VLIW processor ar processor that can execute several INSTRUCCTIONS AT ONCE .but relay a lot on the SCHEDULING by the compiler .(The number of instructions that can be align to be executed) .So this type of machines are better in application where there are a lot of repetitive TASKS .. like DSPs of course.
But it fails dramatically on algorithms where there is a lot of BRANCHING decisions .!
regular dps like the ones implemented in some open sites like opencores are very simple and don't relay on multiple execution UNITS ..a project like that will be MAJOR .
yeah i agree with elton.. VLIW is a failure commercially.. but i want to design a low power version of this processor under Globally asynchronous and locally synchronous (GALS) technology.
Listen FRIEND this was my FIELD of expertise .i worked in this area for 10 years ..
And still i'm always intersted in this AREA .. Even last year with the Event of low priced XILINX powerfull VIRTEX FPGA some of these projects COULD become FEASABLE
But .They require so much STUDY .. I'll give you an ADVICE .. why don't you start with SISTOLIC architectures .. there are plenty of applications and they are much easier to IMPLEMENT.
Get Systolic parallel PROCESSINg by N.Petkov .. is a great COMPILATION of hardwired systolic alorithms !
But i loved so much this area that i'm always ready for a good fight!
Cheers