1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

F1 2010 i5 gaming rig info and benchmarks

Discussion in 'F1 2010 - The Game' started by Paul Lock, Apr 24, 2011.

  1. A while ago I mentioned I was building a pure gaming rig for Wibblemonster, he wanted best bang for buck, and primarily something that would play F1 @ 2550x1440 around 60FPS on his 27" mac monitor.

    So anyway, no need for hyperthreading i7's for gaming, so I built a rig around the SandyBridge i5 2500K with 2x 6950 GFX cards. I chose these components as due to the new architecture, the stock 3.3ghz i5 can go anywhere up to 5ghz on air, and the 6950's can be flashed to 6970 spec saving around £70 per card. Lots of free performance!

    I built the rig yesterday, and conducted some early benchmarking using a combination of different resolutions, DX09 and DX11, and single and crossfire cards with F1 2010 and figured it may be useful to share the information.

    All tests conducted with maximum graphic detail and 8x AA on a 27" Iiyama monitor.
    I havent used V-sync as this locks the framerate to the monitor refresh rate so will be capped at 59/60fps and im aiming a lot higher than that.

    Sapphire 6950 cards are running at stock 6950 spec.

    1680x1050 - i5 @ 3.7ghz (XMS Overclock)

    DX09 Single Card - av_fps="92.511124" min_fps="74.258224" samples="11395"
    DX11 Single Card - av_fps="60.293808" min_fps="54.774097" samples="7431"
    DX09 CrossFire - av_fps="105.830482" min_fps="86.995621" samples="13080"
    DX11 Crossfire - av_fps="94.168121" min_fps="77.545753" samples="11648"

    1920x1080 - i5 @ 3.7ghz (XMS Overclock)

    DX09 Single Card - av_fps="88.973816" min_fps="72.606293" samples="11064"
    DX11 Single Card - av_fps="56.759933" min_fps="52.037720" samples="7046"
    DX09 CrossFire - av_fps="106.148705" min_fps="84.426926" samples="14064"
    DX11 Crossfire - av_fps="92.746231" min_fps="76.040787" samples="11301"

    1920x1080 - i5 @ 4.3ghz (Manual Overclock)

    DX09 Single Card - av_fps="90.781525" min_fps="74.351570" samples="11208"
    DX11 Single Card - av_fps="57.546658" min_fps="52.548695" samples="7010"
    DX09 CrossFire - av_fps="113.362656" min_fps="96.166153" samples="13715"
    DX11 Crossfire - av_fps="99.679260" min_fps="83.400795" samples="12198"

    To any prospective buyers, as it stands, the tests show you can grab a relatively cheap i5 with a single 6950 and enjoy F1 with all the eye candy turned on 1080p at 57fps, which is only 3 frames away from the maximum most monitors can display, or your eye distinguish.

    Tomorrow I plan to flash the 6950 graphic cards to 6970 spec and see how that impacts the scores.
    This will increase the GPU clock and memory speeds by about 100mhz each, and increase the shaders from 1408 to 1536 - for free!
  2. This will be awesome, thanks for building this mate, top notch!
  3. Interesting to see the huge 30-40 fps increase when using crossfire and dx11, F1 does not use tessallation i think ?
    Not so big in dx9, and overclocking isn't really standing out either, but all results are really good even on a single card.
    Good of you to share this info, even though it makes me want to spend money on upgrades :D
  4. The thing that stands out is the games dependence on raw GFX power rather than CPU power.

    i5 @ 3.3ghz (I just went back to stock to see what happened)
    1920x1080 DX11 Single = 56 FPS
    1920x1080 DX11 CrossF = 82 FPS

    i5 @ 3.7
    1920x1080 DX11 Single = 57 FPS
    1920x1080 DX11 CrossF = 93 FPS

    i5 @ 4.3
    1920x1080 DX11 Single = 58 FPS
    1920x1080 DX11 CrossF = 100 FPS

    Once again proving that these console ported games get the performance boost from GFX rather than CPU, but that makes sense as the game is coded and designed to run on the relatively low power console CPU and would require a massive coding overhaul to make use of higer spec PC chips.

    This is why it is poinless spending over £100 more on a i7 2600K if all you want to do is game. That 100 is better spent on GFX.
    At the beginning I was tempted to go as close to 5ghz as I could with the i5, but looking at the benefit in gaming FPS vs the risk of stability, power consumption, heat and CPU life expectancy, it would seem that going over 4ghz for a 24/7 OC would be wasteful on a pure gaming rig. I may ping it up to 5 and see what happens - I know Viktor is running his at 5ghz, but as the PC is for Wibblemonster, I need it to be rock solid.

    Incidently, Battlefield Bad Company 2 @ 1920x1080 maximum detail etc -

    i5 @ 3.3ghz
    Single = 70 FPS
    Cross = 135 FPS

    i5 @ 4.3ghz
    Single = 74 FPS
    Cross =145 FPS

    So again, almost 100% performance boost with a 2nd GFX card, and bugger all difference on a 1ghz overclock, which involves more voltage and heat thus a harder life and potential shortened life span for your CPU, so still not really looking worth taking the CPU to 5ghz, or even above 4ghz really!

