Hi OC-H.264 team,

This TOPIC is meant to contain information about the "overall progress information", all the deep technical discussion can be discussed under other TOPICS in this forum. Feel free to add new TOPICS.

We are now 28 engineers that have sent in the additionbal information, I'm still waiting for 12 additional engineers to confirm if they are interested to join.

An initial SW-team is now in place and it includes:
Caleb, Chander, Sheng, Wendy, Yevgeny

If anyone wants to help them, let me know, the more the better :-)

One of the first task we need to do is "GCC-profiling of VLC-x264". We basically need to understand what functions are suitable to extract from the VLC-SW and then be implement as HW instead. Depending on the information that the profiling exercise with give us, we then need to discuss the pros/cons of moving different functions to hardware (size/performance issues etc).

So the actual RTL development needs to wait until we have "profiling data" available, so stay tuned!


BR,
Marcus

How about to summarize the status weekly. Like this :


Week
====
Week #1. Aug/21 ~ Aug/26

What we've done
===============
- Opened "OC H.264 project" forum
- Got and compiled VLC x264 on linux
- Got some reference video seq for benchmarking purpose
- Discussing about using Verilog only or mixed Verilog/VHDL, no conlcusion yet
- Discussed about the project direction (target codec). Direction is no change at this moment (H.264)

What we plan to do
==================
- Define the profiling category in terms of tools, machine & input parameters
- Port VLC x264 code to OpenRISC processor using OpenRISC toolchain
- Perform the profiling on various categories
- Analyze the profiling and judge which part of VLC x264 code to be implement in HW

Problem & request
=================
- Need to discuss and define the profiling category
- Need to discuss who will do profiling and porting
- Need to agree a common area to upload project data


Honestly speaking, it is a tedious task to summarize the status weekly like this. But if anyone think that it is useful for the whole development, then I'd love to take such tedious task.


Felix

Hi Guys,
I would like to get my hands on S/W as well. Still need to setup the environment. Any pointers will be helpful.
Thanks very much.

Guru



A repeat of my mail I sent to Marcus earlier, today:


*****************************************************
Hi Marcus,
Would you be able to suggest a template/standard and H/W and S/W requirements for the development environment.
This will be useful to everyone on the project to set it up before things get busy.

Everyone on this project should be familiar with the directory structure, protocols followed for checking in and out, and for finding help.

I'm sure you guys are doing this for over a decade and know very well about how things are done at Opencores. It may be a good time to provide information or links to enable everyone to come up to speed.

Thanks very much.
Guru
*****************************************************

Yes, at least, we should first layout the directory structure and specify which material will go to which directory, make it clear what version of x264 is used.

Thanks.
Brian

yes, Let's disscuse it from now.

Hi Felix and ALL,
Excellent initiative Felix, it's better with "more information" then "to little of information". It's important that all members within the group "sees" and feels that things are progressing. Lake of information always make people think that nothing is happening. So Felix, it would be great if you are willing to "assign yourself" making sure that we get weekly status-reports.


Unfortunately I'm out on a business trip right now, which has limit my time and access to Internet :-(
The group are now quite large and more engineers are "signing up" continuously (which is great), this however require us to put more effort into project-management in order to make sure that we take advantage to all the resources within the group.

So I suggest that we do the following:

1. Assign one or two project "team-coordinator/s" within each sub-group (SW, HW design, HW verification). Then we need to take a conf-call via Skype to align what need to be done.
2. Create a document with the "basic" assumptions, settings, what we aim for, etc. This document makes it easier for all of us to "stay on the same track".

Are there any volunteers for the "team-coordinator/s" task?


Additional status information:
We are currently working on creating a 1:st based system including the OR1200-processor, hopefully we can upload this SoC design in SVN during next week.


BR,
Marcus

Hi all,

I'm a bit busy on this week, so I may not be able to summarize the status report this week, sorry.

Either someone else help to summarize it, or I'll do it together for the coming week activities. Thanks.


Felix.

Week
====
Week #2~4. Aug/27 ~ Sep/19

What we've done
===============
- Agreed to put project data on SVN respo
* http://opencores.org/ocsvn/oc-h264-encoder/oc-h264-encoder/trunk
- Compiled VLC x264 on OpenRISC processor or1ksim and MS Visual C++
* Both need log2f func workaround
- Decided the profiling category
* Version of VLC x264
-- x264-snapshot-20090901-2245.tar.bz2
* Configure VLC x264
--disable-avis-input
--disable-mp4-output
--disable-pthread
--disable-asm
--enable-gprof
* Input parameter when executing VLC x264
--profile baseline
* Machine platform
-- Any
- Checked in the above version compiled VLC x264 binary on SVN respo
* doc/x264_profiling/bin
- Checked in some profiling result on SVN respo
* doc/x264_profiling/gmon_files
- Starting to analyze the profiling result, found some CPU intensive funcs
* x264_pixel_sad_hxw (Sum of Absolute Differences)
* x264_pixel_satd_hxw (Sum of Absolute Transformed Differences)
* x264_me_search_ref (me = motion estimation)
* x264_refine_subpel
* mc_chroma (mc = motion compensation)

What we plan to do
==================
- Continue to perform profiling with the above decided profiling category
- Continue to analyze the profiling result
- Discuss which part of VLC x264 code to be implement in HW
* Need more study on VLC x264 source code

Problem & request
=================
- Need to collect YUV clips common for this project
* Problem is filesize too large to upload
- Need one or two project "team-coordinator/s" within each sub-group
* SW
* HW design
* HW verification