 | |||
|
|
|
|
|
 |
3D Performance with F.E.A.R. Beta June 29, 2005 |
|
 |
|
| |
Summary: With DX9 shaders, per-pixel lighting, and normal maps, Monolith's F.E.A.R. will push today's hardware to levels we haven't seen previously when it debuts later this year. In today's article we provide a preview of F.E.A.R.'s performance with the latest high-end cards, including the GeForce 7800 GTX, 6800 Ultra, and 6800 GT, in both single-card and SLI configurations, while ATI cards tested include the RADEON X850 XT Platinum Edition and RADEON X800 XL. How do the cards perform against one another in the most demanding title we've seen to date? Find out in this article!
 Introduction | Page:: ( 1 / 12 ) |
[image]
| <% print_image("01"); %> | <% print_image("02"); %> |
An unidentified paramilitary force infiltrates a multi-billion dollar aerospace compound, taking hostages but issuing no demands. The government responds by sending in special forces, but loses contact as an eerie signal interrupts radio communications. When the interference subsides moments later, the team has been obliterated.
As part of a classified strike team created to deal with threats no one else can handle, your mission is simple: Eliminate the intruders at any cost. Determine the origin of the signal. And contain the crisis before it spirals out of control.
[image]
| <% print_image("03"); %> | <% print_image("04"); %> |
Like DOOM 3, and as the game’s name implies, F.E.A.R. attempts to combine the immersive elements of a first-person shooter with the horror elements of a good scary movie. F.E.A.R. accomplishes this with a strong emphasis on the paranormal, as well as a pasty white, hunched little girl (eerily similar to the one found in the movie, The Ring) who seems to be quite adept at dismembering people without laying a finger on them. Monolith uses pacing to intensify the game’s creepiness, one moment you may be bored into a lull, walking through an apparently “safe” part of a level, followed by a terrifying supernatural event or an intense firefight. Hopefully Monolith’s pacing will be more reminiscent of Valve’s in Half-Life 2 rather than the “surprise! attack!” Jakub felt in DOOM 3.
[image]
| <% print_image("05"); %> | <% print_image("06"); %> | <% print_image("07"); %> |
The visuals
For F.E.A.R, the Jupiter engine used for No One Lives Forever 2 and TRON 2.0 has been replaced with an entirely new rendering engine, supporting 2.0 pixel and vertex shaders, and normal and specular maps. Monolith has also licensed the popular Havok physics engine for F.E.A.R., providing such features as ragdoll physics, although Monolith’s implementation doesn’t appear to go quite as far as Valve did with Half-Life 2, where a large firefight could send props in the room flying everywhere.
F.E.A.R uses an abundant amount of particle effects to model effects like sparks, one full-auto burst from your assault rifle into one of the game’s many metal grates will send them flying everywhere.
[image]
| <% print_image("08"); %> | <% print_image("09"); %> | <% print_image("10"); %> |
Arguably the most notable feature found in F.E.A.R.’s engine however is its per-pixel lighting model, just like DOOM 3. Monolith uses lighting and shadows in the game’s levels to help intensify the game’s horror mood.
| IQ Compared | Page:: ( 2 / 12 ) |
Adjusting video settings
Softer shadows
One such setting that we were eager to check out, soft shadows, has traditionally brought with it a significant performance hit. In Chronicle’s of Riddick’s 2.0++ mode for example, performance declined by a factor of 3x once soft shadows were turned on! This was the default mode for GeForce 6 users, leading many of them to mistakenly complain of performance problems with the game. You can see screenshots of the effect here:
[image]
| <% print_image("11"); %> | <% print_image("12"); %> |
Pay careful attention to the shadows being cast by the soldiers, particularly the edges of the shadows. Note how hard the edges of the shadows are when soft shadows is disabled. You can also see this in the shadow of the helicopter as well as the shadow of the platform it rests on.
We’ll examine the performance impact of soft shadows later in this article.
Anti-aliasing
As usual, we were eager to see how well ATI and NVIDIA’s latest and greatest cards handled AA in F.E.A.R. Like DOOM 3, the game doesn’t feature an abundant amount of jaggies (which is only helped further by the game’s dark environments), so AA isn’t as crucial as it is in other titles, but it’s still a feature we suspect most gamers with high-end cards will definitely be turning on.
