by Anton Shilov
06/07/2006 | 10:57 AM
Perhaps, Intel Corp.’s graphics cores that are featured in the company’s chipsets are not the best on the planet, but it looks like Intel wants them to be as progressive as possible. According to a media report, the company’s new G965 chipset not only features shader model 4 capabilities, but also has progressive unified shader micro-architecture.
It has already been reported that Intel’s built-in graphics core of the company’s G965 chipset – which will be called Intel Graphics Media Accelerator 3000 (GMA 3000) – will support the DirectX 10 functionality, particularly, shader model 4.0, which requires the part to handle vertex shaders, geometry shaders and pixel shaders. However, so far it was not clear how exactly the GMA 3000 family would implement the innovative features.
According to slides published by HKEPC web-site, the Intel Graphics Media Accelerator 3000 will feature “multi-threaded array of execution units” as well as “dynamic load balancing”, which means that the shader processing engines will be unified and all the features – including vertex shaders and transform and lighting (T&L) engine which have been emulated by central processor – will be implemented in hardware, something, which leaves hopes for significantly increased performance compared to predecessors.
The Intel GMA 3000 will be generally significantly more advanced compared to the current- and previous-generations of Intel’s built-in graphics cores. The new core: it will have improved early-Z technology which reduces the load on memory bandwidth, it will support 16x anisotropic filtering, 32-bit precision floating point calculations and so on. Additionally, thanks to unified shader processing engines, the GMA 3000 will support hardware decoding of H.264 and WMV9b high-definition video streams. In addition, the new integrated graphics core from the world’s largest chipmaker will provide HDMI output. It is unclear whether the new graphics core will support HDCP encryption as well.
Traditional graphics processing units (GPUs) or cores have dedicated pixel shader and vertex shader processors. In cases, where a frame being rendered contains a lot of geometry and just a few pixel shaders to determine colour, the performance is limited by vertex shaders, while pixel processors are standing idle. GPUs with unified shader processors will have special scheduler inside them, which will determine the load from pixel, vertex and geometry shaders and then assign appropriate number of processors for every kind of work. This will allow to utilize all the computing power of the GPU, believes graphics chip specialist ATI Technologies. The company currently has flagship graphics processor Radeon X1900 XTX, which is clocked at 650MHz and has 48 pixel shader processors and 8 vertex shader processors, which – along with the other parts of the chip – cost the company 384 million transistors. Definitely, built-in graphics cores should consist of much lower amount of transistors, however, it means that they will have lower amount of executing engines, hence, it is crucial to use all the available computing performance.
Despite of ATI’s optimism about unified shader micro-architecture, rival Nvidia Corp. believes that implementation of a unified shader engines should be gradual and cautious.
Potentially, a competitive built-in graphics core from Intel may pose some threat to entry-level offerings from companies like ATI Technologies, Nvidia Corp. and S3 Graphics Inc., even though generally Intel has not been known for providing smooth framerates in games. Usually those, who use the low-eng graphics cards do not care much about the best performance and quality, but about the price. In case Intel manages to offer speed similar to graphics cards which cost $50 - $79, graphics chip companies may lose a substantial part of the market.
Intel Corp. did not comment on the news-story.