National Semiconductor Embedded Microcontrollers

The CompactRISC architecture was created from the ground up as a scalable architecture, covering the 16-bit embedded processor domain, while providing the best of RISC (Reduced Instruction Set Computer) and CISC (Complex Instruction Set Computer) architectures. The CompactRISC architecture is enhanced with some typical CISC features, like variable instruction set and direct memory bit manipulation to make it even more effective for embedded applications. A CompactRISC core delivers high performance while maintaining low power consumption, small die size and high code density. Through its modular bus architecture the CompactRISC technology allows designers to build performance-tailored functionality around the CompactRISC core. A separate bus for the on-chip peripheral allows a system-on-a-chip designer to use the same on-chip peripherals and I/O with all CompactRISC derivatives. With common processor architecture and software development tools a software designer can feel comfortable to write identical code for any of the CompactRISC cores.  Key Characteristics 

• Reduced Power: Suitable for portable applications    
• High Performance: Delivering the right performance to meet most embedded application needs    
• Small Code/Data Size: Avoiding the typical code/data expansion of RISC processors    
• Minimal Die Size: Freeing space for on-chip application-specific peripherals and memory    
• Portability & Scalability: Common architecture and development tools for all CompactRISC derivatives    
• Modular bus architecture: Sharability of peripherals between all CompactRISC derivatives    
• Debugging Features: On-chip debugging support    
• Multi-core Solution : A proven system-on-a-chip integration of multiple RISC and DSP    
• Licensing: Available in Synthesizable Verilog HDL 

Traditional RISC architectures are driven by the need to squeeze every bit of execution time from a fast system clock, hence typically trading silicon efficiency for instruction execution efficiency. Overlapping pipelines and highly parallel concurrent logic achieve this goal, but at a cost: far more on-chip transistors and greatly increased die size, at a relatively small performance gain. Because the performance needs of embedded controllers are quite different from those of workstation MPUs, the pressure to squeeze clock performance through every available design trick is lessened. With fewer pipeline stages (just three compared to five or more) the CompactRISC architecture eliminates a large amount of costly pipeline control logic. This shorter pipeline organization allows elimination of branch prediction mechanisms and bypass registers, yet the CompactRISC architecture maintains an acceptable overall performance level for embedded system needs. 

Traditional RISC systems have also utilized large off-chip memory and I/O, so on-chip bus interface units were large and complex. Because embedded systems typically use on-chip memory wherever possible, the CompactRISC architecture was designed with a minimal core bus controller (CBC) for on-chip memories. A bus interface unit (BIU) for external memory accesses is added only when off-chip memory is needed. A separate peripheral bus controller (PBC) is optimized for on-chip I/O. This approach yields maximum resource flexibility with minimum silicon overhead. 

National has now created a second generation 16-bit implementation to provide expanded options for system designers. The new implementation provides for: 

• Expanded program address space of 2 Mbytes   
• Atomic Memory Direct Bit Manipulation of single bits to efficiently handle semaphores, I/O triggers, etc.   
• Multiple-register save and restore instructions   
• Multiple push and pop instructions   
• Addition of a hardware multiplier unit for fast 16-bit multiplication   
• On chip debug features, including hardware breakpoint support   
• Synthesizable Verilog HDL format.

The configuration resgister (CFG) holds control bits, determining the mode in which the core operates. Specifically for the new 16-bit implementation, it controls the dispatch table entry size, allowing for small/large memory-model programmability. The processor status register (PSR) holds status information and determines operating modes for the processor. Bits in this register serve as flags indicating the results of intermediate and final ALU operations such as carry bits, zero indications, and negative comparison bits. Also included are interrupt masking and enabling bits and trace bits for debugging. Processors designed with the CompactRISC architecture include hardware for a variety of debugging and self-test features, as well as hooks for external ISE equipment for tracing instruction execution and pipeline sequencing. 

Three debug-specific registers (DCR, DSR, CAR) control the CompactRISC on-chip debug module, allowing breakpoints on instruction fetch, data-read and data-write. For more information, see the Debug Support below.

The CompactRISC architecture provides on-chip, hardware breakpoint instruction tracing, soft breakpoints via breakpoint instructions and external ISE support. 

Three debug-sepcific register control the hardware debug module. In addition for the new 16-bit implementation, there is enhanced external instruction fetch breakpoint support, allowing breakpoints on instruction fetch or data read/write according to address-compare. In general, the debug control register (DCR) controls the various debug feature activation, the debug status register (DSR) holds the indication bits, identifying the source of breakpoint. The compare-address register (CAR) contains the address to be matched for breakpoint by the debug module.Instruction tracing is used during debugging to single-step through a program. Two bits in the PSR enable and generate trace traps. In addition, a breakpoint instruction (EXCP BPT) can be used to stop the execution of a program at specified instructions, allowing the debugger to examine the status of the program and internal registers. The CompactRISC architecture also supports an input pin that causes an ISE interrupt. The core processor then provides status signals that are activated upon completing an instruction and, in some of the cores, also has on-chip hardware breakpoint support for data and address values. In addition, National Semiconductor's incorporation of JTAG interfaces allows for efficient design and debug of multiple-core implementations, such as two CompactRISC cores, integration of DSP cores, etc. within a single die. 

Key RISC advantages include a streamlined instruction set optimized HLL compilers; single-cycle instruction execution; and fast, multistage pipeline structures. To this, the CompactRISC processors add compact code generation, efficient RAM utilization, low-cost silicon implementations and a scalable architecture that spans the range from 8-bit, to 16-bit, 32-bit and 64-bit versions. 

Over the past three years, the CompactRISC architecture has firmly established itself by filling a previously unmet market gap -- those applications that require the performance of RISC but cannot afford the processing and cost overheads of 32-bit RISC implementations. The 16-bit member of the CompactRISC family has been particularly popular with designers because of its optimal balance of cost and performance, plus the ability to combine its small core die size with other key on-chip functions. With the additional features included in the new 16-bit CompactRISC processors, designers can now take advantage of an expanded code space and even more features targeted specifically for embedded implementations. Plus, with built-in scalability, the migration path to 32-bit implementations, using the same basic CompactRISC architecture, is always open.