Reconfigurable logic limitations

There are three major factors that limits the evolution of the RTR applications:

• FPGA gate capacity is some how small compared to the ASICs [11]. This increases the complexity of the system because designers have to use more than one FPGA and they have to partition their designs between the on-board FPGAs. Interfacing problem arises when multiple FPGAs are used which increases the over all system delay.
  These days the number of gates per FPGA is increasing, for example the new FPGA family from Xilinx ``Virtex'' will have 1 million gates per FPGA.

• Configuration speed is the most important factor that limits the number of RTR applications. New FPGA architectures must be developed to reduce the configuration overhead. Several architectures have been proposed such as the DPGA that uses multiple configuration memory syntaxes⁶.

• RTR applications need highly skilled developers who can deal with the low level design, so these applications are developed in laboratories under the supervision of qualified designers [12]. The only solution to this problem is to develop CAD tools that can reduce the design time.

• Benchmarks for RTR applications is one of the important issues that must be introduced to enable designers to discuss the performance of their designs versus the performance of other designs techniques. M. Wirthlin suggested the functional density as a performance measure for RTR applications [16]. It is a measure of the execution time with respect to the circuit area taking into consideration the configuration overhead.

Casselman the president of the Virtual Computer Corporation ``VCC'' wrote in [5] the issues that must be considered in RC applications:

The General purpose RC system must:

• install easily into current computer systems.
• be compatible with existing standards & transparent to the End-User.
• be accessible from within common application program software.
• be programmable by software engineers unfamiliar with digital design.

Three fundamental issues for the successful insertion of RC into marketplace:

• How do we get the large software engineering community to grasp, develop and utilize reconfigurable computing in their applications?
• How do we introduce the reconfigurable computing component into the current and future computing environment effectively, from both the technological and users side?
• How do we establish a fundamental foothold in this evolving market without redesigning or producing the entire computing platform?
  

So his idea is to use the existing computer systems in conjunction with the RC system to get efficient low cost usable system. The application design should be done using general computing software style without going into the details of the underlying architecture.