Specifications
• Dual programmable Vertex Shaders
• Advanced programmable Pixel Shaders
• nVIDIA Lightspeed Memory Architecture™ II
• nVIDIA Accuview™ Antialiasing
• 3D Textures
• Shadow Buffers
• 4 dual-rendering pipelines
• 8 texels per clock cycle
• Dual cube environment mapping
• 128MB high-speed DDR RAM memory at 550MHz DDR
• High-Definition Video Processor (HDVP)
• AGP 8X with Fast Writes
• AGP 8X / 4X / 2X and AGP Texturing support
• 32-bit color with 32-bit Z/stencil buffer
• Z-correct true, reflective bump mapping
• High-performance 2D rendering engine
• Hardware accelerated real-time shadows
• True-color hardware cursor
• Integrated hardware transform and lighting engine
• High-quality HDTV/DVD playback
• TV-Out and Video Modules
• Multibuffering (double, triple, quad) for smooth animation and video playback
• Microsoft DirectX® and S3TC® texture compression
• 8.8 GB/sec. memory bandwidth (550MHz DDR)
• 125 million triangles/sec. setup engine
• 4.4 billion AA sample/sec. fill rate
• 1.12 trillion operations/sec
Key features in more detail include:
-
nfiniteFX II engine: The NVIDIA nfiniteFX II Engine incorporates dual programmable Vertex Shaders, faster Pixel Shaders and 3D textures. The nfiniteFX II Engine gives developers the freedom to program a virtually infinite number of custom special effects to create true-to-life characters and environments.
-
Accuview Antialiasing (AA): The Accuview Antialiasing subsystem with advanced multisampling hardware delivers full-scene antialiased quality at performance levels never before seen.
-
nView Display Technology: The nView hardware and software technology combination delivers maximum flexibility for multi-display options, and provides unprecedented end-user control of the desktop experience. nView allows end-users to select any combination of multiple displays, including digital flat panels, analog CRTs, and TVs, and to modify the display properties using an intuitive software interface.
The basic specifications show that the revised Ti 4800 SE is just the standard GeForce4 Ti 4400 with the added 8x AGP feature. The Ti 4800, however, is the standard Ti 4600 with added 8x AGP support. So we have the standard Ti 4400 clocks of 275/275 (core/memory) with added 8x AGP support. It would be easy to confuse the Ti 4800 SE for the faster Ti 4800.
We saw the GeForce4 Ti 4200s receive a 8x AGP makeover. More than just a naming change with 8x AGP support, they all shared common attributes of having 128MB of memory and standard memory speeds of 513MHz (up from 444MHz). Nothing other than a faster AGP speed, a feature that won't show its worth for a while yet as we don't saturate the current 1.06GB/s 4x bus, has changed from the incumbent Ti 4400 to the present Ti 4800-SE.
The Ti 4800-SE retains the 4 rendering pipelines, two texture units per pipe, 128-bit memory interface (4 x 32-bit memory controllers). Given that the core operates at 275MHz, this should give us a peak single-texture fill rate of 1.1GPixels/s and a dual-texture fill rate of 2.2GPixels/s. The 550MHz DDR memory speed equates to a total peak bandwidth figure of 8.8GB/s. However, that bandwidth is soon exposed when we factor in the classic bandwidth-eating duo of texture and edge aliasing. Advanced mutisampling antialiasing and adaptive anisotropic algorithms help to conserve as much bandwidth as possible, but compared to the more modern high-end ATi cards, the application of these image enhancing effects isn't quite as efficient as it could be.
Just to summarise briefly the GeForce4 table.
Card | GPU Clock (MHz) | Memory clock (MHz) | AGP speed |
GeForce4 Ti 4800 /4600- 8x | 300 | 650 | 8x |
GeForce4 Ti 4800-SE | 275 | 550 | 8x |
GeForce4 Ti 4600 | 300 | 650 | 4x |
GeForce4 Ti 4400 | 275 | 550 | 4x |
GeForce4 Ti 4200 (8x) | 250 | 513 | 8x |
GeForce4 Ti 4200 | 250 | 444/500 | 4x |