It seems that the Mico32 is rather large and slow compared to a Nios II.
Here's the stats from the Lattice web site:
Performance and Resource Utilization for LatticeECP2/M Devices1:
Configuration LUTs fMAX (MHz)
Basic 1,571 98
Standard 1,816 116
Full 2,158 116
Three Configurations:
* Basic
o No Multiplier
o Multicycle Shifter
o No Cache
* Standard
o Multiplier
o Pipelined Shifter
o 8K I-Cache, No D-Cache
* Full
o Multiplier
o Pipelined Shifter
o 8K I-Cache, 8K D-Cache
Their "Standard" version is about the size of a Nios II/f but the Nios II/f includes a data cache.
The frequency of all cores is substantially slower than any of the Nios II cores. Maybe that is a property of the Lattice FPGAs? It would be interesting to get an apples to apples comparison to Nios II by implementing a Mico32 on an Altera device.