Conclusion

As we already mentioned, this article marks the start of a new series, and we're going to show you test results und 3-Way-SLI scaling. Comparing to our 2-Way-SLI articles the conclusion is going to be quite a bit shorter. The reason is, that only with theoretical benchmarks we find consistent scaling and in case of games, well, they're just not optimized for 3-Way-SLI. Apparently there are no rules without exceptions.

Referring to 3DMark we see that the GPU socre is between factor 1.41 (standard clocks) and 1.40 (4.5 GHz) higher than with one card. In other words this means, that adding a third card helps improving performance by roughly 40 percent. A closer look at Unigine Heaven reveals that in case of the basic preset scaling factor is 1.02 at stock frequencies and 1.00 with the CPU at 4.5 GHz. When it comes to the Extreme Preset then we see that the GPU has much more influence again (obviously), since there is much higher load on the VGA(s). In this case scaling factor with standard clocks is 1.54 and at 4.5 GHz we get 1.38.

Games on the other hand are a completely different story. Scaling factors at low resolutions vary from 1.00 to 1.17. 1 basically means, the fps didn't change at all and 1.17 marks a 17 percent increase in FPS. A 17 percent gain was measured using Skyrim and running the CPU at standard clocks.

Now lets focus on high resolution and high details, where the games cause quite some load on the two graphics cards. In this case the lowest scaling factor we found was 1.04. In fact, most of the games fps rates don't scale with adding a third graphics card. There is only Call of Duty Black Ops 2 where we've measured a significant increase, which means, the performance went up by 38 percent (a factor of 1.38).

Putting more than two graphics cards into a gaming PC is basically not going to affect frames per second in most games. There are only very few titles where triple SLI is supported. The only reasonable thing to use triple SLI for is breaking world records and benchmarking in general. Only when running theoretical benchmarks we come accross consistent scaling and with most of today's games, there is no difference at all. As for the different frequencies - stock and 4.5 GHz - they also don't account for significantly lower or higher results.

Page 1 - Introduction	Page 9 - Call of Duty Black Ops 2
Page 2 - Test Setup	Page 10 - Sleeping Dogs
Page 3 - 3DMark Fire Strike	Page 11 - The Elder Scrolls V: Skyrim
Page 4 - Unigine Heaven 4.0	Page 12 - Metro: Last Light
Page 5 - BattleField 4	Page 13 - Power Consumption
Page 6 - BattleField 3	Page 14 - Scaling Factor
Page 7 - Bioshock Infinite	Page 15 - Conclusion
Page 8 - Crysis 3