Field-Programmable Gate Arrays: Reconfigurable Logic for Rapid Prototyping and Implementation of Digital Systems

3.2   CHANNEL-TYPE FIELD-PROGRAMMABLE GATE ARRAYS

The first class of FPGA devices represents a straightforward attempt to bring the advantages of field programmability to gate array technology, by supporting an essentially unchanged design paradigm (netlist of library parts) from early PCB technology, and hence appeal to practicing systems engineers. The two key commercial examples of this class of device are the Xilinx and Actel FPGAs.

3.2.1   Distinguishing Architectural Features

The distinguishing architectural features of today's FPGAs are:

1. Gate-array-like wiring channel and logic block structure: logic and communication resources are clearly separated and distinct. This is reflected in the CAD system where placement of logic blocks on the logic array and routing between them are treated as separate phases of design.
2. Support for bidirectional and multiple source wires. Bidirectional wires are problematic in RAM programmed structures because the fundamental switching elements have a high impedance and intermediate buffering is required. They are, however, commonplace in PCB and gate array design styles where they are efficiently supported primitives. Designers of FPGAs include them in their architectures to maintain continuity with previous technology.
3. Function units with many inputs. Large function units with many inputs are provided which can implement entire blocks of logic. Making the logic blocks more complex than those in conventional gate arrays reduces the number required ...