AMD Mantle Real World Performance

AMD has been talking about the potential performance increase that their new Mantle technology is going to provide, we have all seen the numbers that AMD and EA-DICE has been releasing.  AMD has finally released CATALYST 14.1 beta drivers, so now we can do our own independent testing to check out performance with and without Mantle enabled.

AMD Mantle API

We were given early access to the Catalyst 14.1 beta driver, there are a few hoops to jump through during the installation process. We were able to do a successful install, but it looks like the installer is causing some issues for AMD.  Here is a look at Catalyst Control Center that confirms we are now running Catalyst driver package 14.1.

AMD Catalyst Control Center 14.1

Currently there are only two ways to test the AMD Mantle APU. You can use the latest build of the highly popular Battlefield  game title or there is a new synthetic benchmark called StarSwarm that is now available to download and use. We tried both benchmark options out on what we consider a mainstream gaming platform (Intel Core i5-3570 'Ivy Bridge Process w/ a Sapphire Radeon R7 260X video card).

 Let's see if the numbers that have been released is pure marketing, or whether Mantle performs as well as we're told it does.

Test System

Before taking a look at the benchmarks, let’s take a look at the test system that was used. All testing was done with a fresh install of Windows 8.1 Pro x64, no additional software was running during the tests.

Sapphire R7 260X Front

Intel Z77/LGA1155 Platform

Z77X-UP5 TH CPUZ Screens

The Intel Z77 system used to test the video cards done on a system with an Intel Core i5-3570K 'Ivy Bridge' processor on the Gigabyte Z77X-UP5 TH motherboard with the F11 BIOS from 09/03/2012. Installed for memory was 16GB Kingston Hyper-X Blue 1866MHz, running at the rated speed with 9-11-9-27 2T timings. An Intel 520 series 180GB SSD was used for the OS and benchmarks. Testing was completed in a room with an ambient temperature of 65F.
The Intel Z77 Test System

Component

Brand/Model

Live Pricing

Processor Intel Core i5-3570K Click Here
Motherboard Gigabyte Z77X-UP5-TH Click Here
Memory 16GB Kingston Blue 1866MHz Click Here
SSD Intel 520series 180GB Click Here
Cooling Phanteks PH-TC12DX Click Here
Power Supply Lepa G650-MAS Click Here
Operating System Windows 8.1 Pro 64-Bit Click Here

Sapphire R7 260X Video Card GPU-Z Information 

Sapphire R7 260X GPU-Z Base

Battlefield 4

For Battlefield 4, you will need the beta drivers and a large patch to enable Mantle support, this downloads automatically when you start the game; in game you will also have to enable the Mantle API in the Video section and then restart the game.  For more information See Here.  

Battlefield 4 Mantle API

Sadly, FRAPS does not support Mantle at this time, so DICE has enabled a new console tool to do benchmarking; "PerfOverlay.FrameFileLogEnable 1" to start the recording and "PerfOverlay.FrameFileLogEnable 0" to stop the recording.  While we would like to thank DICE for this tool, it would be great if they would do an automated benchmark tool.  The tool saves the results as a .CSV file, and with a little analysis we can determine the FPS.  As is common practice, the same portion of the mission will be run three times and the results averaged.

Battlefield 4 Mantle

Benchmark Results:  In Battlefield 4, using DICE's new benchmarking tool we received a 15.8% performance boost once Mantle was enabled at a screen resolution of 1920x1080. This is not a bad performance uplift and we'd consider a system with an Intel Core i5-3570 processor along with a Radeon R7 260X a pretty good real world platform for gamers. 

 

Star Swarm

Star Swarm is the newest benchmark focusing on Oxide Games next generation Nitrous engine.  Nitrous was designed for hardware found in modern PC's, Playstation 4 and XBox One. Built on a 64-bit multicore engine, it will allow a high number of 3D objects to be drawn and rendered at the same time.  

Star Swarm Menu

The benchmark has various settings, we will be using the Extreme setting and allowing it to run for 6 minutes.  Due to Star Swarms random nature, each of the tests will be run three times, with the results averaged.

Sapphire 260X Star Swarm Follow

The follow scenario focuses on a single unit to provide more consistent results.  The scores went from 26.56 in DX11 to 32.86 with Mantle; a 23.7% performance boost.

 Sapphire 260X Star Swarm

Attract uses randomized camera movements for a greater amount of variation.  A huge performance boost in this test scenario, 45.4%!

 Sapphire 260X Star Swarm RTS

Utilizes a long range camera as we would see in a RTS game.  Another high performance boost with Mantle, coming in at 29.1%.

Sapphire 260X Star Swarm SHMUP

SHMUP is a top down view for "shoot 'em up" style games.  While not as impressive as the previous couple scenarios, a 16.1% performance boost is still a great advantage.

Benchmarking Results:  Receiving a performance boost between 16% and 45%, with an overall average of 28.58% performance boost in Star Swarm with Mantle.

Final Thoughts and Conclusions

Mantle is a new unproven technology by AMD for their HD7000, R7 and R9 video cards that are Graphics Core Next (GCN) compatible.  We were lucky enough to test the first beta release driver by AMD we saw a significant performance improvement with the first driver release. As the technology matures we hope to see bigger performance gains from Mantle and we really hope that AMD can work with FRAPS to get the benchmark utility to work with the Mantle API. Pretty much every hardware review site relies on FRAPS, so hopefully that will come sooner rather than later.

AMD Mantle API

Both AMD and EA-DICE released some performance improvement numbers and shows substantial performance gains at 1080P in Battlefield 4. AMD found an estimated average of 13.28% performance boost in BF4 at 1920x1080.  Our testing of the driver on a platform with the previous generation Intel Ivy Bridge processor and AMD Radeon R7 260X resulted in a 15.8% performance boost, just slightly higher than AMD's average.  Taking into account the differences in our test system and their configuration this is a pretty good average and a heck of a performance boost from a driver and game updates. 

For StarSwarm, we received a 28.58% performance boost across all four of the scenarios when looking at DirectX versus Mantle performance on Catalyst 14.1 beta drivers  However, the very first time that we ran StarSwarm we used Catalyst 13.12 drivers and the difference between DirectX 11 performance on Catalyst 13.12 and Mantle performance on Catalyst 14.1 beta was found to be a staggering 85% at 1920x1080. (We were also told by AMD that a StarSwarm update was pushed out on steam, so that could be a major reason too)  Those scores were not shown, as we figured that once the Mantle drivers come out, everybody will be utilizing the drivers whether running DirectX 11 or Mantle.  

Legit Bottom Line:  Overall, Mantle is a very interesting technology for GCN compatible video cards.  As more games support Mantle we will be able to see exactly how this new API compares, and will make comparing nVidia cards to AMD cards a little more difficult.