I am very surpised that anandtech didn't show minimum fps numbers. I would have thought that mantle would help a lot in this case and provide a much more consistent experience. The developer even said that this is what they had achieved in a video (It was from the AMD 30 live event: http://www.amd.com/en-us/who-we-are/corporate-information/events/amd-30-live). I wonder if Mantle helps reduce the time to process a turn, unfortunately anandtech didn't test this.
Also, they show that the SFR solution did not scale with multiple GPUs. It would be great if Firaxis and AMD had spelled out what the advantages are of this solution and why they implemented it There are ways to use multiple GPUs to improve the experience beyond pure performance, maybe they did something like that in this case.
I am a big fan of the concept of Mantle. However, as I understand it, games are still being developed around the limitations of d3d 9-11.
If the draw call count were increased to a level where d3d would struggle, then I would imagine that there would be much more impressive gains from using Mantle even on very high end CPUs. However, it would be commercial suicide for a developer to do this when most customers cannot benefit or even run the game acceptably if this were the case, so I understand why they do not at the moment.
I find it frustrating that reviewers use very high end overclocked CPUs to test games that are built around the aforementioned limitations, thereby nullifying most of the gains that Mantle would have in these games where there aren't many draw calls anyway.
Hi ltron,
You make some very good points.
We will put up a blog soon and explain more about the benefits of SFR, as well as the Mantle implementation and EQAA. I think you will find it very enlightening. Certainly there is a lot more to it than average fps. However it should be noted that currently multi gpu is not working correctly. (regardless of API) I'm under the impression that a patch is due imminently to address such issues.