ATI has launched recently its new chip family, Radeon X1900, formerly known as R580. It is based on Radeon X1800 core, but brings a few new features that are worth mentioning. In this article we will explore the main differences between Radeon X1900 and Radeon X1800 families.
As stated, Radeon X1900 is based on Radeon X1800 architecture. To better understand Radeon X1800 family, please read our article ATI Radeon X1000 Series.
While Radeon X1800 chips have 16 pixel shader processors, Radeon X1900 brings 48 pixel shader processor. Yes, you read it right. It has three times more pixel shader processors than Radeon X1800 and two times more than NVIDIA’s GeForce 7800 GTX.
The architecture of this new chip family you can see in Figure 1.
Figure 1: Radeon X1900 architecture.
As you can see, the only difference between this architecture and Radeon X1800’s is the number of pixel shader cores (in orange). Each “quad core”, as ATI calls them, holds four pixel shader processors. So you will see 12 of these quad cores in Figure 1.
The architecture of each Radeon X1900 pixel shader processor is the same as Radeon X1800, as you can see in Figure 2.
Figure 2: Inside a Radeon X1900 pixel shader unit.
Fetch4 Shadow Map Acceleration
One new feature from Radeon X1900 is what ATI calls Fetch4, a technology to accelerate shadow maps. Shadow map is a technique to create more realistic shadow effects.
From ATI technical paper: “This method of rendering shadows works by first rendering the scene from the point of view of a shadow-casting light source. The results are not displayed, but instead stored in a special shadow map texture where each value represents the distance of the nearest object to the light source. The scene is then rendered from the standard viewpoint, and each pixel is checked against the shadow map to determine if there are any objects between it and the light source. If the result is true, the pixel is in shadow and can be darkened, otherwise it is lit normally”.
Fetch4 technology allows the GPU to sample simultaneously four values from adjacent areas when dealing with shadow maps, thus increasing texture sampling rate by a factor of four when processing this kind of texture.
Another difference between Radeon X1900 and Radeon X1800 is that X1900 has 50% more on-chip memory for hierarchical Z than X1800, which should improve performance on games played in very high resolutions such as 1920×1200, 2048×1536 and 2560×1600.
Models
So far two models were released: Radeon X1900 XT, working at 550 MHz and accessing its memory at 1.45 GHz (46.4 GB/s transfer rate), and Radeon X1900 XTX, working at 650 MHz and accessing its memory at 1.55 GHz (49.6 GB/s transfer rate). Both access their memories at 256-bit rate.
Summary
Radeon X1900 = Radeon X1800 + more pixel shader processors + Fetch4 shadow map acceleration + 50% more on-chip memory for hierarchical Z.
Leave a Reply