3DMark Vantage - The Performance Report

Has always been the publication of a new 3DMark made by Futuremark an important event in the community of hardware enthusiasts, so that the 3DMark is Vantage in the future by many as a measure of the performance of their system and used primarily the graphics card - whether this is justified and realistic, even to be seen. appear that the 3DMark Vantage sooner or later would be, was already no longer a secret and even the exact release date Futuremark Announces now on your own website in advance based on a countdown to get the attention of the media and enthusiasts safe. The eighth incarnation of the now follows in terms of naming the 3DMark PCMark and calls itself as simply "3DMark Vantage". After they had invested in 3DMark06 just a year of development, lie between the 3DMark Vantage and 3DMark06 now another two-year intervals as in the previous versions. In the now traditional "Performance Report" at the publication of new 3DMark we want to present not only our own results with the graphics cards from Nvidia and ATI in the new product from Futuremark, but return them to the reader the possibility to submit their results in the forum to the creation of an overview and comparison.

In addition to a free trial version, which only once running can be, however, the Basic Edition is based and allows the result of the 3DMark Vantage in the Online Result Browser (ORB) to send, Futuremark offers another three versions of 3DMark Vantage to the. The cash is again ready for customers to benefit from some additional options. The Basic Edition is for all 'meant those who want to run the benchmark once more only, for which, however, already 6.95 U.S. dollars must be paid. To do this you can then use the benchmark as often in the so-called performance-Preset and send the scores to Futuremark's Online Result Browser. In the Advanced version for $ 19.95 then, all the other presets released, it will feature six additional tests available and all settings are accessible. The purpose intended for commercial Professional version also offers technical support, command line scripting and additional options.

The above-mentioned (posted by "Aamin")preset defined by the following basic requirements:
Presets

• Entry - DX10 graphics card with 128 MB and a resolution of 1280x1024 pixels
• Performance - DirectX 10 graphics card with 256 MB and a resolution of 1280x1024 pixels
• High - DX10 graphics card with 512 MB and a resolution of 1680x1050 pixels
• Extreme - DX10 graphics card with 512 MB and a resolution of 1920x1200 pixels

Who appropriate version of 3DMark Vantage buys can also select other resolutions, but an entry in the Online Result Browser is related to one of predefined presets only. This means that for the High and Extreme presets a widescreen monitor to a basic hardware requirement.

With 3DMark Vantage, Futuremark has not set to well known scenarios, but all the tests developed from scratch. The first graphics test is called "Jane Nash" and represents a game scene in closed rooms with complex characters, multiple dynamic lights and complex surface lighting. It uses several hierarchical rendering steps, including reflections on the water surface and the refracted light, and the physical simulation of collision detection. The second game test New Calico "is a powerful scene in space with many moving objects is a particularly solid but has the scene of a giant planet and a dense asteroid belt. In addition to these many moving and often instantiated objects marked effects such as Parallax Occlusion Mapping, True Impostors and volumetric fog the scene.

The Al-test (artificial intelligence) has a workload with many cooperative maneuvering and away-place calculations of Artificial Intelligence (AI). As the scene Futuremark chose a race track for aircraft, which all try to navigate through a series of gates, but without touching each other or the ground. The algorithm for finding the way is guided by three characteristics: the doors have to be flown to the series, it predicts a physical flight model of a fixed wing aircraft, and it avoids collisions with other aircraft. The load is achieved by planning the movement of each aircraft and can be parallelized, so that even multi-core processors busy. A generator function randomly generated a large number of candidates of possible routes, after which the best candidate is selected on the basis of characteristics such as proximity to the base and proximity to other aircraft. Faster processors can calculate routes for the more subtle planes, because the number being considered as routes longer fails on slow processors.

