Vhdl jpeg decoder

Ok. You should sure that you saturated in c code optimization) no any suggestions

If as you mentioned and you use appropriate data types and computing based on tabulated values

You can verify according the standard optimization for loops, multiplication and even try some techniques for high performance computing cluster

I think if it applicable increasing speedup of algorythm

I read some reports about parallel processing for jpeg ls format

But unfortunately I can't advise ready made for your problem