Watch

fix(window): size the swapchain for its present mode, and key frames in flight by frame instead of image

Window
- CreateSwapchain asked for exactly numFrames (3) images whatever the
  surface said, read them into a three-element array, and keyed the
  command buffers, the fences and the descriptor-heap copy holding "its"
  swapchain image by the acquired image index. A swapchain of another
  size fell back to FIFO (asked for MAILBOX or IMMEDIATE) or threw, and a
  surface whose minimum is above three got a create info below it
  (#160). Mesa's MAILBOX needs four images: Turnip on phoc, llvmpipe on
  any Wayland compositor.
- The present mode is picked first, then the swapchain is sized by
  SwapchainCapabilities: the surface's counts for that mode where the
  instance has surface_maintenance1. With swapchain_maintenance1 the
  swapchain names its mode (VkSwapchainPresentModesCreateInfoKHR), which
  is what makes the mode's own count the one that applies, at creation and
  to the forward progress of a UINT64_MAX acquire; without it the stricter
  of the mode's and the surface's counts for all modes. It asks for
  numFrames or that minimum where more, never past the maximum
  (SwapchainMinImageCount). Turnip's FIFO keeps its three images and
  MAILBOX gets four. The FIFO fallback and presentModeUnfit are gone.
- images, imageViewInfos (was imageViews: they are create-infos), the made
  imageViews that descriptor sets take, imageInitialised and the
  render-finished semaphores are per image, as many as the driver made.
  The views are made once per swapchain. The semaphores only grow, since
  a present of the old swapchain may still wait on one.
- Frames in flight stay numFrames, keyed by currentFrame, frameCounter %
  numFrames: the command buffer, the fence, the acquire semaphore, the
  heap copy (or set), RenderingElement3D::tlases, the frameIdx a pass is
  recorded with. currentImage is the acquired image. currentBuffer is
  gone, so code that used it stops compiling instead of picking the wrong
  one; RenderPass::Record's frameIdx keeps its name and is now the frame
  slot, as RenderPass.cppm says.
- The frame's fence is waited before the acquire. The acquire semaphores
  need no spare then, and a frame waits for exactly the one numFrames
  before it, which the deletion queue and the staging rings count on and
  fences keyed by image did not promise once images come back out of
  order. The swapchain slots are written before the fence is reset.
- From the acquire to the present the window is recording: a Resize or
  SetPresentMode from a listener (a VSync key ticked in onUpdate) takes
  effect after the present, through RebuildSwapchain, together with a
  present that found the swapchain out of date. Rebuilding mid-frame used
  to destroy the image the frame was drawing into; with images sized per
  swapchain it would also index past the new one's end.

DescriptorHeapVulkan
- Frames and images pair up differently from frame to frame, so no heap
  copy can keep one swapchain image. AddSwapchainSlot(slot, window) lists
  an image slot as that window's swapchain image, and Window::Render
  writes the image a frame acquired into its window's listed slots in that
  frame's copy, or its set, before onUpdate, skipping a slot that holds it
  already. A slot is one window's, so two windows can share a heap without
  writing each other's slots under each other's frames in flight. A slot
  listed during a frame its window records is written at once. A slot past
  the heap's image slots, an empty ImageSlot's 0xFFFF among them, is
  refused. Freeing a slot unlists it, as does initialising the heap in
  either binding model. WriteImageSet, the one use of which this
  replaces, is gone, and with it the per-frame dimension of setImageViews.

Device
- VK_KHR_surface_maintenance1 (or EXT) and VK_KHR_get_surface_capabilities2
  are enabled where the loader has them (surfaceMaintenance1Supported,
  vkGetPhysicalDeviceSurfaceCapabilities2KHR), and the matching
  swapchain_maintenance1 with its feature where the device has it
  (swapchainMaintenance1Supported).
- DistinctGpus groups the physical devices by GPU as CRAFTER_GRAPHICS_DEVICE
  numbers them, for ChooseGpu and tests.

UI
- UIRenderer lists its output slot instead of writing every copy itself,
  and again on every resize; the resize listener is gone on both
  backends. The headers' frameIdx is currentFrame.

Examples
- VulkanTriangle, Sponza, SponzaBench and RTVolume list their output slot
  and write only the per-frame TLASes. They wrote imageViews[f] into copy
  f once, which a resize left pointing at destroyed images.

Tests
- SwapchainImageCount: SwapchainMinImageCount on the numbers drivers
  report; every offered mode keeps its mode and gets what the window asks
  for; every frame presents the colour it drew, read back a texel per
  frame; a switch and a resize asked for during a frame wait for its
  present; a slot listed during a frame holds its image at once; two
  windows on one heap draw their own colours. The frame checks run again
  in child processes, through CRAFTER_GRAPHICS_DEVICE, on a GPU whose
  swapchains have more images than frames (this one, else llvmpipe), in
  each binding model the parent didn't cover; a child skips only for a GPU
  that can't run the library, and checks it runs the model it was asked
  for.
- FrameLoopSync checks the frame slot follows the frame counter and counts
  the swapchain's own images. HeapASRead, HeapDynamicIndexHit and
  RTPassPushData list their swapchain slot and drop their resize rewrites.