The physics test is to simulate the physics calculations in future game titles. Owners of a PhysX card by Ageia come at their expense, because the test uses the card for physics computation. The scene shows a race of planes, whereby the gates are arranged very dangerous. The planes take to the smoke, which collide with the obstacles or the ground and each other. In addition, the smoke reacts to pass flying aircraft. As the aircraft all at the same time trying to fly through the gates, there will be collisions. The planes are composed of twelve rigid parts that are held together by eleven fragile joints. The breakup of the aircraft after a collision is not predetermined, but is only made by the physics calculations in real time. In the scene there are two different types of gates: donut-shaped floating gates and those consisting of two cone-shaped towers. Both types of doors are deformable and interact with other objects of the scene. The number of pairs of doors (worlds) is dependent on the number of processor cores and the use of a PhysX card. The one pair of gates is assigned to a CPU. A CPU with two cores thus resulting in two goals pairs.

A CPU with two cores with a PhysX card in your PC already, however, leads to five goals in pairs. Four cores get four goals couples, with PhysX card it had seven. Each pair of gates is its own physical world, so that the worlds not affect each other. Futuremark chose this step to separate all possible worlds and quickly calculate and do not have to vote again and again each other. Thus, the PhysX card is the assigned worlds calculate faster or slower than the processor cores to their assigned worlds. The two new CPU tests are intended to create a game-like scenario and also provide the user during the test, something interesting to view. Both CPU tests use the same graphics engine, which is also used in the two game tests used, but the graphical computational complexity is reduced to minimize the effect of the graphics card on the outcome of the tests. Therefore find no post-processing effects, complex shader or shade use.

Parallax Occlusion Mapping is a relatively new method to take without too much computing power in fitting, the gradient presented in textures, so that they do not 2-dimensional, but realistic. The 3DMark Vantage uses POM to represent a complex geometry under the surface of a triangle. This is done with ray-tracing methods for a large depth-map (4k x 4k), which contains the depth information. In addition, some light sources and render the computational algorithm must determine which geometry is in a visible area. The GPU-Cloth-test takes place a cloth simulation, which is itself calculated by the GPU. This one uses vertex and Geometry-shader programs that will generate multiple render passes (rendering pass) a realistic behavior of the flags. Furthermore, to make you look at the new stream-out capabilities of Direct3D 10-accelerator advantage, so you can quickly copy the individual vertices of a calculation to the next.

With the new 3DMark, the hardware requirements have increased. The new version requires a mandatory first DirectX10-compliant graphics card and hence Microsoft's Windows Vista. Minimum system requirements :

• x86/x64 single core CPU with SSE2 support, performance comparable to Intel Pentium D 3.2 GHz or better
• D3D10 compatible graphics card with 256 MB memory
• 512 MB RAM
• 1 GB free hard disk space
• Windows Vista with Service Pack 1

Recommended system

• Core 2 Duo E6600 or AMD with the same power (Athlon 64 X2 5400 +)
• D3D10 compatible graphics card with 512 MB memory
• 2,048 MB memory
• 1 GB free hard disk space
• Windows Vista with Service Pack 1

After a rather long wait is finally here, the "3DMark Next", which sounded like the latest scion of the PCMark series on the name Vantage. And the expectations of the benchmark were great. Not only that Futuremark programs until now mostly were impressed by a huge spectacular graphics, 3DMark Vantage also was the beginning of the Direct3D API for programming the tenth generation. However, the Fins a real image of fireworks is a success? And above all, could the major criticism being that the end result behaves differently from a real game, eliminate?

Quote:

---

However, the Fins a real image of fireworks is a success? And above all, could the major criticism being that the end result behaves differently from a real game, eliminate?

---

A "yes" or "No" to both questions would be inappropriate. The 3DMark Vantage lot of things right and better especially as the veteran 3DMark06, but remains logically, yet only a synthetic benchmark. Graphically can convince the program. The first game test looks mostly good, but remember off the main protagonist of the cast more of a bumbling cartoon character as a dangerous opponent. That one does it better, shows the main character who is very successful. In the second game test, there is nothing to complain about, however, because it knows fully convince. The graphics are invariably very good and even excellent in some places - like something we have not seen it until now. Send textures, different shader effects, a well-done lighting, fine particle effects and much more spoil the eye.