The key point is a somewhat incorrect initial AMD message regarding Mantle. The company introduced the API as a graphic product that is an alternative OpenGL or DirectX. As a result, the main discussion on the topic of Mantle unfolded about how much the API will increase the performance of AMD video cards, increasing the number of requests for the reaching per second.
It was believed that Mantle would raise the performance of games, making graphic accelerators more efficient. In reality, the founded goal of Mantle — reducing the dependence of the GPU work on CPU and increasing performance due to this. This means that the main increase in the use of Mantle compared to DirectX 11 will depend on whether the power of the CPU is a limiting factor or not.
AMD believes that such scenarios are very common in games, since the existing APIs are too much load CPU and are poorly scaled on multi -core processors. AMD expects Mantle to provide fundamental advantages for most computers equipped with initial and middle -level processors. That is, the main increase from Mantle will be observed on chips like hybrid processors A10-7850K or even Kabini.
Mantle gives the developers more control over the command buffer, the number of CPU nuclei used and provides much better scaling when using several nuclei. One of the main functions of Direct3d 11 was support for multi -flow visualization, but AMD says that the technology does not work effectively enough.
So the main task is Mantle — Increase the overall performance of CPU and increase from scaling. AMD emphasizes that as a result, the technology has less influence on games launched in high resolutions with maximum quality — In these modes, the limiter often does not become the power of CPU, but the performance of graphics and at the API level is difficult to increase the efficiency.
However, Mantle also provides graphic advantages compared to DirectX 11 in the field of command of the command buffer, asynchronous DAM questions for downloading data, asynchronous computing capabilities and advanced technologies of full-screen smoothing. However, the company does not promise due to this fundamental increase in performance.
So far, only two applications are available that can use the advantages of Mantle — This is the game Battlefield 4 (from EA/Dice) and the technical demonstration of Star Swarm (Oxide Games). Therefore, AMD has provided its productivity measurements for these two projects.
- When a CPU is a limiter, an increase in 40.9% (for resolution 1080p) and 40.1% (1600p) when setting quality ULTRA and full -screen smoothing 4XAA — The AMD A10-7700K processor and the AMD Radeon R9 290X video card are used;
- When a GPU is a limiter, an increase in 2.7% (1080p) and 1.4% (1600p) when setting quality ULTRA and FXAA full -screen smoothing — Core i7-4960x processor and AMD Radeon R7 260X video card;
- Averaged growth on different configurations for the resolution of 1080p: 13.28% (290x and 260x graphics cards, I7-4960X processors, A10-7700K, FX 8350 and i5-4670K);
- Averaged growth on different configurations for resolving 1600p: 11.35% (290x and 260x graphics cards, I7-4960X processors, A10-7700K, FX 8350 and i5-4670K).
StarsWarm (Oxide Games):
- When a CPU is a limiter, an increase in 319% (1080p) and 281% (1600p) in the RTS test for Extreme quality settings — AMD A10-7700K processor, AMD Radeon R9 290x graphics;
- when the limiter is GPU, an increase in 5.1% (1080p) and 16.7% (1600p) in the RTS test for Extreme quality settings — Core i7-4960x processor, AMD Radeon R7 260X graphics;
- Averaged growth on different configurations for the resolution of 1080p: 115.65% (290x and 260x graphics cards, I7-4960X processors, A10-7700K, FX 8350 and i5-4670K);
- Averaged growth on different configurations for resolving 1600p: 75.19% (290x and 260x graphics cards, I7-4960X processors, A10-7700K, FX 8350 and i5-4670K).
As you can see, in the demonstration of Star Swarm on the Nitrous engine from Oxide Games, the most impressive increase is ensured. AMD calls Star Swarm a great example of Mantle use. However, it should be remembered that this is only a demonstration of the concept, and not an indicator of growth from Mantle in real games. In games with a mass of calculations and characters (like Company of Heroes or Starcraft 2) on a PC with relatively weak processors, Mantle technology should show the best results. In general, we can conclude that in the heavy modes of modern games on systems where CPU is not a limiting factor, the increase from the use of Mantle will be low. But this is in any case a free plus, especially since the performance of video cards grows faster than processors.
More games with AMD Mantle support should appear this year. We are talking about all future Electronic Arts projects on the Dice FrostBite engine, Sniper Elite 3 Games from Rebellion Entertainment, Cloud Imperium Games Star Citizen and Eidos-MontréAl. Perhaps soon other gameplayers will share their plans for games with support for Mantle.
- Bright Side of News