Video delay when using our own lookup table

Dears,

I am using DM368 ipnc.

When I use our lookup tabel to move YUV data,

I find the cpu loading will up to 9x% and make it delay obviously.

My method is as follow:

1. In VideoSwosdThr.c->VIDEO_swosdTskRun()

I get YUV data from swosdRunPrm.videoInOutAddr,

And memcpy it to a temp buff array (it make cpu loading  from 1x% to 5x%)

Then use our own lookup table to move YUV data in temp buff

Final, memcpy it to original virtual address (it make cpu loading up to 9x%)

2. I create a new thread to do lookup table (such as VNF thread)

but it still useless.

Can anyone give me some advise? Thanks a lot.