Battlefield 3 Benchmarked, GPU and CPU Performance Tested

By Julio Franco ยท 85 replies
Oct 28, 2011
Post New Reply
  1. slh28

    slh28 TechSpot Paladin Posts: 1,706   +172

    Nice review.
    I think it's GPU upgrade time for me, the ultra screenshots look awesome
  2. Any ideas why ATI Radeon HD 4890 sucks too bad there? I mean, it is a great card, runs Crysis 2, Lost Planet 2, CoD: BlackOps and ... pretty smooth (something like Radeon HD 6870 in most benchmarks).

    Can anybody else please confirm these low frame rates with their Radeon HD 4890?
  3. SkitzoPhr3nia

    SkitzoPhr3nia TS Rookie

    I don't suppose they had time to play a few hours with each setup while running these test, but I keep having my display drivers crash while i'm playing. I am running 2 6970's in crossfire with 11.10 drivers with and without CAP 3. Anyone else having this problem? Sometimes I can play for hours no problem, and sometimes it only takes minutes. The only fix I have found is not running crossfire, then it works perfect.
  4. Nice review. Love this game. I'm running it @ 1920x1080 with a 2500K @ 4.5GHz, Two overclocked HD 6950 2GB in Crossfire, and 8GB G.Skill RipjawsX 1600MHz @ 1866MHz, playing on ULTRA.

    Now I gotta get my friend to buy a 6870 or 560 Ti so he can game too. His GTS 250 won't cut it. Get this game... NOW!
  5. Nice. I know it's usually not the norm of this website to do 5760x1080 resolutions, but would have been very interesting to see. Just looking at 2560x1600 UltraQ Settings I can only guess at 5760x1080 judging from the data here and seeing past extreme games at this resolution.

    I can't wait to see how BF3 performs when Kepler and Tahiti are released.
  6. For the 4890's poor showing, it's a combination of the level being testing (why that aircraft one, it's not exactly typical for MP play now is it) and the fact that the reviewing left on High Ambient Occlusion to be fair to the other cards.

    With AO set to off, Motion Blur off (for taste really), no AA and AF x16 you should get about 30-45 FPS on all High with a 4890 DX10 1GB card (plus a good CPU, i.e. 4 core)
  7. Relic

    Relic TechSpot Chancellor Posts: 1,379   +16

    Good review Steve. So far performance wise I'm extremely pleased and running significantly better than I assumed. Getting slightly higher averages than reported here on my 6850 same as the beta but that's nothing to complain about :) . Going to mess around with some OC's to see how much more I can squeeze out but I'm pretty satisfied with my lows being in the 40's on my system during intense multiplayer fights.
  8. Nobody should rely on these benchmarks for bf3. Battlefield games are very tough on video cards and this benchmark comes nowhere near the maximum loads your cards will experience in a 64 player game with lots of action.

    Go to for real benchmarks of bf3. use google to translate the site.

    To the person who did this piece, don't take this the wrong way but sp benchmarking doesn't not produce useful results in this game. 10 minutes of real gameplay will. use that average fps and avoid spending much time on spawn screen.
  9. Sarcasm

    Sarcasm TS Guru Posts: 367   +46

    Umm, based on the findings of this article it is actually pretty accurate. Even on the full 64 player maps when all things go to hell, my FPS doesn't dip below 40 with highs up to 60+. This is running on a stock GTX 580 (no OC) @1920x1080 resolution on ULTRA.

    CPU is irrelevant seeing as any modern decent CPU will run this game.

    And to those looking for the best overall settings, I think setting the Post anti-aliasing to HIGH or MED while deferred anti-aliasing to 2xMSAA. It's the best balance of visual and performance for me.
  10. Steve

    Steve TechSpot Editor Posts: 2,869   +2,039

    To the troll that made this post, thanks for the advice.
  11. red1776

    red1776 Omnipotent Ruler of the Universe Posts: 5,224   +164

    Actually its spot on...and you don't have to translate Steve!....mostly :p:wave:
  12. Whoa Whoa Whoa. Something is very wrong with your 4890 benches! You have the 4890 listed as 20 FPS avg at high settings at 1920X1200.

    This has to be very wrong, I'm running a 1GB 4890, at 1980X1020. It does have a factory overclock to 950 mhz, but that shouldn't make a huge difference as I believe default 4890 clock was 850. Anyways, I'm playing the campaign at all ULTRA settings, mind you, NOT high, with the exception of terrain quality locked to medium (as with all DX 10 cards), and SSAO instead of HBAO, no MSAA, FXAA set to HIGH, motion blur on, 16X AF (settings because I heard HBAO and MSAA are performance killers).

    And I'm getting, in the early parts of the campaign, oh I would say a solid 30 FPS. Very playable. Sometimes in less demanding areas I even sometimes see the FPS shoot up to 45-50, though it often dips to the mid-upper 20's as well in firefights. I havent run a fraps average, but my guess is it would be averaging perhaps 35 FPS, and no lower than 30. That's fully playable for me, I find 30 FPS to be fine (and I console game a lot too where 30 is the norm).

    And the rest of my rig is weak too, it's a Q6600@2.8, 4GB of DDR2.

    So yeah, 20 FPS on lower settings than I'm using, something is wrong. By your chart 4890 is unplayable at high settings+1080P, yet I'm playing it very nicely at mostly ultra 1080p, forget high!