    My 6950 to 6970 bios flash tests will be the next interesting bit of information. Hopefully have a stab at that today.
  5. Yep, it's pretty clear that it's the GPU that's the bottleneck in this game, going from 56 fps to 58 fps with a 1GHz overclock is good evidence of that. As soon as you go crossfire we see a more noticeable increase when overclocking.
    But since it does so well without overclocking the CPU it's hardly any point in overclocking that at all imho.

    Awaiting the results of your gfx flash test :)
  6. The flash is done for both cards. Winflash didnt work in O/S, so I had to follow a workaround and get old skool with DOS prompts and having to physically remove each card to flash individually (meh) -


    Lets see what a (needless) extra £140 would have achieved :D

    Tell you what, im glad F1 is installed on a SATA3 SSD drive, so many restarts...!
  7. Glad to see it worked, think I read somewhere it wasn't possible to do that with all 6950:s.
  8. Nice work Paul, looking forward to seeing what the unlock does to the benchmarks.
  9. Ok, so here are the comparison results at various clock speeds after running the flashed 6970's - seems that in F1 a single 6970 maxes out at around 63 FPS regardless of the CPU speed. I tested the crossfire 6970s at some higher CPU speeds, so I dont have the 6950 stats to compare to.

    I would say flashing a reference board 6950 to 6970 for free is well worthwhile - even more worthwhile if you got 2 of them to crossfire, but really im not sure if paying £70 more for a 6970 out of the box is a good investment for very few FPS in F1 gain. Better gains are achieved in BFBC2.

    i5 @ 3.3ghz
    6950 - 1920x1080 DX11 Single = 56 FPS
    6970 - 1920x1080 DX11 Single = 63 FPS
    6950 - 1920x1080 DX11 CrossF = 82 FPS
    6970 - 1920x1080 DX11 CrossF = 83 FPS

    i5 @ 3.7
    Didnt bother with this one.

    i5 @ 4.0
    6970 - 1920x1080 DX11 Single = 63 FPS
    6970 - 1920x1080 DX11 CrossF= 98 FPS

    i5 @ 4.3
    6950 - 1920x1080 DX11 Single = 58 FPS
    6970 - 1920x1080 DX11 Single = 63 FPS
    6950 - 1920x1080 DX11 CrossF = 100 FPS
    6970 - 1920x1080 DX11 CrossF = 102FPS

    i5 @ 4.5
    6970 - 1920x1080 DX11 CrossF = 103FPS

    i5 @ 5.0
    6970 - 1920x1080 DX11 CrossF = 110FPS

    BFBC2 -

    i5 @ 3.3ghz
    6950 Single = 70 FPS
    6970 Single = 96 FPS
    6950 Cross = 135 FPS
    6970 Cross = 182 FPS

    i5 @ 4.3ghz
    6950 Single = 74 FPS
    6970 Single = 100 FPS
    6950 Cross = 145 FPS
    6970 Cross = 190 FPS

    i5 @ 4.5ghz
    6970 Cross = 192 FPS

    i5 @ 5.0ghz
    6970 Cross = 202 FPS

    Think I am settling on a 4.0ghz overclock @1.25v. Temps idle at 28 to 32 degrees, and fully loaded max out below 60 degrees under Prime95, and low 50's with gaming. With 2x 6970s this thing kicks my i7 6gb DDR3 2x 4870 rig into the gutter.

    I could keep testing till the cows come home. I wanted to do more to stabilize the 5.0ghz O/C but it was unstable even at 1.5v, so I decided not to continue. A nice 4.0ghz should give Wibble headroom to go higher should he ever need it. For now it seems overkill.
    I fully believe despite not being able to test at 2550x1440, it will easily pull 60 FPS.
    If I get chance to grap a MiniDP to DVI connector, I will test Eyefinity and that should give a definitive answer as it would be a higher resolution again.

    Here's a picture of the rig..


    There you go Craig, just add your G27 for some fun!!
  10. Thank you Paul.
  11. That was some interesting numbers you got, that must have taken a while to test, very nicely done :)
    Also interesting to see that BC2 reacts so well to OC and Crossfire, it must be much less demanding on the system and/or simply better coded than F1...
    Anyway, it was kind of you to share this with us and Mr wibbblemonster will get a real good gaming rig to play around with :D
  12. Yes it was time consuming. Best part of 2 days. Hopefully the information might help people make an informed decision when they come to upgrade, as its easy to throw money down the toilet for very little gain thinking that more expensive = better performance.
  13. :good: Porn for geeks !
  14. May as well let you guys know the outcome from the rig. Wibble confirmed the following once he got the rig home and put it on his stupidly high resolution mac monitor -

    i5 2500k @ 4.0ghz - 2x 6970's @ 2550x1440 resolution is pulling in a solid 93 FPS average using the F1 benchmark test.

