Gaming CPU Shoot-Out Redux: Multi-GPU Scaling - DiRT: Showdown

Submitted by skipclarke on 3 September, 2014 - 02:37

Article Type: 
Article
Teaser Icon: 
None

DiRT: Showdown

DiRT Showdown was tested at the Ultra setting pre-set, with 4X MSAA enabled. The test was run three separate times, with the testing done on the third pass. FRAPS was manually started with the benchmark fly-by of the track. It was subsequently stopped right before the car crossed the finish line, prior to the slow motion scene of the finish.

Please note that the DiRT Showdown test is not a canned demo like Metro Last Light or Unigine Valley. It is a sample race in a figure-eight style track. The cars jockey for position with the driver AI, and each run is inherently different.

Small variances between runs should be discounted, due to this variation between the actual tests. I did toss out any test run that had any crashes or a stuck car.


1080p Widescreen

DiRT: Showdown - 1080p HD       DiRT: Showdown - 1080p HD CrossFire    
R7 260X     R9 270X     R9 280X     R9 290 R7 260X CFX     R9 270X CFX     R9 280X CFX
Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max    
AMD FX-9590 39 46.1 60     53 63.2 80     64 76.8 96     72 85.4 111 71 81.9 112     71 84.8 124     71 84.2 125
AMD FX-9370 38 45.6 60     51 61.0 80     63 77.2 97     71 82.2 113 63 79.0 106     67 81.8 121     70 82.1 113
AMD FX-8350 35 42.7 59     47 56.4 79     61 73.6 99     69 79.9 113 64 75.7 105     59 78.5 117     59 78.5 121
AMD FX-6350 36 43.4 59     47 56.4 79     61 73.6 99     69 79.9 113 59 70.7 102     60 72.5 105     59 72.6 109
AMD FX-4350 34 42.7 59     44 57.1 77     55 63.4 96     55 67.1 102 48 60.8 92     48 60.6 92     48 61.3 97

Again, as with the previous article, there is a bit of performance impact on the upper GPUs. We see improvements in averages and minimums across the CPUs, when paired with the R9 280X and especially the R9 290. These are the only instances where a better CPU makes any difference. With the FX-6350, the minimums break the 60fps barrier.

Pic

With a CrossFire configuration, there are noticeable increases with each CPU. The FX-4350 continues to be a performance inhibitor. In many instances both the averages and the minimums are above fps. DiRT: Showdown has never been a demanding game, and many of the configurations would benefit from a 120Hz monitor.

Pic


1080p Eyefinity

DiRT: Showdown - Eyefinity       DiRT: Showdown - Eyefinity CrossFire    
R7 260X     R9 270X     R9 280X     R9 290 R7 260X CFX     R9 270X CFX     R9 280X CFX
Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max     Min Avg Max    
AMD FX-9590 17 20.4 29     25 29.5 41     34 39.9 53     39 45.3 61 32 38.3 53     46 52.9 72     62 71.4 94
AMD FX-9370 17 20.5 29     26 29.6 41     35 40.3 53     39 44.9 61 32 38.2 54     44 52.3 73     60 70.5 94
AMD FX-8350 16 20.0 28     22 27.4 39     32 39.3 54     39 45.7 62 31 37.4 51     46 53.2 74     60 69.7 94
AMD FX-6350 15 19.9 28     23 27.2 39     30 39.0 54     39 45.3 62 30 37.3 51     45 52.4 74     56 65.9 91
AMD FX-4350 16 19.9 29     22 27.5 39     31 38.7 54     38 45.2 62 17 20.3 30     43 50.0 69     43 56.4 89

As with the previous article, the results across each CPU is basically flat. Even looking at the minimums and maximums, the variations are within the margin of error.

Pic

With CrossFire, we do see some performance improvements with a pair of R9 280X cards. The FX-4350 does hamper performance, especially on the R7 260X. The FX-4350 carries parity with the other CPUs with the pair of R7 270X. Even the max frame rate with the R9 280X is on par with the other CPUs. However, the minimum and average scores are much lower.

Pic