Freeware Files: Five Punishing Game Benchmarks for your GPU!

46

Comments

+ Add a Comment
avatar

Shckr57

overall average = 99.935

highest = 364 

lowest = 52 

 

i love my pc, might try the other benchmarks as well 

avatar

K0BALT

here's my results with my Phenom 955 @ 3.57ghz and my (2)GTX 260 Core 216's.....

Run #1- DX10 1900x1200 AA=16x, 64 bit test, Quality: VeryHigh ~~ Overall Average FPS: 30.35

... not too shabby. I'm pretty happy. Let me knowif I missed anything. Thanks!

avatar

StonedMagician99

Great game. Please buy Resident Evil 5 (not just the benchmark). Well worth $50.

avatar

Xylogeist

Eh Very High setting, Phenom II 3.2GHtz Dual 4890s....

 Crysis Very High: 16 FPS... backto thedrawing board :cry:

avatar

K0BALT

I always wondered how to get an "actual" FPS score for Crysis. I just ran the benchmark and got an average 32.14FPS with (2)GTX 260's on the Very High setting @ 1900x1200. I'm pretty stoked. Thanks!

avatar

TheMurph

Did you crank all the extra settings too?  I'd be curious to see your score after jacking up all the performance bits and pieces to their higest settings.  :D

avatar

bingojubes

i wish all games came out with on, especially when all i get are crappy screenshots full of game site logos n stuff. it's nice to be able to run it, and look at what the games really look like.

they really don't do justice SOMETIMES, but for the Resident evil 5 benchmark, it was really sweet to watch. i do still prefer game demos, but some games just have a benchmark, i guess, even if releasing these kinds of programs ends up being kinda like an unplayable demo, but i like them because they show me an idea of what it may or may not look like when i finally get to play.

i remember when the SF4 benchmark came out, and on wel-built PCs, at lowest res and lowest settings with every little detail OFF, players were getting funny FPS @ 200-300+ . i find that pretty funny, since i got up to 600 FPS on the SF4 benchmark at those lowest settings.

i found the RE5 bencher over @ Guru3D.com, but i didn't check to see if they had an extensive listing or not. I did find the SF4 benchmark utility, there.

avatar

TheMurph

That's actually why I didn't include the SF4 benchmark -- although useful for comparison's sake, I suppose, I'd rather punish a system with a demanding game than try to make a comparison between card A's score of 280 FPS and card B's score of 290...

avatar

lunchbox73

Oh man, The Murph is fired up today.

avatar

TheMurph

Nah, I just think it's important to keep the discussion going about these things.  : )

avatar

YoshiHNS

Thanks for the list. I was trying to find different games that would give a good benchmark for my graphics card.

A big +1 to TheMurph for sticking around.

avatar

TheMurph

Thanks!  Let me know if you see any other good ones online, and I'll be sure to include them in a future roundup!

avatar

nekollx

 What about Homeworld?

Not sure it's viable in the demmo buuuuuuuuuuuuuuuuuuut

 

[IMG]http://i182.photobucket.com/albums/x202/NekoLLX/homeworldmission16.jpg[/IMG]

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

Black Lable 69

If I figure out a way to run one of those on a "First Gen" Eee 901 4G Surf do I win a prize?

avatar

TheMurph

Yes!  Yes you do!  I will make you a lolcat.

avatar

nsk chaos

rwar! good article lol

avatar

Modred189

This has been proven time and time again. Benchmarks do not accurately show how your game will actually play in real life. Sure, you ca run the crysis demo, but it's not running AI or physics. It may test your GPU, but your CPU is not nearly as loaded as in the real game. 

Not to mention that green and red BOTH code for the most-used benchmarks in order to squeak out as many fps as possible during those particular scenes. 

 

Benchmarks can be fun to watch, and are often great ways to show off the core capability of your computer, but, personally, I never hang my computer, and my $50, on a benchmark. I'll play the demo first, and if I really want to get scientific, I'll have FRAPS record my fps.

 

for reference, they put it best:

http://www.hardocp.com/article/2008/02/11/benchmarking_benchmarks 

avatar

TheMurph

Sorry, but I would rather take a repeatable series of benchmarks spread across numerous games than one guy's estimation about how he came kind of close to running through a real level the same way.

"After a week of trials and a lot of practice, we finally got to a place where we could pull repeatable real world gameplay run throughs. You only get one shot at recording a “good” real world run through to later use as a demo. If you screw it up, you have to start over."

Not to make this a pissing match, but Hardocp also compares videocards running on completely different settings against each other.  How is that, in any way, a true apples-to-apples comparison of what each card can deliver?  Also, that tells me nothing.

avatar

Modred189

Just to make sure, did you look at the article? 

