HDMI 4k Video: Lessons Learned
summary
Built a conferencing codec box with 3 HDMI inputs and 2 outputs and it must handle 4k capture and presentation modes. Key lessons: 600 MHz for 4k60 RGB requires verified cabling and oscilloscope eye diagrams; HDMI quirks broke outputs (vendor firmware misinterpreting blanking no-op packets); EDID random reads by Apple forced FPGA/driver fixes; quantization mismatches (full vs limited range) require advertising YUV420 in EDID to cover all laptops; adapters and cables are poorly labeled and failing Vbus/HPD cause silent failures—test with edid-decode and Linux.
AI-written summary · about
chapters (15)
- 0:00Introduction to Cisco Norway
- 2:27HDMI connector technicalities
- 5:28Understanding 4K resolutions
- 8:534K data rates and challenges
- 12:19Signal integrity and testing
- 14:02Protocol layer issues
- 16:52Status and control data
- 18:02Quantization range explained
- 22:56Debugging HDMI issues
- 23:53Laptop behavior differences
- 26:574K EDID and timing issues
- 28:44Sink driver reliability
- 29:44Adapter and cable problems
- 33:37Certification and labeling
- 36:00Future trends and Q&A
related talks
45mImproving the Linux Display Stack Reliability
42mExtended Display Identification Data (EDID): Problems, Pitfalls and Complications
52mHDMI CEC: What? Why? How?
45mV4L2 & CEC Status Report
24mUHD Four-O
37mIntegrating Hardware-accelerated Video Decoding with the Display Stack
33mStateless V4L2 Video Encoding
46mV4L2 Video Decoder/Encoder Development Journey