6. NVIDIA GeForce FX Go5200/5600, Continued

NVIDIA GeForce FX Go5200 features:

DirectX9 Chip clock up to 300 MHz Memory clock up to 300 MHz Voltage Range: 1.2-1.3V 4 pixel pipelines (4 pixelshader operations per clock signal) 0.15 micron fabrication process PixelShader 2.0 support, full floating point precision (128 bit) VertexShader 2.0 support 4 FSAA (MultiSampling) 8x anisotropic filtering PowerMizer3
Power management for the chip, which can be manually controlled by the user, or which can be defined according to the performance requirements of the application running. Dynamic Voltage Scaling: automatic voltage adjustment as needed Dynamic Clock Scaling: automatic, dynamic clocking of the chip (on-the-fly) HDTV-Out (?) - according to NVIDIA, this is supported, although the specs only mention TV-out with 1024 x 768 Resolutions up to QXGA (2048 x 1536)

The differences between the two chips can only be seen if you take a second, closer look. Apart from the slower clock speed, the Go5200 suffers above all from the less optimized memory controller. Roughly speaking, the memory controller of the FX Go5200 corresponds to that of the GeForce 4 Ti chip, which leads to hefty losses in performance when using FSAA and anisotropic filtering. The FX Go5600, by contrast, has the optimized memory interface of the "big" NV 30 chip, including color compression and other features. We've already talked about how this affects performance for the desktop variant in a previous article: NVIDIA GeForceFX 5600 Ultra & FX 5200 Ultra: Performance.