In our original Mantle Showdown series we looked at the performance of AMD’s Mantle API across a variety of different platforms. In a surprising turn of events someone at AMD actually read all of those articles and dropped me a line asking why I didn’t test Mantle on an all-AMD high-end platform. We’re back on this topic again, this time with a Mantle versus DirectX comparison using AMD’s FX-8320E CPU.
There are a few differences between this article and the previous round of testing. First off our old testing was done without the benefit of AMD’s Omega driver. We’re also using a different OS this time around. Previously we were running the Windows 10 Technical Preview. But after three updates that technical preview has become an unstable, buggy mess that no longer offers the reproducibility that we expect for our benchmarks. We’ve moved back to testing on Windows 8.1 x64 due to those issues.
Here’s our testing platform:
For the sake of transparency we want you to know that AMD provided the R9 290X GPU we’ll be testing today along with codes for all the games we’ll be looking at. AMD also specifically provided the CPU, motherboard, and RAM for use in this article because we blew up our last AM3+ motherboard when our $100, 1000 Watt power supply shorted. All the other parts we’re using were purchased at retail without the knowledge or consent of those companies. As always you can find our raw testing data on OneDrive. We took no outside input for this article other than the suggestions and support coming from our lovely forums, thanks guys.
Games that Support Mantle
Let’s run through what games we’ll be benchmarking today. First up we have the recently released turn-based strategy game Civilization: Beyond Earth. This is a game that launched with Mantle support and is very CPU intensive. It also showed very significant gains with Mantle in our initial review. Next up with have Plants Vs Zombies: Garden Warfare which also launched with Mantle support. Running on the same Frostbite 3 engine used in the Battlefield series Garden Warfare is a third person shooter with a more upbeat atmosphere than most games. Then there’s Thief, a first person stealth game that patched in support for Mantle after launch.
Following those games we have Sniper Elite 3 which just recently patched in support for Mantle. Rebellion, the developer of gratuitous headshot simulator 2014, was very up front about the performance gains they were seeing in the Mantle version of their game. Finally we have the very first Mantle title, Battlefield 4. As you might expect these games support both the Mantle API and DirectX 11 so we can do as close to an apples to apples comparison as possible.
As part of our procedure we cook our GPUs before benchmarking them by running the benchmark in loop for five minutes prior to capturing our data. Gathering benchmarking data in DirectX 11 applications is easy thanks to the existence of a tool call Fraps. In comparison gathering data from a Mantle based application is a much more difficult task because of the lack of a tool with similar functionality to Fraps. Instead we have to use each game’s built-in benchmarking tools. For some games like Sniper Elite 3 this means after the built-in benchmark runs it will save a nice formatted copy of the results so that we can analyze them in Excel.
But for other games with less thoughtful developers this means that we have to try and measure rendering performance in any way we can. Plants Vs Zombies was the game we ran into the most trouble with due to a lack of documentation and no console commands. Those issues limited us to capturing only a snapshot of the game’s average frame rate from one point in time. If anyone is thinking of building a third-party Mantle benchmarking tool, please do.
We’re going to be looking at two resolutions: 1080P and 1800P or 3200 by 1800 thanks to AMD’s virtual super resolution technology. This way we can get a feel for how AMD’s new VSR feature affects our Mantle and DirectX performance results.
Mantle Improves Performance on High-end Hardware
There are a couple of interesting things to note here; first off the two games from EA have the weakest Mantle performance. Actually regressing from their DirectX 11 numbers at 1080P and only showing small gains with VSR. More importantly, and despite some major per game differences, on the balance Mantle shows similar scaling at both resolutions. Does Mantle improve performance over DirectX 11, yes. On our R9 290X and Intel Core i7-5960X test bed we saw a 121% performance advantage using Mantle. This matches nicely with our results from our all-AMD platform.
The take away here is that no matter what CPU vendor you chose, if you have a Radeon graphics card you will see performance benefit from using Mantle. AMD or Intel; it doesn’t matter. Mantle offers performance gains for users of both.S|A