The Inquirer-Home

Battlefield 2: You won't need a new PC

Part Two Tests
Wed Jun 15 2005, 23:39
IF YOU'RE HOPING to get the most out of the latest game in the very successful Battlefield series, Battlefield 2, you might need a new graphics card, but you probably won't need a new PC - according to tests conducted by The INQUIRER.

Battlefield 2, successor to the hugely popular Battlefield 1942, was developed by Digital Illusions in Sweden and distributed by Electronic Arts. It will be released next week.

The game, a multiplayer first person shooter combined with vehicular combat and squad-based tactics, looks likely to be as popular as its multi-million selling predecessors, but will you need to spend a fortune on upgrades to play it? Well, maybe it depends which of these two pictures you would prefer to see:



The INQUIRER has tested the demo version of the game with various hardware configurations. These include high-end systems and a low-end system, the latter designed to emulate the average legacy PC purchased two to four years ago. Note that the tests used Fraps to measure frames per second while playing back the same recorded demo on all systems - something that wasn't possible with earlier games in the Battlefield series. We expect performance of the game to be very similar to the demo, since the game 'went gold' several weeks before the demo was released - suggesting there is little difference between their respective graphics code. Of course, patches are likely, possibly as early as the day of release - the in game server browser is currently a nightmare.

To our surprise, we found that the game does not need an especially powerful CPU, and can run okay in 512MB of memory. We were able to run the game acceptably on a CPU underclocked to only 1GHz. What Battlefield 2 does need is a decent graphics card, and plenty of main memory bandwidth. That latter characteristic means it responds very well to front side bus overclocking. The results of the memory and CPU tests are towards the end of this article.

According to Electronic Arts, the minimum required hardware for the game is a 1.7 GHz CPU, and 512MB of main memory. Graphics cards older than the Nvidia GeForce 5600 series are not supported. Officially, EA says you need Windows XP to play, though we had no problem running it on Windows 2000.

At the moment, it is apparently impossible to play the game online on Nvidia GeForce 4 cards. Although hacks exist to make the game work with these cards, according to TweakGuides those hacks are detected as cheats by the game's Punkbuster anti-cheat system - so using them will get you kicked or banned from public servers. Note that the game's graphical quality on GeForce 4 series cards is likely to be even lower than the low-end cards described in this article

Low end systems
You can play this game on a fairly low end system, however it will look far better on a more powerful PC. We first tested the game with a Gigabyte GeForce 5700LE graphics card on an Abit IS7-V2 mainboard with a 2.1GHz Pentium 4 and 1GB of memory running at a non-standard 133MHz bus speed (see the end of the article for more details of the test hardware). This is close to the minimum required system specifications for Battlefield 2. At low resolutions (1024x768), with most of the DirectX 9 effects turned off, the game ran well. However, with these settings, the graphical glory was sadly diminished.


At these low quality settings, the visuals were actually very similar to those of an earlier game in the Battlefield series, Battlefield Vietnam, which was released over a year ago. Look at the topmost screenshot at the beginning of this article for an example of how Battlefield 2 looks with all the expensive effects turned off. The screenshot below that shows the game running on a GeForce 6800GT card.

When we tried enabling some of the graphical eye candy on the 5700LE, performance slumped to under 20 frames per second. That's well below the 30 to 40 FPS necessary for playability. Some of the effects, such as dynamic shadows, simply didn't seem to work on the 5700LE. Nvidia's new budget chip, the GeForce 6200, produced slightly better results, but still struggled to show the game in all its magnificence.


High end systems
On a high end system, we saw much better results. This Biostar 330P SFF PC is based around the Nvidia Nforce 4 chipset, and contains an AMD Athlon 64 CPU and 1GB of memory. A GeCube ATI X700 Pro graphics card made it obvious that this is a game written for DirectX 9 hardware. The ATI Radeon X700 Pro was capable of running Battlefield 2 anti-aliased at high resolutions (1280x960) with most effects at their highest settings. Despite having only 128 MB of memory, the card was able to handle the game fine at its medium texture details setting. However, high detail textures were too much for it, leading to severe drop outs in frame rate when textures needed to be shifted from main memory. Informal tests showed that a 128MB Geforce 6600 card delivered similar results to the X700 Pro.


If money is no object, then the Nvidia GeForce 6800GT is probably the ideal card for Battlefield 2 (the ATI X800 XL cards generally give similar performance in DX9 games, but unfortunately we didn't have one available to test). Despite having 'only' 256MB of memory on board, the Albatron GeForce 6800GT had no problem with the highest texture detail settings and ran the game with all graphical quality settings maxxed out at 1280x960. We tested this card in both the Biostar 330P system, and on an Albatron K8SLI motherboard. These are both based around the Nvidia Nforce 4 chipset, and with the same CPU and graphics card, performance was almost identical.