Docs
- README, AGENTS.md and AGENTS.consumer.md: frames and images, the
  swapchain slots, rebuilds held to the present, the new optional groups.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

2026-10-31 18:14:18 +01:00

Examples

Each example is a self-contained crafter-build project that depends on the parent Crafter.Graphics via LocalProject. To build and run any of them:

cd examples/<name>
crafter-build -r

Index

HelloWindow

Minimum viable program: open a window, run the event loop. No Vulkan rendering. Useful as a smoke test for Device::Initialize + Window + the platform backend.

Ray-traced single triangle through vkCmdTraceRaysKHR. Shows the full ray-tracing setup: DescriptorHeapVulkan with image and buffer slots, PipelineRTVulkan from raygen / miss / closesthit SPIR-V, BLAS via Mesh::Build, TLAS via RenderingElement3D::BuildTLAS, direct vkWriteResourceDescriptorsEXT for the per-frame TLASes, AddSwapchainSlot for the output image, RTPass on window.passes. Smallest test of the bindless ray-tracing path.

HelloUI

Compute-shader UI demo using all three UI tiers:

Tier 3 components: DrawButton, DrawSlider, DrawProgressBar, composed via Rect::SubRect for resize-safe layout.
Tier 2 standard shaders: DispatchQuads for the background and components, DispatchCircles for a cursor-tracking dot, DispatchText for the button label (with the FontAtlas wired up to UIRenderer).
Tier 1 is available too — any custom ComputeShader registered on the same heap can be dispatched alongside the standard ones.

Hit-testing and animation are user code (see the EventListener subscriptions on window.onMouseMove / onMouseLeftClick); the library does not track widgets or focus.

Drop a TTF in this directory as font.ttf before running (the example loads it via Font("font.ttf")).

InputSystem

Guided tour of Crafter::Input: name actions ("Jump", "Move", "Fire", "Look", "Zoom"), bind them to keys / mouse / gamepad (with composite bindings for WASD-as-Vector2 and analog sticks), and consume them as events. Demonstrates:

The compile-time Key(CrafterKeys::Space) helper that folds to a per-platform raw scancode — bindings stay cross-platform-readable in source while runtime data stores raw codes only.
All four action types (Button, Axis1D, Vector2) with multiple bindings per action (any-of semantics).
Map::StartRebind — press R, then press any input to remap "Jump" at runtime. Captured input is filtered out for that frame.
BindingToString / BindingFromString round-trip — print the default bindings as the on-disk format.
Gamepad hot-plug events: plug a controller in mid-run and the bindings start firing immediately.

Console-driven (no UI rendering needed); focus the window and watch stdout.

CustomShader

Tier 1 demo: a user-authored compute shader (inverse-circle.comp.glsl) running alongside the shipped drawQuads. The custom shader inverts RGB under each item-circle — exactly the kind of effect attempt #2's closed shader couldn't express. Shows:

Defining your own item POD struct in C++ + matching std430 struct in GLSL.
#include "../../shaders/ui-shared.glsl" for the bindless heap declarations + UIDispatchHeader push-constant contract.
ComputeShader::Load for the .spv, UIRenderer::RegisterBuffer for your SSBO, FillHeader to populate the standard prefix, and UIRenderer::Dispatch to launch — the same pattern the standard shaders use under the hood.
The inter-dispatch SHADER_WRITE → SHADER_READ|WRITE barrier is inserted automatically, so the custom shader sees the colored stripes drawn by the prior DispatchQuads and reads/writes the swapchain image safely.

RayQueryPick

Regression test for the WebGPU software ray-query shim. Builds a 512-instance TLAS and shoots one ray through a rayQuery=true PlainComputeShader, reading the committed hit back to the host and checking it against the analytically-known answer. Guards against the hardcoded-leaf-start TLAS-traversal bug (issue #25) that made every rayQuery pick miss for realistic instance counts. WebGPU/DOM only.

HDRBloom

Cross-backend-shaped HDR bloom on the WebGPU backend (issue #27). The RT pipeline's RESOLVE writes linear radiance into a user rgba16float texture (StorageImage2D, PipelineRTWebGPU::Init(..., RGBA16Float)), two PlainComputeShader passes threshold + blur it through float storage targets (UICustomBindingKind::StorageTexture), and a UI custom shader composites scene + bloom with a Reinhard tonemap onto the canvas. Exercises the three primitives an HDR post-process chain needs: float textures, write-only storage-texture bindings, and pre-tonemap radiance out of the wavefront. The threshold/blur passes run from onBeforeUpdate so each gets its own queue submit — the storage-write → sampled-read barrier WebGPU only provides between submits (or between passes), never within a single compute pass. WebGPU/DOM only; the same chain is wireable on Vulkan today via an offscreen HDR heap image + a composite RenderPass (the present path records passes generically and barriers between them).

RTMultiShadow

Multi-light shadowing through the wavefront RT pipeline (issue #30). Five pillars on a checkered ground, four colored point lights; the closest-hit emits one shadow ray per light from the same invocation, so up to four rays per pixel resolve in a single SHADE pass. Exercises both halves of the >1-ray-per-pixel-per-bounce widening: the atomic rtAccumulate (per-channel f32 CAS — concurrent same-pixel adds don't race) and RTPass::raysPerPixel (scales the ray/hit/payload buffers so the per-light emits aren't capacity-dropped). Any regression shows up as flickering dark noise or a missing shadow color in the overlap regions. WebGPU/DOM only.

5.7 KiB Raw Blame History

Examples

Index

5.7 KiB

Raw Blame History