by Rory Buszka on October 1, 2007 in Motherboards
Gigabyte’s new GA-MA69GM-S2H board has HTPC written all over it, with no fewer than six possible video output types, optical Toslink output, and AMD’s 690G chipset with Radeon X1250 IGP. In our review, we find more than a few reasons to recommend it to anyone who’s considering a new HTPC.
The next three tests will evaluate the admittedly limited 3D performance of the two motherboards in our comparison. IGPs still have a good distance left to travel before they become viable solutions for all but the very lightest 3D gaming loads (the Radeon X1250 graphics core of the 690G chipset being roughly equal to half of a Radeon X700), but 3D gaming still remains one of the best ways to test a system’s overall performance, because 3D rendering stresses every subsystem of a machine. All of these tests were performed at 1024×768 resolution without anti-aliasing or anisotropic filtering to ensure that meaningful frame rates could be sustained.
Each release of Futuremark’s popular 3DMark 3D gaming benchmark utilizes the latest technologies available during that year. In this case, 3DMark2003 represents games that are between three and four years old. 3DMark2003 was the first edition of 3DMark to support Microsoft’s DirectX 9 (version 9.0a).
This score incorporates the full set of game and feature tests, and provides a picture of what sort of performance can be expected in the earliest DirectX9 titles. Both the 690G and GeForce 6150 boards essentially flew through these tests, which weren’t particularly strenuous. It’s realistic to assume that games of this vintage would be playable at the low resolution we used in testing.
This 3DMark release featured support for DirectX 9.0c, as well as Shader Model 2.0. The X700-derivative X1250 video core of the RS690 northbridge only supports up to SM 2.0, while the GeForce 6150 IGP supports SM 3.0 and HDR, which may mean more for improved image quality.
Once again, we’re simply running the default suite of tests to get an overall score. The DirectX 9.0c tests will stress the video card with an increased level of detail and complexity characteristic of two-year-old games, as well as Shader Model 2.0 features and improved image quality.
From these results, we can see two things – first of all, playability suffers dramatically with the increased complexity of the 3DMark2005 tests, with scenes like Return to Proxycon chugging along at between 5 and 10 frames per second. Scores are also significantly lower. As a comparison, the same CPU, memory and hard drive loaded into a system with a 256MB Radeon X1650XT discrete video card scored 7443 3Dmarks, nearly seven times the score of the Gigabyte board’s X1250 integrated graphics core.
Also apparent is an increasing performance rift between the GeForce 6150 and Radeon X1250 graphics systems, with the 690G possessing a 34% performance advantage. Let me take this moment to remind you that both video processors are operating with their default settings – there’s no overclocking or foul play going on here. In fact, we tried to see if we could increase the GeForce 6150 IGP’s score and narrow the margin, but nothing we tried (short of overclocking) had any meaningful effect.
If you’ve ever toyed around with benchmarking apps to any major extent, you’ve probably noticed that many of the tests in 3DMark 2006 look similar to those in 3DMark 2005. In fact, many of the same tests appear to have been recycled in 3DMark2006, but look closer – the 3DMark2006 tests feature full Shader Model 3.0 usage, including HDR (High Dynamic Range) technology. Here, we expect to see better image quality with the GeForce 6150 chipset, but will that added rendering complexity throw a major bottleneck into the works?
The default battery of tests was used again, allowing 3DMark2006 to select the tests most appropriate to the hardware we were using. HDR and Shader Model 3.0 tests were automatically enabled for the GeForce 6150 IGP, while the Radeon X1250 IGP was allowed to skip those tests.
Both boards turned in scores that were depressingly low, in addition to near-unplayable frame rates. However, again the GeForce 6150-based motherboard suffered an approximately 34% performance deficit, likely penalized for its attempting to render complex SM3.0 and HDR scenes. The Radeon X1250 IGP’s significantly higher performance in this test demonstrates the advantage in efficiency of simply jettisoning SM3.0 and HDR support altogether. In an IGP, graphics processor engineers need to fight for every FPS in order to achieve a result that is barely playable, even if it means that features need to be thrown out in the process. Here the cleverness of the decision to omit SM3.0+HDR is proven.
As part of every 3DMark test battery, a CPU-rendering test is also included. This score gives the GPU a short breather while the CPU huffs and puffs its way through rendering a simple scene. These tests tax the CPU to its maximum, as well as stressing the memory subsystem and chipset.
As we see, the results are fairly even, with the Gigabyte GA-MA69GM-S2H having a slight advantage in this test. It’s likely that memory bandwidth played a part here.
The synthetic gaming benchmarks tell part of the story, but what about real-world gaming performance? Next, we’ll compare the two motherboards’ performance in a pair of popular games.