Battlefield 2 GPU Performance Analysis

by Derek Wilson on 7/7/2005 3:17 AM EST


Back to Article

  • sethborg - Thursday, December 22, 2005 - link

    What monitors support these high resolutions, the article doesn't say which one they used! Reply
  • TokyoBrit - Tuesday, August 09, 2005 - link

    I wouldn't of minded seeing the 6600GT, 6800GT and 6800U in SLI configuration in the article, as those are a more likely configuration than the 7800GTX SLI that did get included.

    Anyone care to comment on the expected performance improvements of 6600GT SLI compared to a single 6800GT, 6800GT SLI compared to a single 6800U, and 6800U SLI compared to a single 7800GTX?
  • CrystalBay - Sunday, July 31, 2005 - link

    Thanks Derek, ! Reply
  • ShadowVlican - Sunday, July 17, 2005 - link

    much better with the update, thanks derek Reply
  • JNo - Monday, July 11, 2005 - link

    Guys, I think most of you are missing the point - this is not meant to be a *comprehensive* test of BF2 (with CPU, *all* midrange cards and picture detail analysis etc) but a general look at how a game that will become part of anandtech's new benchmark suite, performs in general across a range of cards. It does this very well and gave me an impression of how a variety of core cards scale with the game.

    Again, the article is called "BF2 **GPU** Performance Analysis" so stop whinging about lack of CPU scaling tests and not knowing how the game performs on true midrange systems - that is not the point of the article!! Read a general BF2 performance article elsewhere for that...
  • Powermoloch - Sunday, July 10, 2005 - link

    Replying #61

    Just recently I had received my X850XT for about $ 350. I'm still amazed about the benchamrks and hopefully will getting this copy of the game. It's performance is dazzling.

  • Powersupply - Sunday, July 10, 2005 - link

    Almost forgot...I whonder how much DICE/EA payed Anandtech for this one... Reply
  • Powersupply - Sunday, July 10, 2005 - link

    I rarely write any comments but i feel obliged to do so this time around because this article stinks. Quoting what #45 said:

    " The more I think about what information this article *should have* presented, I come to realize that it is just fluff. The basic conclusion is that higher-priced cards get better FPS. No duh?"

    That's the simple and ugly truth.

    Sadly I feel that an increasing percentage of articles are just fluff here on Anandtech.
  • beorntheold - Sunday, July 10, 2005 - link

    the last game i played from EA was NFSU2. just as in NFSU the time was not constant... in a racing game. think about it.

    BF2? well, the demo won't start in fullscreen mode no matter what i do (and i did spend a while working on the problem). oh, and that patch fiasco... 8|

    makes you wonder--how do Blizzard do it? :)
  • yanman - Sunday, July 10, 2005 - link

    BF2 that is Reply
  • yanman - Sunday, July 10, 2005 - link

    When I tried out BF 1.0 with 77.72 drivers and 2 Galaxy 6600GTs in SLI i had major issues in 1920x1200. I had to revert to 1600x1200 or another square res to fix this. Any idea why I'd have these problems? (basically the image was skewing everywhere at half-way point so obviously a SLI issue) Reply
  • xtknight - Saturday, July 09, 2005 - link

    And I almost forgot to mention...great little review/test here. Reply
  • xtknight - Saturday, July 09, 2005 - link

    I would have liked to see some 6800 Non-Ultra benchmarks...the AGP version of this card is a much hotter deal then the 6600GT...I saw one for $156 the other day. Reply
  • DanDaMan315 - Saturday, July 09, 2005 - link

    Honestly this is the most incomplete GPU preformance test review I have ever seen out of AT. I am disappointed. Reply
  • bobsmith1492 - Friday, July 08, 2005 - link

    What I get out of this is that ATI cards struggle at higher resolutions, but thrive with AA; Nvidia is the opposite.

    It makes sense for ATI to go for the AA since most people have smaller monitors, although higher resolution is really better all around.
  • dornick - Friday, July 08, 2005 - link

    Thumbs up for updating with x800xl and 6800gt Reply
  • Pastuch - Friday, July 08, 2005 - link

    Canadian Graphics Card Pricing:

    ATI x800xl: $329 AGP ( $379 PCIE (NCIX)

    6800gt: $379 at NCiX if you can catch a sale.
    $400 PCIE (NCIX) Basically if you want a 6800gt you have to spend $400.