I ask, not to be an intarwez jerk, but because you do not address the main point of the article- Benchmarks do not represent the performance you get from the game in real-world play.  That alone means that using benchmarks to compare cards, ESPECIALLY between brands and drivers, is useless. If ATi codes a driver to pump out higher clocks for the 1.5 minutes of a given demo (short enough to not risk overheating) but not during the actual game, how does that represent the performance of the card in real gameplay?

 Also, what about CPU load? It's not as big of an issue as it used to be, especially on Corei5/i7 systems, but on the systems that most of us use, it still has an impact, especially on dual-core systems (i.e. the majority of systems).  Without physics, AI, pathing etc running, Benchmarks load the CPU and ram differently/less. 

Furthermore, go peruse [H]ardOCP's graphs, you can see the different runs and how similar the loads are, marked by savepoints and in-game movies' effects on the fps (recorded by FRAPS, btw). And while I have never seen [H]ardOCP compare systems of different specs, it's a moot point, as the article I linked used one system for the whole test, and has no relevance to my point in general:

 

Benchmarks do not represent the experience you will have in actually playing the game. They remove too many variables and loads, and present too much of an opportunity for driver devs to cheat. 

If you disagree, fine, but deal with these ideas, not whether you like the website or not... 

avatar

nekollx

Hardocp also compares videocards
running on completely different settings against each other.  How is
that, in any way, a true apples-to-apples comparison of what each card
can deliver?

we may disagree on some aspects of the bench but this is totally +1

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

Modred189

just for reference, [H] uses fraps to record their runs, and they are very consistent, see their graphs and my comment, above...

And I have never noticed them comparing runs from different systems.... I just don't think that's true... 

avatar

nekollx

 that's what i was saying. First Party Benches aren't unbiased, that why i like FRAPS

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

habuza

I'm so tired of seeing crysis. It's old cheese. My rig runs it at 50 fps on max graphics. Next please.

avatar

TheMurph

Seeing as a single ATI HD 5870 card can't even hit 40 fps by itself when Crysis is maxed out, uh, what benchmark would you prefer to use to test a videocard to its limits?

Heck, just look at the chart in the review -- no other game Maximum PC uses tests a graphics card as much as Crysis.

avatar

lancethepants

Just curious as to what Resolution 'max' is.  Do you run it at 1920 x 1200?

avatar

TheMurph

I should have specified -- maxed out, in this case, meant the 1920x1200 resolution that MPC uses for its benchmark.  That said, the card wouldn't hit 40 FPS on the 2560x1600 of a 30-inch display with all the graphics settings cranked up.

avatar

szore

What kind of rig are you running?

avatar

nekollx

 FRAPS

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

TheMurph

...is shareware, with a 30-second maximum limit for any recording.  Plus, how are you going to go about constructing the actual benchmark run without a paid-for copy of the game, assuming you can even do that for some of the titles you might want to test (like a number of games on this list).

avatar

nekollx

 but you can get FPS for free and always on, no need to hunt aroud for some demo add on, run the demo. Run fraps. See FPS. no need for 13 different bench marking tools for each game. one FPS to rule them all.

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

Muerte

Now why would you use all of these games to do basically the same test.  If you really want to just do the benchmark choose the one that best fits the rig your testing and use it.

But if you want to check out how all these games look just for shits and giggles dl them all.

 

avatar

TheMurph

You're going to base the performance of your rig on a single benchmark?  In that case, you might as well use the Windows Experience Index number and save yourself the trouble of having to install anything.

Without getting too deep into things, a wide variety of benchmarks allows you to get a complete picture of how your system--or graphics card--is actually performing.  Just running one benchmark isn't enough for accuracy's sake, as certain card and driver combinations might natively be able to handle one game beter than another.  How will you know if your card Kicks Ass...  or if it's just a factor of the game being better coded / a better fit for an individual GPU platform?

That's why Maximum PC runs many, many benchmarks on the items it tests -- and it's why you see five different games in this list instead of just one big "RUN CRYSIS" headline.

avatar

TheMurph

...you'll note that all of the aforementioned programs *are* free.  In fact, all you have to do is run the game then click on the subsequent "benchmark" or "performance test" mode. 

In your situation, you'd have to run each game, fire up fraps, record your average FPS over a limited 30-second window...  you're essentially doing the exact same thing that these games themselves do, only less of it.  Plus, you'll have to create your own benchmark runs for whatever game you're using if you want valid, repeatable results -- since the benchmark runs are identical in the games listed above, this step has already been done for you as well.

