3D Performance with F.E.A.R. Part 1: Mainstream Cards October 20, 2005 Brandon Sandman Bell

Introduction

Page:: ( 1 / 9 )

Up for today is our F.E.A.R. performance with mainstream graphics cards article. Since getting our hands on a beta copy of F.E.A.R. a few months ago, we’ve been throwing every new graphics card on the market at this game. As many of you have seen with the single-player and multiplayer demos VU Games released earlier this year, F.E.A.R. is capable of taxing even the latest hardware. Monolith has replaced the Jupiter engine used in their previous games such as TRON 2.0 and No One Lives Forever 2, with a completely new engine, which is filled with all kinds of eye candy.

Monolith uses high resolution normal and specular maps throughout the game’s dark environments, while particle effects are used for effects like sparks, which can go flying everywhere in the middle of a heated firefight. F.E.A.R. also features a per-pixel lighting model, just like DOOM 3 last year. It’s perhaps because of the game’s per-pixel lighting model and Monolith’s extensive use of shadows that F.E.A.R. is a game that pushes hardware like never before. Between Call of Duty 2, Quake 4, and Serious Sam II, F.E.A.R. seems to be the most taxing. This doesn’t necessarily mean F.E.A.R. is the most advanced engine out there, as the game certainly lacks support for a few key visual features, but it is something you’ll want to keep in mind before playing the game.

With this in mind, we’ve decided to buck tradition and focus on the performance of the various mainstream card offerings first, before moving on to the high-end cards like the GeForce 7800 GTX and RADEON X1800 XT (normally we go in the other direction). There were a few key topics that we wanted answers to, we’ll quickly go over them:

Topic #1: Medium Quality vs. Maximum Quality

Obviously with F.E.A.R. requiring so much graphics horsepower, the number one question gamers with lower-end cards are probably asking is what’s the difference in performance and image quality between the two settings?

We’ll provide a more in-depth response to this question in an upcoming game optimization guide, but until then we wanted a quick answer, which you’ll find in the upcoming pages. Fortunately our first impressions of the game in medium quality mode are that you don’t lose too much in terms of eye candy, but as you’ll see in the benchmarks, you don’t gain that much additional performance either. Take a look at these screenshots:

[image]

Topic #2: 512MB vs 256MB graphics card

While this probably isn’t pertinent to the mainstream graphics card segment, we still wanted to get a quick answer to this question. Our testing with Sapphire’s RADEON X800 XL 512MB showed some pretty surprising performance gains with the 512MB board. In fact, the Sapphire X800 XL 512MB outperformed the RADEON X850 XT Platinum Edition once 4xAA/16xAF was turned on.

While we’re running with lower image quality settings in today’s article, we wanted to see if the same performance trends continued to hold up.

Test Systems

Page:: ( 2 / 9 )

System Setup

AMD Athlon 64 FX-57

ASUS A8N32-SLI Deluxe

1GB OCZ DDR400 SDRAM

ATI RADEON X800 XL 256MB
ATI RADEON X1600 XT
ATI RADEON X700 PRO 256MB
Sapphire RADEON X800 XL 512MB
Driver version 8_183_1017.zip (beta of CATALYST 5.10a)

NVIDIA GeForce 6800 GT
NVIDIA GeForce 6600 GT
MSI GeForce NX6800 256MB
Driver version ForceWare 81.85

250GB Maxtor Hard Drive Maxline III SATA Hard Drive w/16MB Cache

Windows XP Professional SP1

DirectX 9.0c

Benchmarks

F.E.A.R. 1.02

F.E.A.R.