    X850XT: $499 (Bestbuy Sale - It just started today and until now this card cost $650)

    Asus 7800GTX: $669 (NCIX sale)

  • ss284 - Friday, July 08, 2005 - link

    BTW, x850xt's are hitting closer to the 375-400 price mark now. The 7800 is still 550-600. Reply
  • yacoub - Friday, July 08, 2005 - link

    " has been a constant struggle to ... properly render resolutions above 1600x1200 on analog monitors."

    Everyone should be running LCDs for high resolutions these days anyway. Much easier on the eyes and weigh maybe a tenth as much as some giant friggin' CRT.

    Thanks for adding the X800XL and 6800GT to the comparisons, they were much desired. Now if only the tests were done on more common systems so we could actually expect to get the same FPS results as your tests (FX-55).
  • tgjp - Friday, July 08, 2005 - link

    Just like to say, I've bought a system with 7800GTX in SLi and I am very excited about using this machine, and perhaps playing battlefield2. Reply
  • Pastuch - Friday, July 08, 2005 - link

    Normally I really enjoy Anandtech articles but this one was sorely lacking. I agree with the other twelve posters. How could you leave the X800XL and 6800GT out of this comparison? Who is going to buy an x850xt when they can purchase a 7800gtx for almost the same money? No One. So why include it in the test? I'm assuming it was because that is the fastest ATI solution available but I still dont think it was a good idea. I thought the purpose of this article was to help people decide which card to purchase. Why did you choose cards that have almost zero interest to your typical Anand reader?

    I also dont like the naming conventions used for each test. Why not say 2xAF instead of "med" for texture filtering settings. Its just confusing.

    To your credit though I did enjoy seeing the two 7800gtxs in SLI do 2000+ resolution above 60fps. That was something I thought was impossible. If you own a Dell 2405FPW or a 2005FPW then SLI is what you need. Native res on the 2405 is 1900x1200 and apparently a couple 7800s is the key to gaming euphoria.

    If you want to run BF2 in widescreen all you have to do is change the shortcut command line to:

    "C:\Program Files\EA GAMES\Battlefield 2 Demo\BF2.exe" +menu 1 +widescreen 1 +szx 1680 +szy 1050

    This will force the resolution you desire, however the POV can not be changed. This means that your verticle vision will be cropped. Also, you wont get the usual advantage of more peripheral vision due to the wide aspect. If you want to read more about widescreen gaming in BF2 read

  • legoman666 - Friday, July 08, 2005 - link

    Do you guys think that you could run a few benchmarks on Windows XP Pro x64 Edition? I downloaded and played the demo and on my comp: (2.4ghz A64, x800pro, 1gb ram, at 1024x768, with a 3mb dsl connection) the game lagged so badly and had the most horrible fps ever. Almost as bad as when i tried to play doom3 @ 1600x1200 at 4xaa 16xAF with a 9700pro.

    or if anyone else has had experience playing the game on x64 windows and would like to share it, thatd be good too.

  • ElFenix - Friday, July 08, 2005 - link

    you're surprised that you can still see jaggies at high resolutions? resolution doesn't keep the jaggies from happening, it only makes them smaller. i really don't know why you'd be surprised at that. and, of course, if you're running an 24" LCD that can do 1920x1440, your DPI is worse than a 19" CRT running 1600x1200(~95 for the LCD and ~110 for the CRT). so the jaggy is ever MORE noticeable than it was on the CRT, and AA is even MORE important than at the lower resolution, but higher DPI, CRT. and jaggies are plenty noticeable at 110 DPI. Reply
  • Powered by AMD - Thursday, July 07, 2005 - link

    Is this game CPU Bound with my Thoroughbred B at 2300 Mhz, 400 Mhz FSB? and 9800 Pro.. Reply
  • bobsmith1492 - Thursday, July 07, 2005 - link

    For crying out loud people.... look at ANY other video card review. Your card won't change dramatically in relationship to the other cards reviewed. If an XL performs just under an 850XT in ____, then it will be the same here. Reply
  • eBauer - Thursday, July 07, 2005 - link

    Add me to the list who wants a CPU performance analysis. Reply
  • bob661 - Thursday, July 07, 2005 - link

    Jesus guys...LOL!!! Ease up on the man! Reply
  • buck - Thursday, July 07, 2005 - link

    It would be nice to see this review updated or done over again the right way. Reply
  • buck - Thursday, July 07, 2005 - link

    I agree that leaving out the x800 series (specifically the x800xl) was a bad move. I am very dissapointed in this review. What the hell was he thinking? Reply
  • bob661 - Thursday, July 07, 2005 - link

    Gotcha. Thanks.
  • crazyeddie - Thursday, July 07, 2005 - link

    I would like to add that I don't understand omitting X800XL from comparisons. It's ATI's most popular mid-range video solution. Reply
  • yacoub - Thursday, July 07, 2005 - link

    Add me to the list of people looking forward to a more complete article with CPU scaling and a 9800 Pro and an 800XL included in the tests. Reply
  • formulav8 - Thursday, July 07, 2005 - link

    I just bought a Regular 6800 Vanilla AGP and thought/hoping that should/would be included.

