Last month's RADEON 9700 launch sent shockwaves throughout the entire 3D industry. Not only was ATI first to market with a next generation DirectX 9.0 product, ATI's RADEON 9700 also boasted many features that were exponentially greater than any product that was either currently available on the market or had even been announced on paper.
Clearly everyone in the hardware world was eager to see how RADEON 9700 stacked up to the competition from NVIDIA, but at the time ATI was unwilling to oblige. Well folks, today that day has come! We're going to deviate from the norm and dive straight into the specs (as well as our thoughts on them), as we have quite a bit of information to cover. This article will stick to the technology behind RADEON 9700 and its performance, which we'll also be covering in additional detail in subsequent articles. Lets take a look at what makes the RADEON 9700 PRO such a groundbreaking product:
3D Graphics Features
Eight parallel rendering pipelines process up to 2.6 billion pixels per second
Four parallel geometry engines process up to 325 million transformed and lit polygons per second
High precision 10-bit per channel framebuffer support
Full support for Microsoft DirectX 9.0 programmable pixel and vertex shaders in hardware (complete feature set also supported in OpenGL via extensions)
2.0 Pixel Shaders support up to 16 textures per rendering pass
2.0 Vertex Shaders support vertex programs up to 1024 instructions with flow control
New 128-bit per pixel floating point color formats
Shadow volume rendering acceleration
New technology processes up to 15.6 billion anti-aliased samples per second for unprecedented performance
Supports 2x, 4x, and 6x anti-aliasing modes with programmable sample patterns
Advanced anisotropic filtering
Supports up to 16 bilinear samples (in performance mode) or trilinear samples (in quality mode) per pixel
Hierarchical Z-Buffer and Early Z Test reduce overdraw by detecting and discarding hidden pixels
Loss-less Z-Buffer Compression and Fast Z-Buffer Clear reduce memory bandwidth consumption by over 50%
2nd generation Npatch higher order surface support
Discrete and continuous tessellation levels per polygon for dynamic LOD
DirectX 9.0 displacement mapping
Seamless integration of programmable pixel shaders with video data
Integrated MPEG-2 decode
Back-end scalar delivers top quality playback
Dual integrated display controllers drive two displays simultaneously with independent resolutions and refresh rates
Dual integrated 10-bit per channel palette DACs operating at up to 400MHz
Integrated 165MHz TMDS transmitter supports resolutions up to QXGA (2048x1536) and complies with DVI and HDCP specifications
Integrated TV-Out support up to 1024x768 resolution
Comprehensive AGP 2X, 4X, and 8X AGP support
High performance quad-channel DDR memory interface supports 64/128/256MB configurations
110 million transistors
0.15-micron manufacturing process
325MHz core clock/310MHz memory (effectively 620MHz)
The features list of the RADEON 9700 PRO is very impressive: 110 million transistors built on a 0.15-micron manufacturing process, making the RADEON 9700 PRO VPU the most complicated graphics processor ever created. In fact, RADEON 9700 PRO sports twice as many transistors as Intel's "Northwood" Pentium 4 processor (55 million), and while it's built on a 0.15-micron process that is similar to the one used for GeForce4 Titanium (which contains 63 million transistors) it runs a core clock frequency of 325MHz, 25MHz faster than GeForce4.
Clearly ATI went to great lengths to make sure that not only was the fundamental architecture more advanced than anything currently available, but so was its manufacturing implementation. We also believe that RADEON 9700 PRO's flip-chip packaging (which has been used on CPUs for years) plays a role here as well. Flip-chip packaging allows to core of the processor to come in direct contact with the heatsink, resulting in more efficient cooling.
As you can see, the RADEON 9700 PRO supports up to 256MB of memory, but for now all RADEON 9700 PRO cards will ship in the 128MB configuration. With its 256-bit interface (twice that of GeForce4 Titanium), it provides up to 19.84GB/sec of bandwidth. In comparison, GeForce4 Ti 4600's Lightspeed Memory Architecture supplies up to 10.4GB/sec to the GPU.
Over the course of the next several pages, we'll go over all of these features in much greater detail. We'll get started by discussing RADEON 9700's SMARTSHADER 2.0 technology.