*** The official FEAR demo benchmarking thread

[nitromullet]

Originally posted by: Stangs55
And here are my results from the Opening Cutscene.

From Eyeball -> Just got out of the car

Frames - 39230
Time - 294673
Min - 0
Max - 1276
Avg - 133.131

However I don't think that this will be a very good judge at all b/c it takes into account the blank screens when nothing but text is displayed as well as the loading screen. People with faster computers but identical video cards to others should have lower scores b/c their loading screen will load faster thereby decreasing the amount of extremely high fps's that are avg'ed into the final score.

Are you using SFR, AFR, or AFR2? How about AA options? You can make the game look coniderably better IMO by running with 4x transparent SSAA than by using the AA available in the game. I would assume that there is a performance hit with SSAA though.

 

[hans030390]

Just to those who dont re-read edited posts or aren't aware of them, I updated mine with more benchies.

 

[Stangs55]

Originally posted by: nitromullet

Originally posted by: Stangs55
And here are my results from the Opening Cutscene.

From Eyeball -> Just got out of the car

Frames - 39230
Time - 294673
Min - 0
Max - 1276
Avg - 133.131

However I don't think that this will be a very good judge at all b/c it takes into account the blank screens when nothing but text is displayed as well as the loading screen. People with faster computers but identical video cards to others should have lower scores b/c their loading screen will load faster thereby decreasing the amount of extremely high fps's that are avg'ed into the final score.

Are you using SFR, AFR, or AFR2? How about AA options? You can make the game look coniderably better IMO by running with 4x transparent SSAA than by using the AA available in the game. I would assume that there is a performance hit with SSAA though.

I have the nvidia panel set to application controlled everything and normal AA rendering (as opposed to SLI rendering). No SSAA...no Transparency AA

 

[Munky]

Originally posted by: hans030390
Just to those who dont re-read edited posts or aren't aware of them, I updated mine with more benchies.

I just updated the OP with some new ideas based on your custom settings. I want to isolate the shaders and test the benefits of sm3. So, try running the demo again with the new custom settings.

 

[nitromullet]

I have the nvidia panel set to application controlled everything and normal AA rendering (as opposed to SLI rendering). No SSAA...no Transparency AA

try the transparent 4x SSAA... It looks great.

 

[hans030390]

Thats a really good idea munky (not exactly how well it will test the Sm3 part, but I'm assuming it will). Anyways, I just updated my previous post with results from my tweaked settings (the new one's set by munky that is). This should be stickied for now too...

 

[Ronin]

Query: Doesn't the SP demo have the same thing the MP demo did, where it did about 2 minute run to figure out what settings were best to use?

EDIT: Just checked..obviously, it's not there heh.

 

[hans030390]

No. It has auto detect which takes 1 second.

 

[hans030390]

OH! I have a suggestion (for those with 6 or 7 series cards). There was a program that edited the config file/s for games....it allowed you to tell the program you were messing with what video card you were running. Now, you could use that to tell the game you're running an ATI card, and it should put it in Sm2. Benchies would probably confirm or deny this.

It wasn't my idea, but someone's on my Sm3 on fear post. I'm gonna look for that program. I used it once before. And yes, it does work.

Be back in a bit.

 

[hans030390]

Possible way to play Fear in Sm2 with Sm3 cards (not tested yet...but it should work)

Go to google, look for 3danalyze...i dont remember exactly how to use it. But that should do the trick.

crap im not seeing a sm2 option, i think you need to tell it to run it as a 9800 or something.

I did benchies with it running as a "9800" and the results were the same as the 6600gt.
I'm thinking that 3danalyze wont run right for FEAR. We'll probably have to find some command to change shaders because I've never seen a case where Sm3 provides no performance boost, but I know it benefitted the 6800gt in the FEAR mp beta...hm.

Welp, i'm stumped now.

 

[MBrown]

Frames, Time (ms), Min, Max, Avg
1357, 46371, 17, 61, 29.264

 

[Munky]

I updated my benches with the second set of options. So far I havent seen anyone with a 6600gt match the x800, but I'm wondering if a 6800 vanilla might pull ahead thanks to sm3. We need more benches!

 

[Munky]

I also ran the benches on my bro's rig:

A64 3200
6600gt
1 gig ram

Test 1:
Avg: 26.503 - Min: 10 - Max: 109

Test 2:
Avg: 29.814 - Min: 16 - Max: 72

 

[Munky]

Originally posted by: CrystalBay
did you run up the lumber pile or go around to the steps?

