A quantitative study and estimation models for extensible instructions in embedded processors
Proceedings of the 2004 IEEE/ACM International conference on Computer-aided design
Hi-index | 0.00 |
We present a new approach to estimate speed, area, and power of a parameterizable, soft IP. By running the ASIC implementation flow only on selected configurations, we predict the performance for any arbitrary configuration. We exploit performance function decomposability to address the combinatorial explosion challenge. The estimator has been used successfully to configure Xtensa processor cores for numerous embedded SOC designs.