DVRRDK question on allocating buffer in DSP

I want to create a buffer with size of the frame in DSP, so I can use it for processing and temporary storage.

I have tried following the SCD module using Utils_memAlloc(frame_size,  SharedRegion_getCacheLineSize(SYSTEM_IPC_SR_CACHED) ); to allocate buffer.

but then it is cached so I have to call ti_sysbios_family_c64p_Cache_inv and ti_sysbios_family_c64p_Cache_wb which is seems to use up a lot of DSP loading?

I tried just calling the Cache_inv and Canche_wb without other processing in between and the DSP loading increase by 30%. is this normal? I am passing 16ch D1 + 16ch CIF to DSP but only the D1 channels would call the Cache_inv and Canche_wb during my test.

or I should allocate a buffer from some where else? Thanks.