    My work here is done. :D
  15. I'm thinking about getting either the Ati 6950 or Nvidia's 570. Anyone has an idea which one would be better for games like F1 2010?
    my current rig:
    i5 760 @ 3.3ghz OC'd
    4gb ddr3
    p7p55 lx board
    nvidia 260 GTX
    650W Antec PSU

    edit: I'm getting horrible FPS atm. like 15- 20 max on low settings, still very laggy tho.
  16. Without being rude Paul, what sort of sum would it require to knock that system together? You always read people saying you need to spend £1500 and then someone saying you really only need to spend £500 and still play the game at it's max, but you've done the research so your opinion has some weight behind it!
  17. Yesterday I installed i5 2500k with new mobo and 4GB DDR3. Previously I had Core2Duo E8200 (OC to 3.2GHz) with GeForce GTX 460 and avg. 38 fps in game with Medium detail. Now with High details it's avg. 98 fps (1920x1080, 8xMSAA). But when I set Ultra details fps drop to 40 fps.
    When I using FRAPS to record video fps drop to 60 fps (it's no problem for me because normally I use Vsync - with Vsync on and FRAPS fps is still 60).
    I'm thinking about SLI (2x GTX 460), maybe I will need it for F1 2011 ?

    [EDIT] For CPU, motherboard (not cheapest one) and RAM I paid in Poland about £370. This GeForce costs another £150.

    My G27 work now much better. Rev lights on logitech steering wheel and lights on steering wheel in the game now blinks exactly in the same time, with no lag. But I change system and drivers from 32 to 64-bit. Maybe this is it?

    Sorry for my English :rolleyes:
  18. I'm running an i5@4.6ghz on Asus PB67 with 4GB ram and a GTX580. The game is smooth as silk at max settings; displayed on a BenQ 24" monitor, 1920x1200 res.

    Does F1 2010 gain anything from a SLI/Crossfire configuration? I know many games don't. I'm an avid flight simmer and twin cards do nothing for FSX, probably the most demanding of PC games.

  19. Not rude at all. It certainly wasnt £1500, and also bear in mind it did include some decadent parts, like a Sata 3 SSD which weighed in around £250 alone, a conventional Sata 3 drive (which is still fast!) would have been more like £50. Plus the case was close to £100, and Win 7 Retail was around the £100 mark too.

    So really I would say to get the same gaming performance (without the funky HD, case and OS) you could pull a rig together like the crossfire'd i5 system I built here for around the £1000 mark easily. If you wanted to get the same thing 'off the peg' ready built, they would happily take £1500 from you for less spec. I looked around to see what ready built 'gaming' PCs would be similar to the one I built for Wibble, and it was close to £2000, and you know for a fact that wouldnt be with the sort of quality parts you would source to self build.

    That seems strange to get such a performance hit putting a few settings on ultra! But yes you would certainly benefit from a second graphics card. I doubt you will 'need' it for F1 2011, surely CM have enough on their plate making a game that isnt riddled with game breaking bugs and all the features missing from 2010. Not heard many complaints about the graphics not being up to scratch, so hopefully they will spend their time more wisely on this release!

    Do you have your Fraps set to record at 60fps? If you do, its probably capping your game frame rate. Try turning it up to record at 70, and then 75 until things look smoother in game. I thought increasing the recording FPS would hit game performance, but its kinda the otherway round! Well it worked for me anyway!

    Mate look at the other pages on this thread where I was benchmarking F1 in single and crossfire, lots of info there, but in a nutshell -

    F1 2010 everything maxxxed @ 1920x1080

    i5 @ 3.3ghz (I just went back to stock to see what happened)
    1920x1080 DX11 Single = 56 FPS
    1920x1080 DX11 CrossF = 82 FPS

    i5 @ 3.7
    1920x1080 DX11 Single = 57 FPS
    1920x1080 DX11 CrossF = 93 FPS

    i5 @ 4.3
    1920x1080 DX11 Single = 58 FPS
    1920x1080 DX11 CrossF = 100 FPS

    Battlefield Bad Company 2 @ 1920x1080 maximum detail etc -

    i5 @ 3.3ghz
    Single = 70 FPS
    Cross = 135 FPS

    i5 @ 4.3ghz
    Single = 74 FPS
    Cross =145 FPS

    Basically, overclocking your precious CPU achieves very little - adding an extra graphics card achieves a lot!

    This is because games like F1 are ported from consoles, as such arent optimised to run on anything better than a console processor. Yes you will get more frames with a better processor, it doesnt get capped, but if you look at recent benchmarks on sites like TomsHardware, Anandtech, Techpowerup etc, you will see there is a couple frames difference between the current i5 2500K and the £100 more or so i7 2600K. A lot of the tech you get with an i7 (like hyperthreading and bigger L2) just doesnt make financial sense if you are building a gaming rig.

    Phew - right i think thats everyone answered!!