Top