    I am happy that mine unlocks the pipes and shader just fine! (Well, so far anyways)

    I ran 3DMark03 and got 8086 with stock pipes and got 8778 with 16x6x pipes. Not to bad with a fairly slow cpu. Thats a 8% increase with no oc or anything, and with a fairly slow cpu that is bottlenecking most likely.

    I will try ocing it next week when I get a cpu/mobo that will push it alittle more.

    I am running a Athlon XP with 128KB L1 and 128KB L2 Cache at 2ghz. And yes, that IS the right L2 Cache numbers on this cpu :)

    Are there any instructions that I can find to run my own benchmark numbers for BattleField2?? Anyone know???

  • pcfountain - Thursday, July 07, 2005 - link

    And while I'm on the subject of price/performance, how about a graph of street price vs. FPS for each card, so we can see the "value" of each card visually? The more I think about what information this article *should have* presented, I come to realize that it is just fluff. The basic conclusion is that higher-priced cards get better FPS. No duh? Reply
  • pcfountain - Thursday, July 07, 2005 - link

    I agree with many others, leaving the X800XL out of this roundup (and in fact EVERY X800-series card) is a HUGE oversight. The X800XL a very popular card and probably the best price/performance ratio for the high-end right now. IMO the article is useless without this comparison. Reply
  • dornick - Thursday, July 07, 2005 - link

    Only the X700 and the X850XT tested in the high-end!? What about the half-dozen other cards in between? Reply
  • Killrose - Thursday, July 07, 2005 - link

    Can we get off this FX-55 addiction and use a lower RANGE of CPU's to see how the game scales with a lower class CPU?

    A 6200 or X300 GPU being crappy is really a no-brainer. And pairing it with a world class FX-55 CPU is senceless.
  • L3p3rM355i4h - Thursday, July 07, 2005 - link

    9600 and 8500 are 4 pipes.

    My winnie @ 2.5 1GB of RAM and 9800@ 480/375 runs 10x7 everything on high very well. No lag what so ever. Granted, I have a sign. advantage over a x700pro and a 6600gt.
  • Questar - Thursday, July 07, 2005 - link

    My bad, I thought the 9800 was a four pipe card. Reply
  • drizek - Thursday, July 07, 2005 - link

    i have a 6600gt, gig of ram @360, amd 3200+ and i can play it at 1600x1200 with 2xAA and medium everything else(around 40fps).

    from all the guides ive seen, there is a very small difference in graphics by moving to High settings from medium, but you see a big performance hit. You should tested the low/midrange cards using medium, because i doubt many people are going to crank it up to high and then play at 1024x768.

    here are some screenshots comparing low/medium/high settings
  • saiku - Thursday, July 07, 2005 - link

    The choice of cards to benchmark always baffles me.
    Where are the mainstream cards such as 9800 Pro? What percentage of Anandtech's audience uses 7800 GTX or an SLI setup? Why not focus on the "mainstream" cards such as the 9800s, the 6800GT and non spend tons of stats on $1000 video card setups.

    I love reading anandtech stuff but their choice of cards for benchmarks drives me nuts.
  • JustAnAverageGuy - Thursday, July 07, 2005 - link

    Would have liked to see the 9800 Pro and 800XL included. Reply
  • DerekWilson - Thursday, July 07, 2005 - link

    Sorry, I was mixing up my game engines there -- you are correct and the article has been fixed to reflect the use of the proper engine.

    Derek Wilson
  • Therms - Thursday, July 07, 2005 - link

    The Battlefield 1942 engine had nothing to do with the Unreal Engine.

    "When you're looking at screenshots and movies from Battlefield 1942 you cant stop noticing that it is damn beautiful; could you tell us something more about the technique behind the game? We heard that you developed your own engine, could you please tell us a little bit more about it?

