2025-11-11

Todo

• Fix the github runners for the knowledge base ✅ 2025-11-11
• Take STAB52 notes ✅ 2025-11-11
• CSCC10 discussion ✅ 2025-11-11
• Develop STAB52 cheat sheet ✅ 2025-11-11
• Fix Invidious, diagnose if ip banned
• Review verilog items for tomorrow workshop ✅ 2025-11-11
• Change Homelab CICD for zettelkasten to run twice a day, instead of on push ✅ 2025-11-11
• Review Computer Graphics ✅ 2025-11-12

Thoughts

• OpenCL TLDR:
  • You have these things called hosts that are able to compute a full OpenCL program
  • You subdivide a host with a memory model for units of work
  • A work item is the smallest unit
  • A work group is a collection of work items
  • A kernel is responsible for doing the tasks of a work group, and kernels can run in paralell
  • A context is used to manage kernels, and the context has its own memory
  • The context issues commands to kernels using a command queue
  • The command queue has in order or out of order queue types, but they are all equivalent after Synchronization
  • There is a program object which is the final executable that is created during the program
  • The program object contains references to command queue and context
  • The final OpenCL program compiles and emulates the program object with GPU bindings