The nSIM Pro simulator comes with a state-of-the-art JIT compilation engine that leverages the capabilities of multi-core simulation hosts by offloading JIT translation threads from the core that is running the simulation reducing the JIT compilation overhead and significantly speeding up simulation performance. Embedded Microprocessor Benchmark Consortium (EEMBC) benchmarks on a simulation host running at 3.47 GHz show an 18X speed increase, on average, when JIT mode is activated and JIT translation is offloaded to a separate core of the simulation host. As a result, an average performance of 475 MIPS is reached across all 34 tests of the test suite. And over 1,400 MIPS is reached for the EEMBC bezier01 test that benefits most of the JIT activation.
In the case where a full 100% cycle accuracy is needed for final performance tweaking or verification, an ARC xCAM model can be used. These models are derived from the processor's Verilog to achieve 100% cycle accuracy.
| Interpretive | JIT | |
|---|---|---|
| Average speed | 20 MIPS | 475 MIPS |
| Instruction accuracy | 100% | 100% |
| Cycle accuracy | N/A | N/A |
| Software development & debug | YES | YES |
The nSIM processor model is delivered as a single DLL that integrates seamlessly with the MetaWare and GNU debuggers. The nSIM models come with an industry-standard OSCI TLM-2.0 SystemC interface to support their integration with OSCI standard peripheral models. In addition, the nSIM models implement the required interfaces for deployment in the Synopsys Virtualizer platform, a tool for the creation, assembly and execution of SystemC-based virtual prototypes for pre-silicon software development and software-driven verification.
Quickly identify and access the right IP solutions for your project needs.
Find embedded memory and logic IP for your SoC design.
Find silicon-proven NVM IP for your SoC design needs.