Hi all,

This forum has been dormant for a few weeks, so I'd like to give a bit of an update about what the OpenCores team have been working on.

I posted saying I was able to get the x264 running in OR1ksim, and I've made some improvements on that - primarily being able to dump the encoded data out of the simulator and view it to verify that all is well.

Since then I've mainly spent time ploughing through the x264 code attempting to understand what it's doing, and how each of the specific data structures work. I feel a comprehensive understanding of the code will help us make a good decision about where to draw a line with regards to offloading the software algorithms to hardware. In particular I'm attempting to understand how the data is formatted and used by x264 during encoding and decoding. I'm focusing on understanding the x264_me_search_ref() function in encoder/me.c and what is involved if we aim to move the entire UMH (uneven-cross multi-hexagon-grid search) algorithm into a hardware module. Starting with the simpler DIA (diamond) search as an example would be the first step.

I will put together a small howto on creating a new or1ksim module, and then hopefully release an example of an or1ksim module which does a DIA search, and how it can be used from the x264 software.

After this perhaps we could look at the next step of offloading more and more from the software in the x264_me_search_ref.

Please post with your thoughts and ideas.

Julius