Slashdot Mirror
Benchmarking Your GPU with F.E.A.R.
ThinSkin writes "Monolith's new shooter F.E.A.R. is all fun and games, but it can also be used as a benchmark to test your GPU's performance. ExtremeTech's Jason Cross goes into detail on benchmarking your GPU with this graphically-intensive game. In addition, the article also tests the performance of high- and mid-range cards from ATI and Nvidia to see which scores top marks." It's a tough game; I had to buy a new rig.
I found that if you have a midrange system you can turn everything to max, but keep volumetric lighting and all shadows turned off and the framerate stays pretty constant no matter what other settings are enabled or at what quality. The only thing i saw other than that, that hit my performance significantly was enabling full quality textures because i don't currently have enough ram ("only" a gig) to keep all the textures in. The HD swapped a TON with textures all the way up.
There was a time when a game/program required an excessive amount of computing power that was a BAD thing.
I don't look at a game that requires 350W of computing to run as a "good thing". Sorry, I just don't. Any VB.NET hacker can make an inefficient bloaty game. It takes real talen to do the same with minimal requirements.
If this "F.E.A.R." game really requires a $500 graphic card to play then they can keep it. It's just a game, you'll play it and be bored within a week. Meanwhile you're still out the $500 and your computer is taking "yet more power" to run.
These peeps really ought to develop games for things like a gameboy or PSP first. Then they'd get an idea of what "optimization" means.
Tom
Someday, I'll have a real sig.
With your type of thinking we would still be playing pong just in multiple colors. There is nothing excessive about the game requirements what is excessive is the whiners about how other people spend their time and money.
There are always those who will try to guilt-trip anyone for whatever reason. Most always it boils down to money. Like people who harp about how much gasoline costs, to hummers, to millionaires buying rides to the ISS.
Enjoyment and relaxation come in many forms and how people spend THEIR money is of no real concern to me as long as it does not endanger me in the process.
Computer games are advancing the state of entertainment, attempting to bring realism to the screen. Doing so does require oodles of computer power and we have that luxury these days. People looking at the future would never imagine the power we dedicate to games but looking back 10 years the picture changes.
The amount of power expended by high end PCs is nothing to cry about. In fact it trivializes many other real wastes of power and money.
* Winners compare their achievements to their goals, losers compare theirs to that of others.
If you've already spent that much on a system, going SLI isn't going to kill you too much. It's just another $300-$400 card. Now if you don't have a compatible motherboard and power supply, that might ramp up another $250-$350. Either way, you KNOW you wanna do it. You can't resist. Alma awaits.
Just for kicks, I tried overclocking my rather plain 6800GT card. I saw an almost direct linear relationship between % overclocked and % increase in framerate in F.E.A.R.
At most I managed to push it from 350 to 410Mhz (no special cooler), which is a 17% increase. The average framerate went from 41 to 48, which is a 17% increase...
The in-game 'benchmark' is misleading - it's just a fly-by, with no A.I. load on your CPU at all. Given how much amazing A.I. there is in F.E.A.R, the numbers you get from the in-game fly-by are not at all representative of real gameplay performance. In fact, they are artificially inflated. If you want to see the difference between non-playable fly-by runs and *real* human gameplay experience, I suggest you read bit-tech's review of F.E.A.R. They proved this benchmark was bollocks three weeks ago, so used FRAPS to measure someone physically playing the game. The results are way different. Unfortunately, the Anandtech benchmark review failed to spot this, so those figures are all wrong too
If you have an ati card (R4XX models) you can get a performance boots by renaming FEAR.exe to anything.exe.
Running fear 1.02 on an msi x800 xl
first i ran with FEAR.exe:
1st run (fps):
* min 25
* avg 46
* max 93
* 0% below 25
* 44% between 25 and 40
* 56% above 40
2nd run (fps):
* min 26
* avg 46
* max 91
* 0% below 25
* 43% between 25 and 40
* 57% above 40
then i quit and ran anything.exe
1st run (fps):
* min 22
* avg 42
* max 103
* 1% below 25
* 44% between 25 and 40
* 55% above 40
2st run (fps):
* min 21
* avg 42
* max 111
* 3% below 25
* 37% between 25 and 40
* 60% above 40
not believing it i reverted back to fear.exe and it went back to the first lot of results.
I dont know whats going on, but the max framerate jumped up by 20 fps as well as the percent above 40fps. The min and avg values went down a little.
http://row1.info
... researchers have discovered that "Half-Life 2: Lost Coast" can be used to benchmark computer hardware. In fact, they also discovered that the original "Half-Life 2", as well as "Doom 3", "Farcry", and "Billy Bob's New Shooter" can all be used as benchmarks, because, incredibly, they all use the GPU, for something, sometimes.
Nothing to see here, PLEASE, IN THE NAME OF ALL THAT IS HOLY, JUST move along.
Don't thank God, thank a doctor!