Context Navigation

2022

Timestamp:: Feb 26, 2022, 5:09:32 PM (5 years ago)
Author:: Pierre-Anthony Lemieux
Comment:: --

Legend:

: Unmodified
: Added
: Removed
: Modified

SponsoringPrograms/GSoC/2022

-              v16
+              v17
 '''Qualification Task:''' Add support for detecting Part 15 code blocks to the native JPEG 2000 decoder so that it can fail gracefully.
+'''Mentor:''' pal@palemieux.com
+'''Mentor:''' Pierre-Anthony Lemieux (pal@palemieux.com)
+'''Backup Mentor:''' Chris Hafey (cehafey@amazon.com)
+'''Duration:''' 350 hours
+== GPU (CUDA) accelerate common software video filters ==
+'''Description:''' FFmpeg already has a couple CUDA based video filters, but for a lot of commonly used functionality you have to fall back to software processing. A lot of existing software filters can be ported to CUDA more or less directly. Or completely new ones can be implemented.
+'''Expected results:''' Port or create from scratch one or multiple CUDA based video filters. The number of filters entirely depends on the difficulty and complexity of the filters chosen.
+'''Prerequisites:''' Access to the necessary Hardware. Basic understanding of git and Makefiles. Intermediate to Good C skills and at least a basic understanding of C++. Understanding of filtering algorithms and how shaders/CUDA kernels operate on GPUs.
+'''Difficulty:''' Medium
+'''Qualification Task:''' Port a software filter of your choice to CUDA. Does not need to be fully featured(no extensive pixel format support or options).
+'''Mentor:''' Timo Rothenpieler (''BtbN'' in #ffmpeg-devel on Libera.Chat IRC, timo@rothenpieler.org)
 '''Backup Mentor:''' TBD
+'''Duration:''' 175
+== FATE Tests for Hardware(CUDA/NVDEC)-Accelerated filters/decoders  ==
+'''Description:''' As of right now, no tests exists for any of the CUDA/NVDEC accelerated filters and decoders. This is quite unfortunate, since they can end up with unintended changes in behavior just like anything else.
+'''Expected results:''' Write FATE tests that cover one or several of the CUDA/NVDEC based filters and/or decoders, depending on complexity and the necessary initial boilerplate. The focus on which decoders/filters to cover is free to decide, since creating the initial boilerplate to run those tests will be a large part of the project.
+'''Prerequisites:''' Access to the necessary Hardware. Basic understanding of git, good familiarity with Makefiles and C.
+'''Difficulty:''' Medium
+'''Qualification Task:''' Outline how testing Hardware-Accelerated filters could be tested and implement initial boilerplate and at least one simple test.
+'''Mentor:''' Timo Rothenpieler (''BtbN'' in #ffmpeg-devel on Libera.Chat IRC, timo@rothenpieler.org)
+'''Backup Mentor:''' TBD
+'''Duration:''' 175
+----
+= Unmentored Projects =
+This is a list of projects that students are encouraged to consider if a mentored project is unavailable or not within the students skill or interests. The student will have to find a mentor for the project. A student can also [#YourOwnProjectIdea propose their own project].
+== Checkasm ==
+'''Description:''' Assembly functions in FFmpeg are not unit tested like they are in x264 or dav1d.
+'''Expected results:''' Add tests for most asm functions which are not tested. Less if this is 175 hour duration.
+'''Prerequisites:''' Good C coding skills, basic familiarity with Git. Some x86 assembly knowledge can be advantageous.
+'''Difficulty:''' Easy
+'''Qualification Task:''' Add tests for 2 separate asm functions, and pass review on the ffmpeg-devel mailing list. To pass qualification the code must be accepted into git master.
+'''Mentor:''' tbc
+'''Backup Mentor:''' tbc
+'''Duration:''' 175 or 350 hours
+== ffcall ==
+'''Description:''' Simple tool which shows how to use libavcodec & libavformat for creating a 2 way video conference tool.
+'''Expected results:''' You will write from scratch a tool which allows 2 or more people to communicate with video and audio across the network. The tool can reuse parts from the codebase like ffplay or other input/output related code. The tool also has to use strong end to end encryption.
+'''Prerequisites:''' Good C coding skills, basic familiarity with Git. Experience with writing software communicating over the network.
+'''Difficulty:''' Hard
+'''Qualification Task:''' Write a one way audio or video (not both needed) communication tool with libavcodec & libavformat. The tool must pass code review for the qualification to pass.
+'''Mentor:''' tbc
+'''Backup Mentor:''' tbc
 '''Duration:''' 350 hours
+== GPU (CUDA) accelerate common software video filters ==
