Hi,
I used CCS to compile H.264 encoder and used the profile function to evaluate the cycles.
Then I found the "C64/C64+ Compiler Optimization Tricks" document, it introduced another method "benchmarking" to report cycle counts.
When I used both two to evaluate the H.264 encoder on the EVM board, they had different results.
The cycle counts CCS profile showed are much faster than benchmarking method's.
But when I used HDVICP parallel skill to speed up H.264 encoder,
CCS profile show no improvement but benchmarking method shows cycle counts are really decresed.
So , Which one is more precise on EVM evaluation?
Thank you!
