gpu-rendering-vulkan-guide
/gpu-rendering-vulkan-guide
Use when implementing a Vulkan renderer: the concrete Vulkan API for device/queues, device memory + memory types + staging, images/buffers + pipeline barriers and layout transitions, descriptor sets/layouts + bindless, pipelines + pipeline cache + dynamic rendering, timeline semaphores + fences, command pools/buffers, and the swapchain. Triggers on Vk* types and vkCmd*/vkCreate* calls, VkImageMemoryBarrier2, VkDescriptorSet, VkPipeline, VkSemaphore, VkCommandPool, swapchain acquire/present, even when the user doesn't say 'Vulkan'.
- Author
- xonovex ↗
- Repository
- platform ↗
- Licence
- MIT
- Last repository change
- 08.10.2026
Assessment and security
- Files in this package
- 12
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 08.10.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Use when implementing a Vulkan renderer: the concrete Vulkan API for device/queues, device memory + memory types + staging, images/buffers + pipeline barriers and layout transitions, descriptor sets/layouts + bindless, pipelines + pipeline cache + dynamic rendering, timeline semaphores + fences, command pools/buffers, and the swapchain. Triggers on Vk* types and vkCmd*/vkCreate* calls, VkImageMemoryBarrier2, VkDescriptorSet, VkPipeline, VkSemaphore, VkCommandPool, swapchain acquire/present, even when the user doesn't say 'Vulkan'.
Triggers from the description
Original author description
Use when implementing a Vulkan renderer: the concrete Vulkan API for device/queues, device memory + memory types + staging, images/buffers + pipeline barriers and layout transitions, descriptor sets/layouts + bindless, pipelines + pipeline cache + dynamic rendering, timeline semaphores + fences, command pools/buffers, and the swapchain. Triggers on Vk* types and vkCmd*/vkCreate* calls, VkImageMemoryBarrier2, VkDescriptorSet, VkPipeline, VkSemaphore, VkCommandPool, swapchain acquire/present, even when the user doesn't say 'Vulkan'.
How it works
- Requirements
- Setup
- Resources & sync
- Pipelines & binding
- Commands & present
- Gotchas
- Progressive Disclosure
SKILL.md preview
--- name: gpu-rendering-vulkan-guide description: "Use when implementing a Vulkan renderer: the concrete Vulkan API for device/queues, device memory + memory types + staging, images/buffers + pipeline barriers and layout transitions, descriptor sets/layouts + bindless, pipelines + pipeline cache + dynamic rendering, timeline semaphores + fences, command pools/buffers, and the swapchain. Triggers on Vk* types and vkCmd*/vkCreate* calls, VkImageMemoryBarrier2, VkDescriptorSet, VkPipeline, VkSemaphore, VkCommandPool, swapchain acquire/present, even when the user doesn't say 'Vulkan'." --- # Vulkan Rendering Guidelines The concrete Vulkan API for a low-level renderer. This skill is the Vulkan _how_; apply **gpu-rendering-guide** for the _why_ and the API-agnostic architecture (render graphs, binding model, sync model, frames-in-flight, memory strategy). For the general allocator principle behind device memory see memory-management-guide. ## Requirements - Vulkan 1.3 (or 1.2 + the relevant extensions: `VK_KHR_synchronization2`, `VK_KHR_dynamic_rendering`, `VK_EXT_descriptor_indexing`, `VK_KHR_timeline_semaphore`). - Validation layers + synchronization validation on in development; a frame debugger (RenderDoc-class) for capture. ## Setup - **Device & queues** - `VkInstance` → physical device → `VkDevice`; select graphics/compute/transfer queue families, see [references/device-and-queues.md](references/device-and-queues.md) - **Device memory** - `VkPhysicalDeviceMemoryProperties` types/heaps; few `vkAllocateMemory` + sub-allocation; staging, see [references/device-memory.md](references/device-memory.md) ## Resources & sync - **Images/buffers & barriers** - `VkImage`/`VkBuffer` + views, `VkImageMemoryBarrier2` stage/access masks, layout transitions, see [references/resources-and-barriers.md](references/resources-and-barriers.md) - **Synchronization** - `VkSemaphore` (binary + timeline), `VkFence`, pipeline stage/access masks, submit-time waits, see [references/synchronization.md](references/synchronization.md) ## Pipelines & binding - **Descriptors** - `VkDescriptorSetLayout`/`Pool`/`Set`, update-after-bind for bindless, push constants, set frequency, see [references/descriptors.md](references/descriptors.md) - **Pipelines** - `VkPipeline` (graphics/compute), `VkPipelineCache`, dynamic rendering vs render passes, dynamic state, see [references/pipelines.md](references/pipelines.md) ## Commands & present - **Commands & swapchain** - `VkCommandPool` per thread per frame, primary/secondary buffers, `VkSwapchainKHR` acquire/present, fence per frame, see [references/commands-and-swapchain.md](references/commands-and-swapchain.md) ## Gotchas - `vkAllocateMemory` is hard-capped (`maxMemoryAllocationCount`, often ~4096) and slow; sub-allocate from a few large blocks. The architecture rationale is in gpu-rendering-guide. - A `VkImageMemoryBarrier2` that omits the layout transition, or uses `oldLayout` that does not match the image's current layout, is undefined behavior: track current layout per image/subresource. - A `VkDescriptorSet` updated while the GPU may still read it (without `UPDATE_AFTER_BIND`) is a data race: gate on the frame `VkFence`. - Binding a `VkPipeline` whose dynamic state you forgot to set (e.g. `vkCmdSetViewport`) draws nothing or validation-errors; declare every dynamic state you rely on. - Ignoring `VK_ERROR_OUT_OF_DATE_KHR`/`VK_SUBOPTIMAL_KHR` from acquire/present leaves a stale swapchain after resize: recreate it. ## Progressive Disclosure - Read [references/device-and-queues.md](references/device-and-queues.md) - Load when creating the instance/device or selecting queue families - Read [references/device-memory.md](references/device-memory.md) - Load when allocating device memory, choosing memory types, or staging uploads - Read [references/resources-and-barriers.md](references/resources-and-barriers.md) - Load when creating images/buffers or placing `VkImageMemoryBarrier2`/layout transitions - Read [references/descriptors.md](references/descriptors.md) - Load when building descriptor set layouts/pools/sets, going bindless, or using push constants - Read [references/pipelines.md](references/pipelines.md) - Load when building `VkPipeline`s, the pipeline cache, or dynamic rendering - Read [references/synchronization.md](references/synchronization.md) - Load when using `VkSemaphore`/`VkFence`, stage/access masks, or submit-time waits - Read [references/commands-and-swapchain.md](references/commands-and-swapchain.md) - Load when managing command pools/buffers, the swapchain, or frames-in-flight
Files in this package (12)
- SKILL.md 4 KB
- SOURCES.md 5 KB
- eval-queries.json 6 KB
- evals.json 3 KB
- moon.yml 1 KB
- references/commands-and-swapchain.md 6 KB
- references/descriptors.md 6 KB
- references/device-and-queues.md 4 KB
- references/device-memory.md 5 KB
- references/pipelines.md 6 KB
- references/resources-and-barriers.md 3 KB
- references/synchronization.md 5 KB
Versions
The database has no other skills named gpu-rendering-vulkan-guide.
Database snapshot: 13.11.2026 · package version 7c200c95d42a (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 95vulkan
Use this skill when the user says 'vulkan', 'vkQueueSubmit', 'swapchain', 'render pass', 'dynamic rendering', 'pipeline barrier', 'synchronization2', 'descriptor sets', 'VMA', 'vulkan validation'. Guides explicit Vulkan rendering: instance/device setup, swapchain, dynamic… - 90gpu-rendering-guide
Use when designing the architecture of a low-level GPU renderer on any explicit API (Vulkan/D3D12/Metal/WebGPU): render/frame graphs, shader/permutation systems, the descriptor/binding model, explicit GPU↔CPU synchronization, command recording and frames-in-flight, and GPU… - 100render-graph
Use this skill when the user says 'render graph', 'frame graph', 'transient resources', 'barriers', 'async compute', 'GPU-driven rendering', 'command lists'. Designs render/frame graphs: pass declaration, resource lifetimes and aliasing, automatic barriers, queues and async… - 100managing-metal4-synchronization
Metal 4 explicit synchronization — barriers, fences, events, and cross-API state translation (from D3D12, Vulkan, OpenGL). Use when porting GPU synchronization, translating resource state transitions, or debugging barrier-related visual corruption. - 100metal-game-rendering-workflow
Guide native Apple game and custom-renderer work with Metal 3 and Metal 4. Use when Codex owns MTKView, CAMetalLayer, drawable presentation, render passes, command encoding, MSL shaders, pipeline states, GPU resources, Metal capability fallback, GPU capture, or renderer… - 95translating-to-metal4-api
ALWAYS use as the Metal 4 API reference and cross-API translation map when porting graphics code to Metal 4 — translating Metal 3, D3D12, or Vulkan calls, looking up any MTL4 / Metal 4 API or renamed type, finding the Metal 4 equivalent of a source-API call, or learning what the…