3DMark Vantage Feature Tests

This benchmarking suite isn't new, but its Feature tests support D3D10. Those are generally interesting, because they differ from ours. Perhaps, we'll be able to draw new helpful conclusions not indicated by RightMark.

Feature Test 1: Texture Fill

The first test deals with texture fill rate. It fills a rectangle with values read from a small texture with many texture coordinates, which are changed in each frame.

This Vantage test shows results that differ from the RightMark's. And it seems these numbers are more like it. As you can see, NVIDIA cards use TMUs more efficiently. And though GeForce GTX 480 still lags behind Radeon HD 5830, the lag is small and close to the theoretical value.

But, of course, we're interested in GeForce GTX 460 more. It shows a good result, only yielding to GeForce GTX 480 and Radeon HD 5830 -- all according to theoretical assumptions. If we compare GeForce GTX 460 with GeForce GTX 465, yet again the difference is adequate to technical specifications.

Feature Test 2: Color Fill

This one uses a simple pixel shader, which does not limit performance. An interpolated color value is written into the off-screen buffer (render target) with alpha blending. It uses a 16-bit off-screen buffer of the FP16 format, which is often used in games employing HDR, so this is a timely test.

The performance results demonstrated by this test do not correspond to what we've seen in similar benchmarks, even considering the difference in formats (we used an integer buffer, 8-bit per component, while 3DMark Vantage uses 16-bit floating-point). 3DMark Vantage demonstrates ROP performance, not memory bandwidth.

These test results more or less correspond to the theory. They depend on ROPs and their clock rate, not on memory bandwidth. GeForce GTX 460 performs well compared with both Radeons which have a bit lower theoretical fillrate. And it yields to GeForce GTX 465 for the same reason -- even a cut-down GF100 has more ROP power. Naturally, GeForce GTX 480 is far ahead in this test.

Feature Test 3: Parallax Occlusion Mapping

It's one of the most interesting feature tests, as this technique is already used in games. It draws one quadrangle (to be more exact, two triangles) using Parallax Occlusion Mapping that imitates complex geometry. The test uses resource-intensive operations to trace rays and high-res Z maps. This surface is also shaded using the heavy Strauss algorithm. This test uses a very complex and heavy-load pixel shader with multiple texture lookups during ray tracing, dynamic branches, and complex Strauss lighting algorithms.

This test is different, because it depends on shader power, branching efficiency, and texture fetch rate combined. That is, it takes a balanced GPU and card to reach high speed. The efficiency of executing branches in shaders has a strong effect on results as well.

Unfortunately, GeForce GTX 460 demonstrated its worst result in this test. The new mid-end graphics card lags behind Radeon HD 5770, Radeon HD 5830 and GeForce GTX 465.

The reasons for such low results are not clear. Perhaps, we should blame the lower branching efficiency if GF104. Or perhaps it's all because of lower memory bandwidth. Let's hope that GF104 will do better in physical simulation tests. We'll also see if we are right about the reduced branching efficiency.

Feature Test 4: GPU Cloth

This test computes physical interactions (cloth simulation) using a GPU. It uses vertex simulation with the help of vertex and geometry shaders, with several passes. Stream out is used to move vertices from one pass to the other. Thus, this feature test benchmarks execution of vertex and geometry shaders, and stream out speed.

It seems that rendering performance is affected by a complex mix of parameters in this test. Most likely, the general performance depends on geometry handling efficiency and geometry shader performance. GeForce GTX 460 does well in this test. It lags behind GeForce GTX 465 a bit, but that may be caused by lower memory bandwidth or lower complex shader performance.

In this test GeForce GTX 460 is also ahead of Radeon HD 5830 and Radeon HD 5770. As you can see, GF104 is all right in terms of geometry shaders and geometry performance. It does lag behind GF100, but that's adequate to the theory.

Feature Test 5: GPU Particles

Physics simulation test based on particle systems computed with the help of GPUs. The test also uses vertex simulation, where each vertex is a single particle. Stream out is used for the same purpose as in the previous test. The test computes hundreds of thousands of particles, all of them being animated separately, and their collisions with a bump map. Similar to one of our tests in RightMark3D 2.0, particles are drawn with a geometry shader, which creates four vertices from each point and forms a particle from them. However, the heaviest load falls on shader units (vertex computing), stream out is used as well.

These results are similar to what we saw on the previous diagram, but this time GeForce GTX 480 does a much better job compared with other graphics cards. The GeForce GTX 460 also does well, outperforming Radeon graphics cards and not losing much to GeForce GTX 465 that has more geometry units.

Feature Test 6: Perlin Noise

The last feature test is arithmetically intensive for a GPU. It calculates several octaves of Perlin noise in a pixel shader. Each color channel uses its own noise function for higher GPU loads. Perlin noise is a standard algorithm, which is often used for procedural texturing, it's a mathematically complex procedure.

This math feature test demonstrates pure GPU performance in peak-load tasks. Its results correlate with theory and partially correspond to what we've seen in our own RightMark 2.0 math tests.

GeForce GTX 460 performs on a par with GeForce GTX 465, which is about adequate to the theory (the new GPU should've performed better than it actually did). The lag behind both Radeon cards is the same, and Radeon HD 5830 is the leader.

ATI graphics cards always win in this test. They do better in simple but intensive math, but the picture changes when it gets complex -- e.g. physical calculations above -- and GeForce GTX 460 does quite well.

Conclusions on synthetic tests

Judging by the synthetic test results, we can say that NVIDIA made a good mid-end solution.

The new GPU offers improved performance and features, and graphics cards based on it may be rather attractive in their price range. The GPU itself differs by more geomery units and their parallel operation, which improves geometry processing performance.

NVIDIA GF104 performs well in synthetic tests, involving geometry shaders and physical calculations (cloth and particle imitation of the Vantage suite). It also does a good job in other branching-heavy calculation tests. But ATI solutions remain leaders in simple, well-parallelized tasks in RightMark and Vantage.

It seems that certain changes to the GF104 architecture (compared with GF100) did result in branching performance drop. But it may also be related to compiler peculiarities. So maybe things will improve with newer drivers.

We expect that GeForce GTX 460 should also do quite well in real-life tests that follow. It should be slower than GeForce GTX 465 on general, but a bit faster than the competing ATI Radeon HD 5830, because rendering performance in games often depends on several factors, including fillrate and texturing. The difference shouldn't be big, but in games, which are not limited by math only and that use tessellation, the novelty may outperform Radeon HD 5830 by 15-20%. Let's see if this is true.

Write a comment below. No registration needed!