With that in mind, let’s take a look at the screenshots:
[image]
| <% print_image("13"); %> | <% print_image("14"); %> |
| <% print_image("15"); %> | <% print_image("16"); %> |
As you can see, there’s a bug in the game for GeForce cards that prevents the wall on the left hand side from being rendered properly. Considering the beta nature of the game, this shouldn’t come too unexpected, as Monolith likely still has a number of issues with the game that must be resolved before its release. The NVIDIA board appears to do a slightly better job at clearing up the jaggies on the edge of the catwalk across from you, but as you can see in the screenshots, your position in the NVIDIA shot is slightly left of your character in the ATI screenshot. This could account for some of the difference we’re seeing in both images. We’ll just have to wait for the final game, where we can hopefully record proper demos for testing AA.
| Test Systems | Page:: ( 3 / 12 ) |
System Setup
AMD Athlon 64 FX-55
ASUS A8N-SLI Deluxe
1GB OCZ DDR400 SDRAM
ATI RADEON X850 XT PE
ATI RADEON X800 XL
Driver version CATALYST 5.6
CATALYST A.I. Left at Standard
NVIDIA GeForce 7800 GTX
NVIDIA GeForce 6800 Ultra
NVIDIA GeForce 6800 GT
Driver version 77.72
NVIDIA Image Quality Settings Left at Quality
250GB Maxtor Hard Drive Maxline III SATA Hard Drive w/16MB Cache
Windows XP Professional SP1
DirectX 9.0c
Benchmarks
F.E.A.R Beta Maximum Detail Settings
Notes
For testing the game, Monolith provides a built-in demo for benchmarking, think Half-Life 2 video stress test and you’ll have an idea of what we’re talking about. Unlike Valve’s VST though, Monolith’s test provides a limited amount of gameplay footage, as well as very in-depth statistics. Not only do you get the average frames per second and the minimum/maximum, F.E.A.R.’s built-in demo also provides data on the percentage of frames that ran at below 25 frames per second, the percentage between 25-40 FPS, and finally, the percentage of frames above 40 FPS.
Based on this data, the game then provides basic recommendations on image quality settings. We noticed that the game tends to recommend that you crank up the detail level if the percentage of frames above 40 FPS exceeds 40%. If your system can’t maintain greater than 40 FPS over 40% of the time, then the game recommends that you turn down your settings. Just food for thought, every person will likely have a different ideal frame rate that they’re targeting.
| F.E.A.R. Performance | Page:: ( 4 / 12 ) |
| F.E.A.R. Performance 1280x1024x32 | ||
| Card | Min FPS | Max FPS |
| GeForce 7800 GTX SLI | 29 | 343 |
| GeForce 7800 GTX | 20 | 347 |
| GeForce 6800 Ultra SLI | 12 | 330 |
| GeForce 6800 Ultra | 10 | 314 |
| GeForce 6800 GT | 9 | 281 |
| RADEON X850 XT PE | 11 | 290 |
| RADEON X800 XL | 9 | 269 |
| ||
| F.E.A.R. Performance 1280x1024x32 | |||
| Card | % Below 25 FPS | % Between 25-40 FPS | % Above 40 FPS |
| GeForce 7800 GTX SLI | 0 | 5 | 95 |
| GeForce 7800 GTX | 4 | 29 | 67 |
| GeForce 6800 Ultra SLI | 6 | 16 | 78 |
| GeForce 6800 Ultra | 9 | 42 | 49 |
| GeForce 6800 GT | 27 | 37 | 36 |
| RADEON X850 XT PE | 19 | 49 | 32 |
| RADEON X800 XL | 43 | 34 | 23 |
| |||
| Soft Shadows | Page:: ( 5 / 12 ) |
Notes
Despite all our attempts, we were unable to run the SLI setups with soft shadows once AA/AF were disabled (fortunately we were able to get it to work for all further testing). And while we didn’t include tables with min/max and percentage data, we did note the differences, the max and min data weren’t affected all that much, for example the 6800 Ultra went from 10 FPS minimum without soft shadows to 8 FPS once soft shadows were turned on, while the max frame rate dropped from 314 FPS without, to 282 FPS once soft shadows were turned on. The most important data are the percentages, the 7800 GTX went from frame rates below 25 FPS 6% of the time without soft shadows, up to 31% once the feature was turned on, while the RADEON X850 XT PE’s percentage of frames below 25 FPS went from 19% to 66%. The 6800 GT and Ultra weren’t immune from the performance impact either, going from 27% and 9% respectively without soft shadows to 64% and 55%.
