Skip to content

Cover basic RayQuery methods on the single-triangle BLAS#1271

Open
MarijnS95 wants to merge 1 commit into
llvm:mainfrom
Traverse-Research:inlinert-shader-side-queries
Open

Cover basic RayQuery methods on the single-triangle BLAS#1271
MarijnS95 wants to merge 1 commit into
llvm:mainfrom
Traverse-Research:inlinert-shader-side-queries

Conversation

@MarijnS95

@MarijnS95 MarijnS95 commented Jun 3, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator

Summary

Stacks on top of #1232 and #1245 to add five small InlineRT tests, each isolating one RayQuery method on the existing single-triangle BLAS:

  • miss-status.test — COMMITTED_NOTHING path (ray points away from geometry)
  • ray-t.test — CommittedRayT() returns exact 1.0 for the axis-aligned hit
  • barycentrics.test — CommittedTriangleBarycentrics() at world (0,0,0) returns exactly (0.25, 0.25)
  • world-ray-echo.test — WorldRayOrigin / WorldRayDirection / RayTMin / RayFlags round-trip into a structured buffer; passes -fvk-use-dx-layout so SPIR-V matches DXIL's tight float3 packing and the expected bytes are portable across DX / VK / MTL.
  • tmin-tmax-clip.test — two queries against the same BLAS: one with TMin past the hit, one with TMax before it; both must miss.

First batch out of #1258 (inline-RT test coverage epic) — the easiest wins, no framework / YAML changes required.

Marked draft because it depends on #1232 and #1245 landing first.

Test plan

Local on an NVIDIA RTX 3060:

  • Linux Vulkan (native offloader)
  • Linux D3D12 (Wine + vkd3d-proton + cross-compiled offloader.exe)
  • Windows Vulkan (native offloader.exe)
  • Windows D3D12 (native offloader.exe)

CI (RT-capable runners):

  • windows-nvidia D3D12 (RaytracingTier 1.2)
  • windows-intel VK (VK_KHR_ray_tracing_pipeline)
  • macOS Metal (supportsRaytracing)

@MarijnS95 MarijnS95 force-pushed the inlinert-shader-side-queries branch from 1b77bf5 to e6e89ea Compare June 3, 2026 09:04
This was referenced Jun 3, 2026
Open
Open
Open
@MarijnS95 MarijnS95 force-pushed the inlinert-shader-side-queries branch from e6e89ea to 92f47db Compare June 3, 2026 09:21
@MarijnS95 MarijnS95 changed the title Add shader-side RayQuery query tests Add InlineRT tests for committed-hit and ray-side methods Jun 3, 2026
@MarijnS95 MarijnS95 force-pushed the inlinert-shader-side-queries branch from 92f47db to d864e01 Compare June 3, 2026 09:21
@MarijnS95 MarijnS95 changed the title Add InlineRT tests for committed-hit and ray-side methods Cover basic RayQuery methods on the single-triangle BLAS Jun 3, 2026
@MarijnS95 MarijnS95 mentioned this pull request Jun 3, 2026
Open
7 tasks
@MarijnS95 MarijnS95 mentioned this pull request Jun 3, 2026
Draft
7 tasks
MarijnS95 added a commit to Traverse-Research/offload-test-suite that referenced this pull request Jun 3, 2026
@MarijnS95 @claude
Add first PSO RT test batch: DispatchRays + SBT routing
29c73ee 
Four small tests stacked on top of llvm#1275, each isolating one
shader-observable PSO raytracing surface. They follow the same shape as
the inline-RT batch already in llvm#1271 / llvm#1272 / llvm#1274 / llvm#1276 — one .test
file per behavior, single-purpose shader, exact buffer comparison.

  - `dispatch-rays-index.test` — 4x1x1 dispatch, raygen writes
    `DispatchRaysIndex().x` into `Output[index]`. Confirms the
    dispatch grid plumbs through to the per-lane system value with no
    BLAS / TLAS / hit groups in play (RT-pipeline-only, no AS binding).
  - `dispatch-rays-dimensions.test` — 2x3x1 dispatch, raygen packs the
    constant `DispatchRaysDimensions()` into one uint per lane.
    Confirms every lane sees the host-side `{W, H, D}` even when only
    one dimension > 1.
  - `miss-shader-index.test` — two miss shaders writing distinct
    sentinels (0xAA / 0xBB). 2-lane dispatch picks `MissShaderIndex` 0
    and 1 respectively; rays start far enough from the geometry that
    every ray misses. Verifies the SBT miss region's per-record
    routing.
  - `ray-contribution-to-hit-group-index.test` — two hit groups with
    distinct closest-hit shaders (0xA1 / 0xB2). 2-lane dispatch picks
    `RayContributionToHitGroupIndex` 0 and 1, every ray hits the same
    triangle. Verifies the SBT hit-group region's per-record routing.

The first two have no AS / Miss / HitGroup in their pipeline at all —
just a raygen + a UAV — which exercises the minimum viable RT pipeline
shape (one raygen group, zero-sized miss / hit / callable SBT regions).
The latter two reuse the single-triangle BLAS/TLAS from
`raygen-roundtrip.test`.

