Soft Cores

We briefly examined the Xilinx MicroBlaze and Altium’s TSK3000A soft cores. Neither were supplied in a format that would allow modifications. Many years before we watched other collegues configure Altera’s NIOS, but never invested in Altera tools ourselves. The soft core experimentation was around 2000 for NIOS, 2005 for MicroBlaze, and 2008 for some tracing of the Altium TSK3000A. None of these cores were particularly attractive without source code. We looked into the OpenCores and other sites for free cores, which we might document elsewhere, but more from a trace and performance viewpoint.

The clock frequencies and expense did not compare very well to SoCs from a range of vendors. In the past five or so years, the levels of integration in Flash, RAM, ARM cores and preipherals have made FPGAs less competitive. Xilinx pre-announced their ARM hardcores so far in advance that they now appear pedestrian compared to the Freescale and Texas Instruments devices that are announced almost at the same time as evaluation boards. We will be using highly integrated cores for intelligent I/O devices and to perhaps load up a FPGA, but the use of a soft core will only happen if we can instrument the core. The investment in any new technology is high, so what can you get from a soft core? We think debug infrastructure would benefit, plus any I/O intensive work. The Lattice Semiconductor Mico32 is going to be our target as it satisfies the source code requirements, plus the tools are reasonably priced.

Lattice Semi EPC3 board

 

Lattice Semiconductor’s Versa board will be used to test the Mico32 core