Page:: ( 3 / 9 )

F.E.A.R. Performance 1024x768x32
Card	Min FPS	Max FPS
GeForce 6800 GT	35	138
GeForce 6800	30	87
GeForce 6600 GT	28	91
RADEON X800 XL 512MB	32	109
RADEON X800 XL 256MB	33	110
RADEON X800	20	73
RADEON X1600 XT	29	81
RADEON X700 PRO	18	57

F.E.A.R. Performance 1024x768x32
Card	% Below 25 FPS	% Between 25-40 FPS	% Above 40 FPS
GeForce 6800 GT	0	4	96
GeForce 6800	0	43	57
GeForce 6600 GT	0	50	50
RADEON X800 XL 512MB	0	17	83
RADEON X800 XL 256MB	0	19	81
RADEON X800	3	64	33
RADEON X1600 XT	0	32	68
RADEON X700 PRO	24	64	12

F.E.A.R. Max quality

Page:: ( 4 / 9 )

F.E.A.R. Performance 1024x768x32
Card	Min FPS	Max FPS
GeForce 6800 GT	29	137
GeForce 6800	21	84
GeForce 6600 GT	21	83
RADEON X800 XL 512MB	27	103
RADEON X800 XL 256MB	28	104
RADEON X800	19	70
RADEON X1600 XT	23	83
RADEON X700 PRO	15	57

F.E.A.R. Performance 1024x768x32
Card	% Below 25 FPS	% Between 25-40 FPS	% Above 40 FPS
GeForce 6800 GT	0	11	89
GeForce 6800	3	53	44
GeForce 6600 GT	3	55	42
RADEON X800 XL 512MB	0	28	72
RADEON X800 XL 256MB	0	28	72
RADEON X800	14	57	29
RADEON X1600 XT	3	44	53
RADEON X700 PRO	38	50	12

To see the complete max vs min quality performance results, we recommend you open up a new browser window. We'll provide a quick summary at 1024x768. The GeForce 6800 GT and 6600 GT's performance drops by 7%, while the 6800 sees a drop of 6%. Turning up to max quality only results in a 5% performance dropoff for the X800 XL, while the X800 and X1600 XT see a reduction of 8% and 4% respectively in performance.

F.E.A.R. 2xAA

Page:: ( 5 / 9 )

F.E.A.R. Performance 1024x768x32
Card	Min FPS	Max FPS
GeForce 6800 GT	31	116
GeForce 6800	23	76
GeForce 6600 GT	21	71
RADEON X800 XL 512MB	26	94
RADEON X800 XL 256MB	29	95
RADEON X800	17	63
RADEON X1600 XT	22	75
RADEON X700 PRO	12	48

F.E.A.R. Performance 1024x768x32
Card	% Below 25 FPS	% Between 25-40 FPS	% Above 40 FPS
GeForce 6800 GT	0	26	74
GeForce 6800	3	73	24
GeForce 6600 GT	12	64	24
RADEON X800 XL 512MB	0	42	58
RADEON X800 XL 256MB	0	38	62
RADEON X800	29	53	18
RADEON X1600 XT	5	66	29
RADEON X700 PRO	72	25	3

F.E.A.R. 4xAA

Page:: ( 6 / 9 )

Far Cry Performance 1024x768
Card	Min FPS	Max FPS
GeForce 6800 GT	23	89
GeForce 6800	16	59
GeForce 6600 GT	16	54
RADEON X800 XL 512MB	18	79
RADEON X800 XL 256MB	17	79
RADEON X800	10	48
RADEON X1600 XT	14	60
RADEON X700 PRO	8	36

F.E.A.R. Performance 1024x768x32
Card	% Below 25 FPS	% Between 25-40 FPS	% Above 40 FPS
GeForce 6800 GT	5	58	37
GeForce 6800	48	47	4
GeForce 6600 GT	75	22	3
RADEON X800 XL 512MB	27	48	25
RADEON X800 XL 256MB	24	45	31
RADEON X800	79	16	5
RADEON X1600 XT	51	38	11
RADEON X700 PRO	90	10	0

F.E.A.R. 2xAA/8xAF

Page:: ( 7 / 9 )

F.E.A.R. Performance 1024x768x32
Card	Min FPS	Max FPS
GeForce 6800 GT	33	114
GeForce 6800	23	77
GeForce 6600 GT	21	74
RADEON X800 XL 512MB	26	89
RADEON X800 XL 256MB	26	93
RADEON X800	16	62
RADEON X1600 XT	21	71
RADEON X700 PRO	11	45