All four tests are `# REQUIRES: raytracing-pipeline` with `# XFAIL: Clang`
— Clang (`clang-dxc`) doesn't yet lower `[shader("…")]` entry points to
either DXIL libraries or SPIR-V. With the Metal RT bring-up rebased on
top, all four pass natively on Apple Silicon and Metal is dropped from
the XFAIL list.

Locally verified end-to-end on the user's Linux box:
  - Vulkan via the native offloader against an NVIDIA RTX 3060:
    all four tests PASS.
  - D3D12 via Wine + vkd3d-proton + the cross-compiled offloader.exe
    on the same GPU: all four tests PASS.
And on macOS 15 / metal-irconverter 3.1.1:
  - Metal via the native offloader: all four tests PASS.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@MarijnS95 MarijnS95 force-pushed the inlinert-shader-side-queries branch from d864e01 to bfd904b Compare June 8, 2026 12:31
MarijnS95 added a commit to Traverse-Research/offload-test-suite that referenced this pull request Jun 8, 2026
@MarijnS95 @claude
Add first PSO RT test batch: DispatchRays + SBT routing
1b0010a 
Four small tests stacked on top of llvm#1275, each isolating one
shader-observable PSO raytracing surface. They follow the same shape as
the inline-RT batch already in llvm#1271 / llvm#1272 / llvm#1274 / llvm#1276 — one .test
file per behavior, single-purpose shader, exact buffer comparison.

  - `dispatch-rays-index.test` — 4x1x1 dispatch, raygen writes
    `DispatchRaysIndex().x` into `Output[index]`. Confirms the
    dispatch grid plumbs through to the per-lane system value with no
    BLAS / TLAS / hit groups in play (RT-pipeline-only, no AS binding).
  - `dispatch-rays-dimensions.test` — 2x3x1 dispatch, raygen packs the
    constant `DispatchRaysDimensions()` into one uint per lane.
    Confirms every lane sees the host-side `{W, H, D}` even when only
    one dimension > 1.
  - `miss-shader-index.test` — two miss shaders writing distinct
    sentinels (0xAA / 0xBB). 2-lane dispatch picks `MissShaderIndex` 0
    and 1 respectively; rays start far enough from the geometry that
    every ray misses. Verifies the SBT miss region's per-record
    routing.
  - `ray-contribution-to-hit-group-index.test` — two hit groups with
    distinct closest-hit shaders (0xA1 / 0xB2). 2-lane dispatch picks
    `RayContributionToHitGroupIndex` 0 and 1, every ray hits the same
    triangle. Verifies the SBT hit-group region's per-record routing.

The first two have no AS / Miss / HitGroup in their pipeline at all —
just a raygen + a UAV — which exercises the minimum viable RT pipeline
shape (one raygen group, zero-sized miss / hit / callable SBT regions).
The latter two reuse the single-triangle BLAS/TLAS from
`raygen-roundtrip.test`.

All four tests are `# REQUIRES: raytracing-pipeline` with `# XFAIL: Clang`
— Clang (`clang-dxc`) doesn't yet lower `[shader("…")]` entry points to
either DXIL libraries or SPIR-V. With the Metal RT bring-up rebased on
top, all four pass natively on Apple Silicon and Metal is dropped from
the XFAIL list.

Locally verified end-to-end on the user's Linux box:
  - Vulkan via the native offloader against an NVIDIA RTX 3060:
    all four tests PASS.
  - D3D12 via Wine + vkd3d-proton + the cross-compiled offloader.exe
    on the same GPU: all four tests PASS.
And on macOS 15 / metal-irconverter 3.1.1:
  - Metal via the native offloader: all four tests PASS.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@MarijnS95 MarijnS95 force-pushed the inlinert-shader-side-queries branch from bfd904b to 16c9883 Compare June 11, 2026 14:12
@MarijnS95 @claude
Cover basic RayQuery methods on the single-triangle BLAS
e21ef98 
Adds five small InlineRT tests on top of the existing single-triangle
BLAS, each isolating one RayQuery method:

- miss-status: COMMITTED_NOTHING path (ray points away from geometry)
- ray-t: CommittedRayT() returns exact 1.0 for the axis-aligned hit
- barycentrics: CommittedTriangleBarycentrics() at world (0,0,0) returns
  exactly (0.25, 0.25)
- world-ray-echo: WorldRayOrigin / WorldRayDirection / RayTMin / RayFlags
  round-trip into a structured buffer. Passes `-fvk-use-dx-layout` so
  the SPIR-V backend uses DXIL's tight float3 packing and the expected
  bytes match across DX/VK/MTL.
- tmin-tmax-clip: two queries against the same BLAS — one with TMin past
  the hit, one with TMax before the hit — both must miss.

Part of the inline-RT test coverage epic
(llvm#1258).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@MarijnS95 MarijnS95 marked this pull request as ready for review June 11, 2026 16:56
@MarijnS95 MarijnS95 force-pushed the inlinert-shader-side-queries branch from 16c9883 to e21ef98 Compare June 11, 2026 16:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@manon-traverse manon-traverse manon-traverse approved these changes

@EmilioLaiso EmilioLaiso Awaiting requested review from EmilioLaiso

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@MarijnS95 @manon-traverse