+'''Description:''' FFmpeg already has a couple CUDA based video filters, but for a lot of commonly used functionality you have to fall back to software processing. A lot of existing software filters can be ported to CUDA more or less directly. Or completely new ones can be implemented.
+'''Expected results:''' Port or create from scratch one or multiple CUDA based video filters. The number of filters entirely depends on the difficulty and complexity of the filters chosen.
+'''Prerequisites:''' Access to the necessary Hardware. Basic understanding of git and Makefiles. Intermediate to Good C skills and at least a basic understanding of C++. Understanding of filtering algorithms and how shaders/CUDA kernels operate on GPUs.
+'''Difficulty:''' Medium
+'''Qualification Task:''' Port a software filter of your choice to CUDA. Does not need to be fully featured(no extensive pixel format support or options).
+'''Mentor:''' Timo Rothenpieler (''BtbN'' in #ffmpeg-devel on Libera.Chat IRC, timo@rothenpieler.org)
+'''Backup Mentor:''' TBD
+'''Duration:''' 175
+== FATE Tests for Hardware(CUDA/NVDEC)-Accelerated filters/decoders  ==
+'''Description:''' As of right now, no tests exists for any of the CUDA/NVDEC accelerated filters and decoders. This is quite unfortunate, since they can end up with unintended changes in behavior just like anything else.
+'''Expected results:''' Write FATE tests that cover one or several of the CUDA/NVDEC based filters and/or decoders, depending on complexity and the necessary initial boilerplate. The focus on which decoders/filters to cover is free to decide, since creating the initial boilerplate to run those tests will be a large part of the project.
+'''Prerequisites:''' Access to the necessary Hardware. Basic understanding of git, good familiarity with Makefiles and C.
+'''Difficulty:''' Medium
+'''Qualification Task:''' Outline how testing Hardware-Accelerated filters could be tested and implement initial boilerplate and at least one simple test.
+'''Mentor:''' Timo Rothenpieler (''BtbN'' in #ffmpeg-devel on Libera.Chat IRC, timo@rothenpieler.org)
+'''Backup Mentor:''' TBD
+'''Duration:''' 175
+----
+= Unmentored Projects =
+This is a list of projects that students are encouraged to consider if a mentored project is unavailable or not within the students skill or interests. The student will have to find a mentor for the project. A student can also [#YourOwnProjectIdea propose their own project].
+== Checkasm ==
+'''Description:''' Assembly functions in FFmpeg are not unit tested like they are in x264 or dav1d.
+'''Expected results:''' Add tests for most asm functions which are not tested. Less if this is 175 hour duration.
+'''Prerequisites:''' Good C coding skills, basic familiarity with Git. Some x86 assembly knowledge can be advantageous.
+'''Difficulty:''' Easy
+'''Qualification Task:''' Add tests for 2 separate asm functions, and pass review on the ffmpeg-devel mailing list. To pass qualification the code must be accepted into git master.
+'''Mentor:''' tbc
+'''Backup Mentor:''' tbc
+== Template Project ==
+'''Description:''' Template Description
+'''Expected results:''' Template results
+'''Prerequisites:''' Template prequisites
+'''Difficulty:'''
+'''Qualification Task:''' Template Qualification Task
+'''Mentor:''' someone @ somehwere.net
+'''Backup Mentor:''' someoneelse @ somewhere.net
 '''Duration:''' 175 or 350 hours
-== ffcall ==
-'''Description:''' Simple tool which shows how to use libavcodec & libavformat for creating a 2 way video conference tool.
-'''Expected results:''' You will write from scratch a tool which allows 2 or more people to communicate with video and audio across the network. The tool can reuse parts from the codebase like ffplay or other input/output related code. The tool also has to use strong end to end encryption.
-'''Prerequisites:''' Good C coding skills, basic familiarity with Git. Experience with writing software communicating over the network.
-'''Difficulty:''' Hard
-'''Qualification Task:''' Write a one way audio or video (not both needed) communication tool with libavcodec & libavformat. The tool must pass code review for the qualification to pass.
-'''Mentor:''' tbc
-'''Backup Mentor:''' tbc
-'''Duration:''' 350 hours
-== Template Project ==
-'''Description:''' Template Description
-'''Expected results:''' Template results
-'''Prerequisites:''' Template prequisites
-'''Difficulty:'''
-'''Qualification Task:''' Template Qualification Task
-'''Mentor:''' someone @ somehwere.net
-'''Backup Mentor:''' someoneelse @ somewhere.net
-'''Duration:''' 175 or 350 hours
 ----