    It could be something to do with the level, as I said I'm early and have not reached the "Go hunting" level yet. Still, I'm doubting it's so much more demanding than the rest of the game.

    Another clue is that the 6850 should have similar grunt to my 4890. My 4890 is 800 shaders @950, a 6850 is 960 shaders @775. That's almost exactly equal in power when I did the math (I know shaders arent everything but it should give us an idea). So for 6850 to be averaging 40 and 4890 averaging 20 again is a red flag. And as I showed, I would guess my 4890 is averaging maybe 35, which would put it close to 6850 where it belongs.

    So yeah, imo your 4890 benchmarks are wrong, I'm playing at 1080P mostly ULTRA at 30-35 FPS with a 4890.

    I did notice you have HBAO on as well where I'm using SSAO (then again I have multiple settings at ultra versus your high!), but again I'm not sure that would kill performance that much.

    I'll revisit this comments section in a day or two, double check everything, get a fraps avg to be exact, maybe see if HBAO makes the difference, but as of now must disagree with your 4890 bench.
  13. Hmm, I hadnt read the comments before my last post, and I see somebody else had brought up the 4890 issue before I wrote my long post above about it.

    And he's right. The 4890 may not have the DX 11 bells and whistles but is flat fast as hell due to it's extremely high clocks and I guess, lack of DX 11 overhead. Even today it runs most games on "high" (or above) and is one of the best value cards I've ever owned.

    But yeah, other guest, my 4890 runs the game pretty well, as it's all laid out in my post above, and as us 4890 owners would expect given it's good performance in other games.
  14. Steve

    Steve TechSpot Editor Posts: 2,869   +2,039

    Once AMD release a certified driver I will retest but I agree our results for the Radeon HD 4890 seem too low. I was unable to work out why the card was so slow.
  15. Will my 1 gb version of hd 5870 give same performance and why my pci express 2.0 x16 running only at x8?
  16. OK, same guest as above, I already ran some very quick benches. Again my settings and PC:


    Q6600@2.8 ghz, 4GB DDR2, HD 4890 1GB @950.

    BF3 campaign mission was operation swordbreaker, the checkpoint where you start underground after the QTE-fistfight, make your way uptop to a firefight, then get on the bridge and grab the LMG for more firefight. So, a good mix with lots of firefight should be pretty stressful. Ran two main benchmarks. Using FRAPS. Settings, everything Ultra, no MSAA, high FXAA, 16AF, Motion blur on, SSAO. Ran two benches one with SSAO one with HBAO. Oh and 1080P resolution of course.

    First bench (SSAO)

    5825 frames, min 24, max 72, avg 37.629

    Second bench (HBAO)

    3442 frames, min 22, max 53, avg 31.063

    So there you have it. HBAO did seem to make a big hit, but I'm a little suspicious just because this type of benchmark is very random, and it's possible the 2nd run was just a lot more stressful in terms of "firefighting". But yeah overall HBAO took a hit, but it's still WELL faster than your bench. Even allowing say an extra 10% for my 4890 overclock too. And remember most of my settings are higher. I'll run some more and post them in the next couple days here, short on time now. The only other variable is the level, and I'm not at "go hunting" yet.

    But again if "go hunting" was that much more stressful and 6850 ran it at 40 FPS, then a 6850 should be running the parts I just benchmarked at like 80 FPS, which probably isn't the case.

    Like I said I'll be back for more benchmarks cause this is fun. Like to run one at your "high" settings too for direct comparison.
  17. nice review there. its very good to see bulldozer scaling benchmarks, but i noticed that athlon x3 was left out,i am hoping that in the future it will be included again as it would be interesting to see if games scale beyond 3 cores, especially in games that are not as GPU demanding as bf3
  18. Guest: "will my 1 gb version of hd 5870 give same performance and why my pci express 2.0 x16 running only at x8"

    Maybe you have too many ports/slots/ enabled on your motherboard (ie. USB3.0 and/or "too many" expansion cards that are eating up your PCIe lanes. Check your bios, read the motherboard manual, or google, to see if this is the case.

    I'm using two video cards on my ASUS P8P67 EVO, and if i want to use USB 3.0, then i cannot use one or both of my PCIe x1 slots, and i can only use my 3rd PCIe slot @ x1 instead of x4, if i remember correctly without looking. Also if i want that slot to remain at x4, then i cannot use 2 of my standard USB slots on the back of my case.
  19. Frames, Time (ms), Min, Max, Avg
    3566, 60000, 48, 68, 59.433

    @High preset 1680x1050.
    Core i5-760 (@stock)
    GTX460 Hawk (875/1750/1900)

    Did the same test as this site.
  20. ---agissi---

    ---agissi--- TechSpot Paladin Posts: 1,978   +15

    How about some screenshots without MSAA per the conclusion, so we can see what it looks like & compare.
  21. Excellent article.

    Will come in handy when I'm ready to build my new rig next month.
  22. Really?... no 2500K tested? no Core 2 Duo/Quad? So many AMD's tested i'd think.......
  23. red1776

    red1776 Omnipotent Ruler of the Universe Posts: 5,224   +164

    ummm...I forgot my glasses in the car...could you make out the CPU second down from the top for me?
    so much like a troll post ...i'd think....

    Just kidding!...I don't wear glasses

  24. i mean in CPU utilization. ;)
  25. you did all that for nothing. did you find my C2D/Q? no? aw too bad. must be nice to have all your free time.

Similar Topics

Add your comment to this article

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...