Introduction

One of the traditional bottlenecks of 3D graphics hardware has been the rate at which pixels can be rendered into a frame buffer using conventional DRAM or VRAM. The 3D-RAM emerged from a complete rethinking of frame buffer technology and produces an order of magnitude increase in rendering performance.

The factors responsible for the dramatic overall performance improvement include:

• New Memory Architecture

• 10-Mbits DRAM array supporting 1280 x 1024 x 8 frame buffer

• Four independent, interleaved DRAM banks

• 2048-bit SRAM Pixel Buffer as the cache between DRAM and ALU

• Built-in tile-oriented memory addressing for rendering and scan line-oriented memory addressing for video refresh

• 256-bit global bus connecting DRAM banks and Pixel Buffer

• Flexible dual Video Buffer supporting 85-Hz CRT refresh

• Write Mostly Interface

• On-chip ALU

• Four ROP units supporting 16 raster operations on byte data

• Four Blend units blending the old pixel value with new information

• On-chip hardware acceleration for all OpenGL blending modes(NEW)

• On-chip hardware acceleration for all OpenGL stencil modes (NEW)

• One 32-bit Match Comparator and one 32-bit Magnitude Comparator

• Concurrent operations of DRAM, Pixel Buffer, ALU and Video Buffer

• 32-bit synchronous high-bandwidth data bus interface with rendering controller

• Blending operations in both (8, 8, 8, 8) and (4, 4, 4, 4) color modes (NEW)

• One additional PASS_IN pin for flexible bit plane organization