The Test

Test Setup
CPU AMD Phenom @ 2.6GHz
Motherboard MSI K9A2 Platinum
Video Cards ATI Radeon HD 3870 X2
ATI Radeon HD 3870
Video Drivers Catalyst 8.1 (Modified for CrossFireX Support)
Hard Drive Seagate 7200.9 300GB 8MB 7200RPM
RAM 2x1GB Corsair XMS2 DDR2-800 4-4-4-12
Operating System Windows Vista Ultimate 64-bit


We ran the following games in their corresponding quality settings:

Game Resolution AA AF Detail Settings
Half Life 2: Episode Two 2560 x 1600 8X 16X Highest in-game
Unreal Tournament 3 2560 x 1600 0X 16X Highest in-game
Bioshock 2560 x 1600 0X 1X Highest in-game
Call of Duty 4 2560 x 1600 4X 16X Highest in-game
Crysis 1920 x 1200 0X 1X

High Quality defaults


The graph below shows the performance levels you can get from a single GPU all the way up to four: 

Obviously the biggest jump comes from one to two GPUs, but we still see some reasonable gains going from two to three. The value of a fourth GPU is simply nonexistent in most of the cases.

Number of GPUs HL2 UT3 Bioshock CoD4 Crysis
1 x Radeon HD 3870 39.3 46.7 36.9 25.3 14.0
2 x Radeon HD 3870 (1 X2) 71.9 84.1 63.2 50.0 26.2
3 x Radeon HD 3870 (2 X2 + 1) 93.3 112.7 86.7 72.2 26.4
4 x Radeon HD 3870 (2 X2) 102.2 114.6 92.5 93.2 27.7


Let's put some percentages with the graph above to put things into perspective:

Configuration HL2 UT3 Bioshock CoD4 Crysis
2-way CF Improvement over 1 card 83% 80% 71% 98% 87%
3-way CF improvement over 2 cards 30% 34% 37% 44% 0%
4-way CF improvement over 3 cards 10% 3% 7% 29% 4%
4-way CF improvement over 1 card 160% 150% 151% 268% 98%


The move from one to two cards generally yields a healthy performance improvement, but the gains taper off as we look at the performance added by a third GPU. Call of Duty 4 is the only game that shows solid gains with 4 GPUs (29% over a 3-GPU configuration), the rest of the titles show mostly single-digit percentage improvements.  Once again we see that Crysis simply needs new, faster GPU architectures - four GPUs does absolutely nothing for this game.

Remember our 3-way SLI review from earlier in the year? Although the test systems were very different, the GPU-bound scaling should be comparable (if anything, NVIDIA should have the advantage of being run on an Intel system). Here's a quick look back at some of the comparable benchmarks:

Configuration UT3 CoD4
2-way SLI Improvement over 1 card 84% 90%
3-way SLI improvement over 2 cards 20% 28%
3-way SLI improvement over 1 card 121% 143%


Single to dual GPU scaling is similar with SLI as CrossFire, but the 2-way to 3-way gain is better on CrossFireX. NVIDIA's quad-SLI is not yet available so we're not sure how the 4-way scaling will compare, but so far it seems like CrossFireX is doing quite well in the pure numbers game.

Drivers and Requirements Power Consumption & Final Words
Comments Locked

28 Comments

View All Comments

  • PhantomKnight - Thursday, February 21, 2008 - link

    Sorry getting used to the comment posting.
    This would have helped.
    3 x Radeon HD 3870 (2 X2 + 1)
  • NicePants42 - Thursday, February 21, 2008 - link

    I was also wondering what that meant. Looks like two 3870x2s plus one 3870.

    I was also a little confused about the system setup - looks like they were using nVidia hardware and drivers...
  • tynopik - Thursday, February 21, 2008 - link

    obviously it's a typo and they meant 1 X2 + 1
  • PhantomKnight - Thursday, February 21, 2008 - link

    Am I alone in being confused, i thought that 2 times 2 plus 1 is equal to 5.
  • Anonymous Freak - Thursday, February 21, 2008 - link

    Previous 3870 CrossFire vs. 3870X2 reviews have shown that the X2 performs better than two standalone 3870s in CrossFire. So using "2 GPUs" as the X2 only might be deceptive. I want to see:

    One 3870
    Two 3870s
    Three 3870s
    One 3870X2
    One 3870X2 + one 3870
    Two 3870X2s
    (I don't think there is any board that could take four plain 3870s.)

    Simply because from previous reviews, the "one to two" delta is much higher for X2 than for two plain 3870s. Another question is if the delta from one to two 3870s is the same as one to two 3870X2s. Based on previous X2 reviews, it looks like by all rights, the system really sees the X2 as a single double-fast GPU.
  • Goty - Thursday, February 21, 2008 - link

    The MSI K9A2 Platinum can. =)

    Actually, I think that's the same motherboard used in this review.
  • Anonymous Freak - Tuesday, February 26, 2008 - link

    Except when you use all four slots, they all drop to x8. And, yes, you would need a case with 8 back-panel slots, including one beyond the 'bottom' of the card. (Or you'd need to hack your cards to be single-slot.)
  • strikeback03 - Friday, February 22, 2008 - link

    How many cases have the openings on the back for 4 dual-slot cards?
  • yyrkoon - Thursday, February 21, 2008 - link

    "From AMD's explanation it sounds like there's still a lot of work to be done on the CrossFireX driver. While we can expect to see its public debut in March, it seems like it'll be a while before we're anywhere close to ideal scaling. We've found ourselves in this position with many-GPU designs in the past, at least the players are taking things a bit more seriously this time around."

    You guys, and many others have been talking about how well quad core CPUs scale(or dont) in a none server, are you really all that surprised to find that quad GPUs would be similar ?

    I have been long suspecting that while the PCIe specification is absolute in how much bandwidth a number of PCIe lanes are supposed to have total, that motherboard manufacturers have been skimping as to how much they are allowing those lanes use. Maybe this is just now biting them in the behind ? Or maybe I am just being silly . . . ; ) Either way, system details are not exactly forthcoming here.

  • skiboysteve - Thursday, February 21, 2008 - link

    ... What?

Log in

Don't have an account? Sign up now