when using the cic filter as decimation filter,is a compesation filter needed?
the basic cic structure is composed of two section,integrator section and comb section.I found that the response of integrators for dc input is without bound. so does that imply we can not use the structure in practice?
Yes, quite right. This behaviour is a principle property of CIC filters that utilizes modulus arithmetics. I think, the topic is addressed in any serious CIC literature, you may want to read it somewhat more thoroughly. You get a well defined DC response from the basic CIC structure anyway.
Compensation filters are used if you want a flat frequency characteristics rather than CIC typical (sin(x)/x)^n behaviour. They are present e.g. in today's audio SD converters.