Explain the loader, instance, physical device, logical device, and queue using evidence from your actual Linux machine.
This chapter’s project
A device report, a small selection policy, and one clear unsupported-configuration diagnostic.
A softer light for long learning sessions.
Loading your preference…
Light uses Latte. Your choices stay in this browser.
Volume 05 · 16 flexible weeks
Make every dependency visible.
Learn Vulkan through a small explicit renderer: devices, queues, presentation, memory, descriptors, pipelines, synchronization, compute, and evidence from a captured frame.
Open my weekly workbench ↗Build and debug an explicit rendering path with documented feature requirements and correct resource retirement.
Your sketchbook / 8 chapters
Two weeks per chapter: understand and build, then investigate and finish. Repeat any week.
Explain the loader, instance, physical device, logical device, and queue using evidence from your actual Linux machine.
A device report, a small selection policy, and one clear unsupported-configuration diagnostic.
Build a presentation lifecycle that distinguishes frame slots, acquired images, and window extent.
A three-color window with an event/result log and a resize/minimize walkthrough.
Trace a mesh byte from a CPU owner through an upload allocation into a GPU buffer, including completion and alignment.
A mesh upload diagram, an alignment test, and a log showing when staging storage becomes reusable.
Connect host layout, SPIR-V interfaces, descriptor bindings, and pipeline state without relying on accidental compatibility.
A small image, a host/shader layout table, and an explained diagnostic.
Derive a buffer dependency from producer and consumer accesses, then retire resources only after proven completion.
A resource-use diagram, one correct compute-to-graphics dependency, and a bounded retirement log.
Build a depth-tested textured scene while tracking image subresources, layouts, and attachment formats.
An overlap-order test, an image-state table, and a resize-safe textured scene.
Implement a predictable compute update with explicit dispatch bounds, read/write separation, and graphics consumption.
A deterministic seed study, tail-case result, and compute-to-draw resource diagram.
Finish a small original installation with a reproducible capture, tested failure paths, and an honest performance story.
A runnable Linux project, short demo, annotated capture, measured experiment, and honest case study.
Keep your hands on the work
The companion explains its build steps, tested behavior, and remaining learner work. Keep your own contribution clear as the project grows.