| 4xAA | Page:: ( 6 / 12 ) |
| F.E.A.R. Performance 1280x1024x32 | ||
| Card | Min FPS | Max FPS |
| GeForce 7800 GTX SLI | 16 | 318 |
| GeForce 7800 GTX | 9 | 279 |
| GeForce 6800 Ultra SLI | 10 | 314 |
| GeForce 6800 Ultra | 8 | 255 |
| GeForce 6800 GT | 6 | 254 |
| RADEON X850 XT PE | 5 | 260 |
| RADEON X800 XL | 4 | 231 |
| ||
| F.E.A.R. Performance 1280x1024x32 | |||
| Card | % Below 25 FPS | % Between 25-40 FPS | % Above 40 FPS |
| GeForce 7800 GTX SLI | 8 | 27 | 65 |
| GeForce 7800 GTX | 32 | 37 | 31 |
| GeForce 6800 Ultra SLI | 20 | 31 | 49 |
| GeForce 6800 Ultra | 46 | 35 | 19 |
| GeForce 6800 GT | 58 | 32 | 10 |
| RADEON X850 XT PE | 61 | 21 | 18 |
| RADEON X800 XL | 73 | 14 | 13 |
| |||
| Performance Impact of 4xAA | Page:: ( 7 / 12 ) |
| 4xAA w/Soft Shadows | Page:: ( 8 / 12 ) |
| 4xAA/8xAF | Page:: ( 9 / 12 ) |
| F.E.A.R. Performance 1280x1024x32 | ||
| Card | Min FPS | Max FPS |
| GeForce 7800 GTX SLI | 15 | 319 |
| GeForce 7800 GTX | 9 | 281 |
| GeForce 6800 Ultra SLI | 10 | 304 |
| GeForce 6800 Ultra | 8 | 255 |
| GeForce 6800 GT | 6 | 236 |
| RADEON X850 XT PE | 4 | 235 |
| RADEON X800 XL | 3 | 225 |
| ||
| F.E.A.R. Performance 1280x1024x32 | |||
| Card | % Below 25 FPS | % Between 25-40 FPS | % Above 40 FPS |
| GeForce 7800 GTX SLI | 9 | 27 | 64 |
| GeForce 7800 GTX | 35 | 34 | 31 |
| GeForce 6800 Ultra SLI | 24 | 33 | 43 |
| GeForce 6800 Ultra | 50 | 34 | 16 |
| GeForce 6800 GT | 61 | 29 | 10 |
| RADEON X850 XT PE | 64 | 19 | 17 |
| RADEON X800 XL | 75 | 13 | 12 |
| |||
| 4xAA/8xAF w/Soft Shadows | Page:: ( 10 / 12 ) |
| Transparency AA | Page:: ( 11 / 12 ) |
| Conclusion | Page:: ( 12 / 12 ) |
F.E.A.R. is an ATI Get In The Game partner, so it’s possible that Monolith hasn’t seen G70, and therefore hasn’t had any time to integrate optimizations for the card into their game just yet. Likewise, NVIDIA probably hasn’t had an opportunity to optimize their latest card for F.E.A.R., so it works both ways. This could potentially suggest why the GeForce 7800 GTX card took such a large hit once 4xAA was enabled. In any case, the GeForce 7800 GTX was able to come out on top, although the ratio wasn’t nearly as dramatic as what we saw last week with other titles, only outperforming its predecessor by roughly 20% in most situations for the most part. Again, we attribute this to the beta nature of the software.
The older high-end cards, the GeForce 6800 Ultra/GT and RADEON X850 XT Platinum Edition/X800 XL still perform well with F.E.A.R. as long as you keep the detail settings in check. In terms of performance, our tests reveal a performance edge for NVIDIA cards.
NVIDIA’s stencil buffer performance has always been a strong suit for GeForce 6, so that may not come as much of a surprise to some of you. With 4xAA/8xAF enabled, the GeForce 6800 Ultra and 6800 GT swept all tests, finishing first and second over the ATI cards, even once soft shadows were turned on. Fortunately for ATI the margins weren’t huge, with 4xAA/8xAF the 6800 GT outperforms the X800 XL by 15-25%, with the result depending on resolution, while the X850 XT PE trims the margin down to 13-19% behind the 6800 Ultra thanks to a greater clock speed advantage in favor of the X850 XT card.
Soft shadows still come with a significant performance hit, although as we showed you on page 5 the maximum and particularly the minimum weren’t affected too badly; the most telling story were the percentages. Enabling soft shadows just drags down your overall frame rate. Sure, it does hit your peak FPS badly, but the percentage of frames that dropped below 25 FPS shot up dramatically once soft shadows were turned on. Fortunately the drain wasn’t quite as high as what we saw with Riddick, but we still have a feeling that this setting will likely be turned off by most users, even 7800 GTX owners.
Monolith is still hard at work on tweaking F.E.A.R. though, so we’ll just have to wait and see how the final results play out once the game is available. Hopefully by then ATI’s CrossFire solution should be well established in the driver department (our sources indicate that’s what’s holding up CrossFire right now, the hardware is ready), while R520 will also be out the door. If so, it will be interesting to see how G70 fares against its competition. We can’t wait!