F.E.A.R. Performance 1024x768x32
Card	% Below 25 FPS	% Between 25-40 FPS	% Above 40 FPS
GeForce 6800 GT	0	33	67
GeForce 6800	3	72	25
GeForce 6600 GT	16	64	20
RADEON X800 XL 512MB	0	39	61
RADEON X800 XL 256MB	0	40	60
RADEON X800	38	48	14
RADEON X1600 XT	11	62	27
RADEON X700 PRO	72	25	3

F.E.A.R. 2xAA/8xAF Max vs Med Quality

Page:: ( 8 / 9 )

F.E.A.R. Performance 1024x768x32
Card	Min FPS	Max FPS
GeForce 6800 GT	25	98
GeForce 6800	18	75
GeForce 6600 GT	16	73
RADEON X800 XL 512MB	22	88
RADEON X800 XL 256MB	26	89
RADEON X800	16	58
RADEON X1600 XT	20	67
RADEON X700 PRO	11	43

F.E.A.R. Performance 1024x768x32
Card	% Below 25 FPS	% Between 25-40 FPS	% Above 40 FPS
GeForce 6800 GT	0	47	53
GeForce 6800	14	64	22
GeForce 6600 GT	31	51	18
RADEON X800 XL 512MB	1	55	44
RADEON X800 XL 256MB	0	50	50
RADEON X800	46	40	14
RADEON X1600 XT	16	62	22
RADEON X700 PRO	75	24	1

Conclusion

Page:: ( 9 / 9 )

Between the two cards, NVIDIA’s GeForce 6800 GT comes out on top, which holds up with the results we first showed you in our first 3D Performance with F.E.A.R. beta article back in June. The GeForce 6800 GT generally ran between 8-15% faster than the RADEON X800 XL (although there were a few cases when the margin separating both cards was greater than that), with the result varying based on resolution and amount of AA/AF tested. The RADEON X800 XL trailed the GeForce 6800 GT pretty significantly under base and maximum quality settings without AA/AF, but the margins narrowed once those settings were turned on. We didn’t see any performance benefit from the 512MB RADEON X800 XL however. In fact, it generally fell slightly behind the 256MB X800 XL in performance. We attribute this to the added latency that the additional memory brings.

At the ~$200 price point, the GeForce 6800 and RADEON X1600 XT compete fiercely, with the GeForce 6800 winning early testing without AA/AF, and the X1600 XT generally pulling even of the GeForce 6800 by the time AA and AF are turned on (in fact, it pulls slightly ahead in a few cases). We’d call this one a draw in F.E.A.R. performance.

Of course, the difference here is that the GeForce 6800 can be found at retail now, while the X1600 XT won’t hit shelves for another month (according to ATI). ATI’s also listed an MSRP of $249 for the RADEON X1600 XT, making it more expensive than the GeForce 6800. Factoring all this in, we’ve got to give the nod to the GeForce 6800 at this point.

Keep in mind that we tested a 256MB GeForce 6800 today, the 128MB GeForce 6800 SKUs can be found for about $20-$30 less in general, but we’d recommend spending the extra money and getting the 256MB card as F.E.A.R. becomes pretty choppy on 128MB cards, especially once you crank up the texture resolution setting in the game. This is probably one of the first settings you should adjust when you’re tweaking F.E.A.R for more performance.

The GeForce 6600 GT continues to put up a strong showing, delivering better performance than the 12-pipeline RADEON X800 for the most part, but with the lowest prices hovering at about $150, it’s beginning to be priced too closely to the GeForce 6800. The $100 gap between the two cards that existed just a few months ago has eroded to about $10 or $20 at most e-tailers, as NVIDIA’s had to make room for price cuts to their higher-end GeForce cards with the introduction of the 7800 GT and 7800 GTX.

In fact, the 7800 GT can be found for about $360 pretty easily online, making it a tempting solution if you want to splurge for a little bit more than a 6800 GT. By comparison, one GeForce 7800 GT runs about as fast as two GeForce 6800 GTs, making it a substantial upgrade in performance.

We’ll be taking a look at this in our next F.E.A.R. performance article. Stay tuned…