2005-08-07 23:26:27 - FEARSPDemo
Frames: 3769 - Time: 42099ms - Avg: 89.527 - Min: 53 - Max: 205


Opening cut scene, from the eyeball to the loading screen.

2005-08-07 23:39:18 - FEARSPDemo
Frames: 29692 - Time: 267051ms - Avg: 111.185 - Min: 7 - Max: 1496





A64 4000 @ 2626
2048 ram
bfg6800gt

Those min fps are impressive. I'm thinking the 2 gigs of ram goes a long way to prevent stuttering, even at these medium settings

 

[hans030390]

Hm, you know, is this game even running in SM3? I know the MP beta was...but for all we know, the sp demo might not be.

I mean, theres no option for it and no one knows any commands for it...

Hm...and trying to tell the game i was using a 9800 didn't change anything. Hm...

It's interesting how the 6800gt beat the x850 nicely, yet it seems that similar 6 series cards might not benefit as well? or the above. I don't know.

I think right now, benchies should be less important than trying to figure out how to run the game in different shader modes. argh, im getting confused. Im gonna lay off the whole FEAR benchies and sm3 and sm2 on it for now. play some battle grounds....

 

[blckgrffn]

Hey, just wanted to report in that @ High CPU settings and minmal GPU settings, my Ti4200 @300/550 along with a 2.66P4 @ 2.8 with 1 gig of PC2100 @ 2-3-3-6 ran it fine and playable at 1280*1024. My 9700 Pro had a much harder time of it a just at the the medium GPU level.

That means this game must scale a bit better than Doom3 then, more like HL2 on older hardware. That is a good thing!

 

[hans030390]

Yup, and its just a demo too.

You know, I think the benchmarks should be in a different area of the demo because 1. the first part runs slower for some reason 2. there is not much activity in the first part.

Perhaps we could get a benchy of someone doing the whole thing? i'll probably do that sometime soon.

 

[Munky]

If I could figure out how to record a demo, then we could benchmark it. But playing the game manually while shoting baddies is not a good way to run benchmarks, even less so than running around with no action.

I'm off to tweak some more settings, I need to figure out why the game stutters at medium settings on a 2.5ghz A64.

 

[CrystalBay]

I'm not so sure it's easy to get real accurate numers without a consistant , replayable scene . It just too variable.


This is on high physics, high graphics, soft shadows on

2005-08-08 15:39:27 - FEARSPDemo
Frames: 2493 - Time: 48048ms - Avg: 51.886 - Min: 32 - Max: 112

 

[hans030390]

Originally posted by: munky
If I could figure out how to record a demo, then we could benchmark it. But playing the game manually while shoting baddies is not a good way to run benchmarks, even less so than running around with no action.

I'm off to tweak some more settings, I need to figure out why the game stutters at medium settings on a 2.5ghz A64.

Perhaps you need 2gb of ram? I heard that solves most of the problems. Plus it is a demo.

Ok, here im going to post a nice article in a separate thread. Look for it. You might like it.

 

[Munky]

Originally posted by: hans030390

Originally posted by: munky
If I could figure out how to record a demo, then we could benchmark it. But playing the game manually while shoting baddies is not a good way to run benchmarks, even less so than running around with no action.

I'm off to tweak some more settings, I need to figure out why the game stutters at medium settings on a 2.5ghz A64.

Perhaps you need 2gb of ram? I heard that solves most of the problems. Plus it is a demo.

Ok, here im going to post a nice article in a separate thread. Look for it. You might like it.

Yeah, that was my guess also. Although it's a little disturbing to think that a game at medium settings already requires 2 gigs or ram.

 

[CrystalBay]

Couple of hallway shots2

 

[hans030390]

what were those screens for...

 

[imported_g33k]

Originally posted by: hans030390
Yup, and its just a demo too.

You know, I think the benchmarks should be in a different area of the demo because 1. the first part runs slower for some reason 2. there is not much activity in the first part.

Perhaps we could get a benchy of someone doing the whole thing? i'll probably do that sometime soon.

I already benched the whole demo, look upthread. Average 60fps with all medium settings, 10x7 and 4aa/8af.

 

[5150Joker]

Here's what I got using the revised settings in the OP's post:

Frames: 6392
Time: 46978
Min: 72
Max: 212
Avg: 136.064

System:
A64 3000+@2.6 GHz
2 GB Crucial PC3200
DFI Lanparty NF4 SLI-DR
Leadtek 7800 GTX using EVGA KO bios @ 490/1300