Mirrored_Repos/DarkflameServer
1
0
Fork 0
mirror of https://github.com/DarkflameUniverse/DarkflameServer.git synced 2026-10-02 02:43:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat(ugc): ray_backend=hiprt traces on the GPU with HIPRT and Orochi (optional build)

Browse Source 
HIPRT (MIT) behind the CMake option DLU_HIPRT (off): its headers come from its
SDK (HIPRT_ROOT, else ROCm's /opt/rocm) and are copied next to the servers; its
library is loaded when first used (hiprtew), as HIP or CUDA are by Orochi
(MIT, fetched pinned by hash; CUDA when its toolkit is found). The trace
kernels (nearest hit skipping the triangle a ray leaves, any hit) are compiled
the first time and kept in cache/hiprt. One GPU context for the process
(hiprt_device picks the GPU); the workers take turns on it. When HIPRT, the
GPU or a scene's upload fails, Embree is used instead, and the UGC server logs
why at start.

For a GPU the rays go in batches (UgcRays::Scene gets batch queries; the CPU
backends answer them a ray at a time):
- hidden faces: with a batch backend the paths are traced side by side, a
  bounce at a time (the path code split into Start, Scatter and Bounce, the one
  by one tracing unchanged); the same paths with the same random numbers, so
  the same triangles are decided (tested with builtin side by side)
- the occlusion bake and the denoised icons' traced occlusion always ask in
  batches (the same rays, the same results)

Its symbols are hidden: the servers export theirs (-rdynamic), and HIPRT's
library, which has an Orochi of its own, would otherwise call ours.

Check: configure with -DDLU_HIPRT=ON on a machine with ROCm (or HIPRT's SDK)
and an AMD RDNA or NVIDIA GPU; UgcServer --make-model x.lxfml out hiprt; the
UGC tests (hits, hidden faces and occlusion against builtin); a build without
it leaves everything as before.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Aaron Kimbrell
2026-09-29 11:57:37 -05:00
parent 2e2e8153e2
commit 07dbf30dc8
13 changed files with 719 additions and 92 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
tests/dUgcTests/UgcTests.cpp
View File

@@ -2215,6 +2215,24 @@ TEST(UgcRays, BackendsFindTheSameHits) {
}
}
TEST(UgcHsr, SideBySideDecidesTheSameAsOneByOne) {
// The paths traced side by side (as for a GPU) are the same paths, so the same triangles stay; ground plane too
for (const bool ground : { false, true }) {
UgcHsr::Options options;
options.seed = 5;
options.samples = 2;
options.groundPlane = ground;
const auto mesh = Clutter();
uint64_t points = 0, paths = 0, sidePoints = 0, sidePaths = 0;
const auto oneByOne = UgcHsr::Visible(mesh, options, &points, &paths);
options.sideBySide = true;
EXPECT_EQ(UgcHsr::Visible(mesh, options, &sidePoints, &sidePaths), oneByOne) << ground;
EXPECT_EQ(sidePoints, points);
EXPECT_GE(sidePaths, paths); // it traces a round's other paths too once one escaped
EXPECT_EQ(UgcHsr::Visible(Room(true), options), UgcHsr::Visible(Room(true), UgcHsr::Options{ .groundPlane = ground, .samples = 2, .seed = 5 }));
}
}
TEST(UgcRays, OtherBackendsMakeTheSameModels) {
// The hidden faces and the occlusion with each backend. The paths bounce, so a hit found a rounding further away
// sends a path on from a slightly different point: the rare triangle decided by a path that only just gets out