    The engine is named Refractor2 and is completely home made, just as it's predecessor, Refractor. The reason why we made our own engine, and didn't use any of the well known engines as say the Quake3 engine or the Unreal Tournament engine, is because of the specific demands that Battlefield 1942 has. "

  • CrystalBay - Thursday, July 07, 2005 - link

    Nice article Derek,

    Will AT be making the AT BF2 benchmark downloadable for members ? That would be nice , then members could bench thier cards and share the results. Myself I can figure out how well my GT scales compared to Ultras.
  • dev0lution - Thursday, July 07, 2005 - link

    It doesn't make much sense to include an X700 Pro over an X800XL, as I'm sure the latter is one of the more popular ATI cards. Reply
  • coomar - Thursday, July 07, 2005 - link

    x700 pro is pretty much the equivalent of a 9800pro, i'm a little surprised as to a lack of the x800xl or 6800gt Reply
  • ShadowVlican - Thursday, July 07, 2005 - link

    a 2nd edition of this article would be appreciated (to include CPU scaling & older gfx cards) Reply
  • Yelapaboy - Thursday, July 07, 2005 - link

    One thing I have noticed with my 3.2 @ 3.6 P4 with X800 XT PE main rig is that although I notice a pretty nice difference at very high res with older games and sims I don't notice that much with newer shooters that use shaders such as HL2, Doom 3, and Far Cry. For example I was almost as happy playing Far Cry at 1024x768 with 2X temporal FSAA on my old 9700Pro as 1600x1200 4X with my X800 XT PE, I can't tell much difference in HL2, Butcher Bay or Doom 3 when I go above 1280x960 on my 22" Viewsonic P225 and pretty much run them at that res. I figure my year and a half old main rig with it's year old video card easily has another year left of keeping me wildly happy although I am chomping at the bit for a dual core AMD CPU and the 7800 is certainly awesome. I feel that of the games I play only IL2 and FS2004 would give me a definitely better gaming experience with more GPU and CPU but as it is they run quite well. At any rate great article but I would have liked to see certain other cards tested. Reply
  • Jep4444 - Thursday, July 07, 2005 - link

    the demo doesn't seem to want to install on my computer so i can't see how it'll run on my rig Reply
  • jkostans - Thursday, July 07, 2005 - link

    Yeah the 9800 pro runs fine at 1024x768 with graphics high on my barton 3000+ (2.2ghz). Oh yeah and a gig of pc2700 ram. I would compare the framerate to CS:S but a tad slower in some areas. Oh and for people wanting to find out how their computer performs, there is a demo available. Yeah it's 500mb but it's good. Reply
  • Xenoterranos - Thursday, July 07, 2005 - link

    Where's the Voodo 3 2000/K62 test setup? Come on guys, cater to the poor b@stards out here... :p Reply
  • WileCoyote - Thursday, July 07, 2005 - link

    For you 9800 Pro owners this game runs pretty well on a 3ghz p4 with 2gb of ram.

    I run BF2 at 1024x768, medium, 4xAA and it runs very smooth. I'm sure you could even bump up some of the medium settings to high.
  • Aikouka - Thursday, July 07, 2005 - link


    I mean bias as in a statistical bias, not a personal preference.
  • Jep4444 - Thursday, July 07, 2005 - link

    how bout we seem some tests from either an X600Pro/XT or a 9600Pro/XT, these cards are extremely common and personally i'd like to know how my card fairs in BF2, i don't want to have to wait until the X550 comes out(basically a 9600Pro with 500mhz RAM instead of 600mhz) to find out how my card should perform Reply
  • yacoub - Thursday, July 07, 2005 - link

    Oh I see now. FX-55 system. Heh. Reply
  • yacoub - Thursday, July 07, 2005 - link

    Sweet so if I get an X800XL card for around $300 (the most anyone should have to pay for a GPU to play the latest games smoothly), I can just about handle 1280x1024 with 4xAA/High quality at around 50fps. That's not toooooooo bad.

    Oh wait, I wonder what the rest of the test system's specs are (I didn't see them on any of the pages). If it's like an FX-57 or something then that's not exactly promising for the majority of us running 3000+-3700+ A64 systems who will clearly experience a bit lower performance. =/
  • Avalon - Thursday, July 07, 2005 - link


    An X700 does not have twice the pipes of a 9800 pro, it has the same exact ammount. It also is not clocked much higher. The x700pro is clocked at 425/860 compared to a 9800pro at 380/680. As you can see, core speeds are fairly close, and memory speeds...well, the 9800pro is 256bit while the x700pro is 128bit, so despite the clockspeed advantage, the x700 should actually have less memory bandwidth in the end. This puts the cards roughly equal.
  • OrSin - Thursday, July 07, 2005 - link

    the x700pro is clocked a little higher but has the same number of pipes and only 128 bus where the 9800pro has 256 bus. In truth the 9800pro is faster then the x700 and might be about the same as the 700pro or just little faster. Reply
  • coldpower27 - Thursday, July 07, 2005 - link

    Questar, X700 is the RV410 chip which is 8X1/6 vs the the 9800 Pro R350 chip with 8X1/4. Though the X700 Pro has the advanatge of being clcoked a little higher.