I'm not knocking Fraps.  However, it's a great deal more work than just clicking a button, sitting back, and letting the game take care of the benchmarking and FPS recording for you.

avatar

nekollx

 but in your case you have to run 5 diffenrt benches instead of one (which leads to inconsistency with multiple "versions")

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

TheMurph

What are you talking about?  Inconsistency with what?

The point is that you're running five separate benchmarks to get a completely accurate picture of how your graphics card is performing.  You can't just run a single benchmark and call it a day -- why do you think Maximum PC, or any site worth it's weight in reviewing, runs multiple tests with multiple pieces of software?  To eliminate inconsistencies and verify that a card is going to deliver its performance across a wide range of scenarios.

It would sure be a real shame to find out that a card that rocked the RE 5 benchmark is actually an underperforming piece of hardware--especially if [manufacturer] tweaked the drivers to deliver better-than-expected performance in that single title.

avatar

nekollx

 see that's my concern. How am i to know the"offical RE5 Benchmark" doesnt fudge the number on RE5. Atleast FRAPS is multi programs so theirs no bias. Ideally id run several, 3rd party, benches alas...their aren't many.

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

TheMurph

That's the point -- you run multiple benchmarks across multiple games to ensure that the numbers you're seeing are indicative of a card's performance, not a fluke. To your point, you could certainly run Fraps at the same time you run the RE 5 benchmark and see how the numbers compare. 

Fraps is one way to go about benchmarking, but there's no way to record a timed demo in RE 5.  You would thus lose that game as a benchmark.

avatar

nekollx

 yeah it's just the idea of trusting a first party bench i don't like. As for recording there is this thing called Paper they use to make. Besides whatgood id a recording? You need a average: Normal play, chaos zone/high lag, "idle"your still going to have to extrapoate the average from a recording.

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

TheMurph

Sigh.  Record a timed demo = you create a demo run, whereby your actions are repeated identically each and every time you run the benchmark.  That way, you are eliminating the external variables that could impact the FPS result you obtain, which will give you a truer indication of the card's abilities and a stronger degree of confidence for your benchmark runs.

Simply put, you can't just fire up Fraps and play your favorite level over and over -- you'll be performing different actions and seeing different elements displayed on the screen each time, which will affect your FPS results when you go to make a final comparison.  Apples-to-apples testing -- that's how it's done.

avatar

Airheadq

I have to comment on this. FRAPS isn't a benchmark, its an FPS
monitor. To be blunt, it overlays the game and spits out your fps. It
doesn't stress your hardware in anyway.

There is a need for multiple games since they are coded differently one my stress the CPU
harder than the GPU and vice-versa. FRAPS is just nice to have to
always show the fps and log it.

avatar

lancethepants

Ditto,  Fraps is  a nice application to see real time fps.  Sure you can setup your own benchmark, but getting it to run exactly the same each time would be difficult, and heavily mess up your stats if it doesn't,  Why not run these premade benchmarks which ensure each run is taxing your system exactly the same.  Plus these benchmarks are run by many people allowing for better comparison.  It will also allure you to any inconsistencies that your system may have.  For example, if it seems to be underperforming compared to comparable hardware running the same benchmark.  Additionally,  some benchmarks may be CPU bound, while others are GPU bound, or Memory Bound etc.  So any one benchmark really can't assess the true performance of you PC.

avatar

nekollx

 i think you miss understand. I'm advocating 5 Games + 3rd party bench vs 5 game+ 5 first party (potentially biased) benches.

------------------------------
Coming soon to Lulu.com --Tokusatsu Heroes--
Five teenagers, one alien ghost, a robot, and the fate of the world.

avatar

Jasker

I think you miss the point.  Not the rest of us.  How do you intend recreating a fraps clip more than once to come up with any kind of repeatable test?  You make one 30s run at a single game and call it good?  There isn't a practical way to FRAPS the excat same scene more than once, which you'd want to do.  I think you're taking the "Damn The Man" approach too far.

avatar

Modred189

We do realize that FRAPS for a nominal fee ($40) you can unlimited recording, benchmarking etc. I know that for most of us that might seem like a lot, but for the average MaxPC reader who upgrades cards and buys games like crazy, it's a pittance for the information gleaned...?

avatar

Airheadq

Since these are quote benchmarks/Demos from the developer themselves, they do give an accurate level of how your PC will perform in the
actual game. Granted that the last level in Crysis will tax your
hardware more than the demo level, its still playable. If your PC
performs well in these demos, then that is a good indication that it
will perform similarly in the full version.

Log in to MaximumPC directly or log in using Facebook

Forgot your username or password?
Click here for help.

Login with Facebook
Log in using Facebook to share comments and articles easily with your Facebook feed.