Mobility Radeon от ATI
There is nothing surprising in the fact that the new Mobility Radeon is developed on the basis of the latest ATI graphics chip for desktop systems – Radeon VE, which, in turn, is a stripped-down version of Radeon – a professional chip for desktop systems. This is the normal way to give birth to a mobile graphics chip. Moreover, in general, Radeon VE also has good characteristics, and therefore is used in video cards for desktop systemsThe video card supports the AGP 2x / 4x interface, has 32 MB of DDR SDRAM, which consists of 4 chips that are located on the front side of the circuit board around the graphics chip (operating at a clock frequency of 183 (366) MHz). The graphics chip is closed by a radiator glued to it.
There is no fan, and, as ATI herself notes, “it is not needed, since Radeon VE heats up very little.” On the card, in addition to the usual VGA connector, there is also a DVI output. Both of these outputs, together with the digital-to-analog converters (RAMDAC) integrated in the chip, form the basis of the HydraVision technology.
The Radeon VE graphics chip includes only one pixel pipeline, which has three texture modules, and this allows you to calculate only 3 texels per clock cycle, which is half as much as the full version. Three texture modules of the graphics core will be used extensively in future games, but in the games currently existing it is not so important how many texture modules a graphics chip has (pixel pipelines of other Nvidia and Matrox graphics chips consist of two texture units), and on performance it does not affect. The fact is that modern games (Quake III Arena, Unreal Tournament, Soldier of Fortune, Heavy Metal: F.A.K.K. 2, Flying Heroes) use two texture modules, therefore, only two of the three modules will be involved in Radeon VE. For example, when rendering a scene in the popular Quake III Arena game today, the Radeon VE core will only process two texels per cycle, since the additional texture module is simply not used (there is no algorithm that would use the third texture block in the calculations). ATI, releasing the Radeon accelerator and its modifications, hopes that game developers will write new algorithms to support the third texture module, which will ultimately lead to increased productivity and improved rendering quality of scenes in 3D graphics. Id Software founder John Carmack and many other OpenGL programmers (Brian Hook, John Cash) support the idea of using additional texture in games. But for now, the fact is that all modern games use only two of the three Radeon VE texturing modules of the three, and this, of course, does not allow the chip to show its maximum performance.
In Radeon VE, there is no Charisma unit – the trademark name for the geometric coprocessor, which is in the full version of Radeon. Excluding the Charisma block, ATI essentially deprived Radeon VE of T&L support. And this is against the backdrop of the ATI statement made seven months ago that their T&L unit outperforms similar competitors.
The biggest disappointment due to the lack of a T&L unit in Radeon VE is the inability to transfer coordinate transformation and lighting calculations from the central processor of the system to a specialized geometric processor, which would free up CPU resources for calculating the physics of object movement or artificial intelligence of bots. Transferring the coordinate transformation and lighting calculations to a specialized graphics processor would undoubtedly affect a decrease in the flow of data transmitted from the central processor, which would lead to less CPU power consumption. Currently, T&L support does not affect the concept of the game, but it seems that in the near future, with the advent of new games, the situation will change. Nevertheless, now the absence of the T&L unit has practically no effect on the promotion of Radeon VE and even reduces the energy consumption of the graphics chip. The situation can only change with the release of a new generation of games and applications in which T&L support will already be a prerequisite.
But the developers also added blocks, the appearance of which will be gladly received by users. These modules allow you to output images to a second receiver (a second RAMDAC and a block for converting an analog signal to digital form have been added).
ATI’s developers also included tile-based HyperZ technology in the Radeon and Radeon VE cores (splitting the screen into square fragments). The ATI chip that uses this technology draws the polygon in the usual manner, and then in tile mode, and when the tile closes the polygon, the latter is discarded and does not take part in rendering in the future. According to ATI, this technology saves 20% of the time when rendering game scenes.