ATi Radeon X800 XT & X800 Pro - Heart Burn For The NV40

Article Index

ATi R420 Architectural Overview


At the chip level, ATi has given their new R420 VPU architecture a complete overhaul to allow for better scalability and efficiency.  This core VPU is the same engine that drives both the Radeon X800 XT and Radeon X800 Pro series of graphics cards.

Radeon X800 XT and Pro Series Architectural Overview
Scalable Parallel Computing  on a chip


X800 Pro - 3 Quad Pixel Pipelines

X800 XT - 4 Quad Pixel Pipelines

As we noted earlier, with the X800 XT and X800 Pro VPUs, ATi has essentially scaled up their current core architecture to 16 pipe and 12 pipe implementations.  In doing so however, they've also beefed up many areas of the architecture providing significantly higher overall throughput and a few new features as well.



The first thing to note here is the obvious GDDR3 memory interface that ATi has implemented on both the Radeon X800 XT and X800 Pro, running at 1.12GHz and 900MHz for the X800 XT and X800 Pro respectively.  This will enable a theoretical peak memory bandwidth of 35.8GB/sec for the X800 XT and 28.8GB/sec for the X800 Pro.

In addition, the X800's design team added two more Vertex Pipes to the architecture expanding total resources to 6 Vertex Pipelines.  They've also tweaked their pixel shader engines significantly, offering a new type of compression technology, which we'll cover for you shortly.  The new X800s also make use of  "Hyper Z HD", ATi's next generation Z culling and compression algorithms, for solid gains in available memory bandwidth by minimizing overdraw.

Finally, ATi has also enhanced their AA engine coining it "SMOOTHVISION HD".  It brings with it support for DX 9.0c Centroid sample AA and a new type of AA technology called "Temporal AA", that we'll cover for you in the pages ahead as well.

Tags:  ATI, Radeon, ATI Radeon, heart, art, XT, pro, RT, EA, BU, X8, AR, burn

Related content