    The X700 Pro should be close to the 6600 GT. And certainly does not have twice the pixel pipes or vertex pipes of 9800 Pro.
  • Questar - Thursday, July 07, 2005 - link

    Who told you that? An x700 has twice the pipes of a 9800pro, and is clocked much higher.
  • bob661 - Thursday, July 07, 2005 - link

    Oh and I guess I'll be turning up the settings on my 6600GT. Reply
  • bob661 - Thursday, July 07, 2005 - link

    I don't know about bias, but it would defintely not be scientific given the purpose of the article.
  • Aikouka - Thursday, July 07, 2005 - link

    The point of these articles is to see how well the GPU scales as you change models. The article's focus is not in overall system performance in different price ranges, but how well different video cards perform against eachother and to do that, you kind of have to use the same hardware (with the exception of the card) or you really help to create a thing we like to call "bias." Reply
  • blckgrffn - Thursday, July 07, 2005 - link

    I think that they could do a whole article with this engine and CPU performance + ram usage, let's hope if they do they don't neglect the AXP like they did for the HL2 benchmarks...

  • jm20 - Thursday, July 07, 2005 - link

    You have just benchmarked the highend, why not now cater to the rest (95%+) of the market. How many people pair a 6200TC with a FX-55? Do some realistic tests with a few different setups.

    Budget: AthlonXP 2000+, 512Mb ram, 8500/9000 pro
    LowMid: AthlonXP-M @ 2.2-2.6Ghz, 512Mb ram, 9800 pro
    HighMid: Athlon64 3200-3500+, 1Gb ram, 6800GT
  • Hacp - Thursday, July 07, 2005 - link

    "(disclaimer: this is not actually possible)"

  • Backslider - Thursday, July 07, 2005 - link

    My 9600Pro does the job fine with med/high settings. When using 1GB of memory.. Reply
  • jkostans - Thursday, July 07, 2005 - link

    The X700 pro is essentially a 9800 pro but pci-e. So there's pretty much no difference in performance. Reply
  • geekfool - Thursday, July 07, 2005 - link

    Where does the X800 Pure/XL/Pro fit in? And I'm sure that anyone with an Athlon FX 57 doesn't use GF 6200/ X300. Reply
  • Tiamat - Thursday, July 07, 2005 - link

    #6 that was what I was gonna ask :/

    also, 9800pro would be cool...
  • reactor - Thursday, July 07, 2005 - link

    frallen, 6800 ultras in sli perform about the same as a single 7800, so take that as your benchmark. so either way it would be the same performance, but if you get the 7800 then you have the option to go sli later and get even better performance.

    nice article, hope to see it updated when ati releases their cards.
  • 100proof - Thursday, July 07, 2005 - link

    No X800XL benchmarks? =/ Reply
  • Frallan - Thursday, July 07, 2005 - link

    Since I went SLI and 1*6800 GT in the begining I would like to see how a 6800gt SLI or a 6800 U SLI setup does. The 6800U does 42.5 in 16*12 which is just about playable so how will a SLI setup do? (the question for me will be to either sell my 6800gt doing Ultra+ speeds and get a 7800 or get another 6800 gt). Reply
  • RobFDB - Thursday, July 07, 2005 - link

    Think it's time for me to up the resolution. I've been playing on 4xAA and high settings @ 1024 on my rig. My 3500+ and x850 XT PE should be able to handle higher. Does anyone know how to run this time demo to benchmark their own system? Reply
  • Cavedweller - Thursday, July 07, 2005 - link

    No test of Radeon 9700/9800? 8( Reply
  • DigitalDivine - Thursday, July 07, 2005 - link

    i'm playing bf2 fine on my 1.6ghz duron and 512 mb of ram, with a 9200 non pro at low settings 800x600.

    at low settings, that game is still very damn playable. kudos to them.

    btw, for those wanting their 9200s to run with bf2, set your agp speed to 4x. for some odd reason that works. *game freezing after the bink videos.

    now back to playing with my also antiquated 2.8 p4 with 9800pro *sigh, only at medium settings.
  • R3MF - Thursday, July 07, 2005 - link

    i could play the demo ok on my rig at 1164x896 (or some such random number), which is the highest resolution it permitted.

    however i understand the retail game still won't ship with support for 1680x1050, so i won't be buying it.

Log in

Don't have an account? Sign up now