SLI: do you need three megapixels?
If one 6800GT isn't enough for you, how about two? Nvidia's drivers already support SLI for Battlefield 2, and the demo version, with Alternate Frame Rendering. So we tested a pair of Albatron 6800GT cards in SLI mode. Frankly, this is magnificent overkill - unless you have an extremely high resolution screen capable of displaying two or three million pixels, or demand framerates consistently over 70 FPS. Still, the results are interesting, and hopefully give an idea of the performance boost SLI might give if you have a pair of lesser cards.


LCDs at 1280x1024
Why didn't we benchmark the game at the common LCD monitor resolution of 1280x1024? Because the game wouldn't let us. You need to set that resolution with command line options (+szx 1280 +szy 1024), and although we could play the game at 1280x1024, it crashed when we tried to play back a demo at that resolution. Anyway, we did test at 1280x960 and there is almost no difference in performance between that resolution and the LCD's preferred 1280x1024.

Memory and CPU
Most of our tests were conducted with 1GB of main memory. When we reduced this to 512MB in a couple of our test systems, we were surprised to find that performance impact was fairly minor, at around ten to fifteen percent. The exception to this was when we set high texture detail, which had a severe effect on frame rate. What this means is, if you have low end graphics card, 512 MB is enough. That's because you're already restricted to low detail textures by the card's limitations.

We tried underclocking our Athlon 64 CPU. Interestingly, reducing the CPU frequency from its stock setting of 2.4 GHz down to 1 GHz had little effect on frame rate in our high end system. What did have a dramatic effect on performance was reducing the main memory frequency. We haven't yet repeated these types of tests on our low end system, since it is less simple to separate the effects of CPU and memory clocks on the Pentium 4 - but we will do so when this article is updated next week. In the interim, we would strongly recommend that if you're running the game on an older system, you investigate increasing the memory bandwidth by overclocking the front side bus. This seems to give substantial gains in peformance.

Some explanation of the graphs: Generally these tests were conducted with all graphics quality settings inside Battlefield 2 set to either low, medium or high. The setting used is stated on the graph. In all cases, however, view distance was increased to 100 percent. Anti-aliasing was not used except where noted on the graph. Vertical refresh synchronization was turned off in the drivers, or by using RivaTuner. Frames per second were averaged at one second intervals during the course of the two minute long recorded demo. There's a distinct upper limit on some graphs because Battlefield 2 will never redraw the screen faster than 100 frames per second. Pay particular attention to the lower frame rates recorded on the Fraps graph - FPS falling below about 30 makes the game noticeably harder to play.

We will update this article with further tests when Battlefield 2 is released next week.

Drivers used
Nvidia cards: Forceware 77.30 (included with the Battlefield Demo)
ATI cards: Version 5.6

Hardware tested
Low end system
Abit IS7V-2 mainboard (Intel 848P chipset) with 1GB of DDR 400 SDRAM (2x512MB). Pentium 4 (Northwood) 1.6A CPU, overclocked to 2.1 GHz (by increasing bus speed from 100 to 133MHz) for all tests, except where noted. Dual 7200RPM, 8MB cache hard disk drives. Topower 370W PSU.

AGP 8X Graphics Cards: Gigabyte GeForce 5700LE graphics card and Albatron GeForce 6200 graphics card. All tests involving these cards used the ABit IS7V-2 system described above.

High end SFF system
Biostar 330P SFF PC (Nvidia Nforce 4 chipset), with 1 GB Kingston DDR 400 SDRAM (2x512MB) and AMD Athlon 64 4000+ CPU. 7200RPM, 8MB cache hard disk drive. PCI Express graphics cards: GeCube SilenCool ATI X700 Pro graphics card, Albatron PC6800GT graphics card.

High end standard system
Albatron K8SLI (Nvidia Nforce 4 chipset), with 1 GB Kingston DDR 400 SDRAM (2x512MB) and AMD Athlon 64 4000+ CPU. 7200RPM, 8MB cache hard disk drive. PCI Express graphics cards: GeCube SilenCool ATI X700 Pro graphics card, Dual Albatron PC6800GT graphics cards. Thermaltake Purepower 460W PSU. µ


Share this:

blog comments powered by Disqus
Subscribe to INQ newsletters

Sign up for INQbot – a weekly roundup of the best from the INQ

INQ Poll

Heartbleed bug discovered in OpenSSL

Have you reacted to Heartbleed?