Usually we would use FRAPS to test video performance, however FRAPS does not support Mantle at this time, so that makes things a little more difficult. Thankfully Square Enix has included a benchmark with Thief. The benchmark takes us down a street during a storm, with people and guards interacting.
To test Mantle, Thief was installed through Steam on a freshly imaged test system running Windows 8.1 Pro. This test system included the following hardware:
Enabling Mantle is done through the Thief Options menu when you start Thief. All other settings remain the same whether you select Mantle or not. There is no option to switch to Mantle once inside the game, however you can deselect it when you restart Thief.
Within the Thief Options, there are six preset options for the video quality, Very Low, Low, Normal, High, Very High and Custom. We will be testing Thief using three of these, Normal, High and Very High. The integrated benchmark will be run three times and the scores averaged.
Using the normal graphic settings, Mantle took the average FPS from 61 to 101.4; this is a 24.87% increase in performance.
Without Mantle, the average was 53.6FPS, with Mantle enabled, it increased to 81.3; a 10.26% increase. The Maximum FPS did not change from the Normal to the High graphic settings, making me question the results a little, it’s as if Thief is limited to around 61FPS.
With the settings set to Very High, the average FPS increased 4.2%. Without Mantle the FPS was 60.5, which is a small decrease from the High settings, and again throws the results into question.
The Maximum FPS seems to be stuck around 61FPS without Mantle, if that’s the case, then Mantle changes that and allows it to go higher. According to this testing, the Average FPS increased from 4.2% to 24.87% depending on the graphic setting chosen.