Command Palette

Search for a command to run...

Architectures

A quick overview of what each architecture tends to be good at, based on how they show up on our current benchmarks.

Geometry-aware transformer

GeoTransolver

Transformer with multi-scale geometry injected at every layer. Strong on large surface/volume CFD and some transient structural cases; multi-scale context adds compute cost.

Strengths

  • Persistent geometry conditioning across depth
  • Multi-scale neighborhoods for non-uniform engineering meshes
  • Surface, volume, and transient structural examples in one framework

Limitations & caveats

  • Published accuracy is strongest on large accelerator configurations
  • Multi-scale context can roughly double throughput cost
  • Crash evidence is limited to short horizons and a small number of benchmark families
Compute
Multi-scale ball queries add material compute and memory overhead. Our DrivAerML reference run trained on one grossular-1 node (1× NVIDIA A10).
Preprocessing
Requires geometry point clouds, neighborhood queries, and task-specific point features.

Submissions (2)

ProblemSubmittedRun
ROEB-Aero: DrivAerML (Open subset)Jul 22, 202652bd9
ROEB-Crash: Rescale Tutorial (OpenRadioss Car)Jul 8, 2026984c9

Showing 1–2 of 2

1 / 1

Physics-attention transformer

Transolver

Transformer that learns physics tokens over mesh or grid points without explicit multi-scale geometry injection. Often faster to train than GeoTransolver; less geometry context at depth.

Strengths

  • Strong baseline on moderate-size CFD meshes (airfoil, plate)
  • Attention over learned physics tokens scales to many point-cloud formulations
  • Lower preprocessing burden than full geometry-token pipelines

Limitations & caveats

  • Less explicit multi-scale geometry context than GeoTransolver
  • Native multi-million-node industrial meshes may need capacity tuning
  • Not a drop-in substitute for GeoTransolver on every benchmark row
Compute
Typical Rescale grossular-1 runs finish in tens of minutes to a few hours depending on benchmark and epochs.
Preprocessing
Point or mesh features plus normalization statistics from the training split.

Submissions (2)

ProblemSubmittedRun
ROEB-Aero: Rescale Tutorial (2D Airfoil)Jun 24, 202668434
ROEB-Stress: Rescale Tutorial (Plate with Hole)Jun 24, 20269b52e

Showing 1–2 of 2

1 / 1

Geometry-aware transformer

Two-Stream Transformer

Dual-stream transformer for steady CFD on surface or volume fields. Strong on airfoil and plate-scale problems; crash / fea-deform use is emerging as Neon packs land.

Strengths

  • Strong on 2D/ moderate 3D steady CFD with clear global scalars
  • Handles native open meshes on several benchmarks without heavy decimation
  • Published Rescale AI Physics recipes for airfoil and plate

Limitations & caveats

  • Best-known recipes target steady single-frame fields
  • Transient crash evidence is thinner than for MeshGraphNet or GeoTransolver today
  • Very large volume meshes may need the same capacity planning as other transformers
Compute
Typical grossular-1 training runs from tens of minutes to a few hours.
Preprocessing
Mesh or point features, case parameters, and normalization from the training split.

Submissions (21)

ProblemSubmittedRun
ROEB-Stress: NAFEMS D1 (hole diameter)Sep 10, 2026ktno7
ROEB-Stress: NAFEMS D1 (hole diameter)Sep 9, 2026peak7
ROEB-Stress: NAFEMS D2 (diameter + thickness)Sep 9, 2026area7
ROEB-Stress: NAFEMS D2 (diameter + thickness)Sep 9, 2026are13
ROEB-Stress: NAFEMS D2 (diameter + thickness)Sep 9, 2026_mesh

Showing 1–5 of 21

1 / 5

Multi-scale neural operator

DoMINO

Decomposable point-cloud operator for large surface and volume CFD. Good when you can invest in SDF-style preprocessing; surface and volume models are often trained separately.

Strengths

  • Local evaluation can avoid processing every output point simultaneously
  • Multi-scale geometry encoding captures near and far context
  • Public surface and volume recipes are available

Limitations & caveats

  • Preprocessing is substantial and dataset-specific
  • Surface and volume models are commonly trained separately
  • Very large volume cases create significant storage and I/O pressure
Compute
Heavy preprocessing and long wall times are common. Our DrivAerML reference run used one grossular-1 node (1× NVIDIA A10) for ~31 hours.
Preprocessing
Signed-distance fields, nearest-neighbor features, bounding boxes, normalization statistics, and optional cached geometry encodings.

Submissions (4)

ProblemSubmittedRun
ROEB-Stress: NAFEMS D1 (hole diameter)Sep 11, 2026DoMIN
ROEB-Aero: DrivAerML (Open subset)Jul 21, 2026569e9
ROEB-Stress: Rescale Tutorial (Plate with Hole)Jun 18, 20261e4d5
ROEB-Aero: Rescale Tutorial (2D Airfoil)Jun 17, 20264dfb6

Showing 1–4 of 4

1 / 1

Graph neural network

MeshGraphNet

Message passing over mesh connectivity. Natural for transient rollout on a fixed topology. Vanilla MGN builds one graph over every node and edge, so large industrial surfaces usually need heavy decimation or partitioning before they fit on a GPU — and that changes the problem.

Strengths

  • Mesh connectivity provides a strong local physical prior
  • Natural fit for transient rollout on a consistent topology
  • Mature message-passing ecosystem and multiple scalable variants

Limitations & caveats

  • Long-range effects require depth, hierarchy, or multi-scale variants
  • Full industrial meshes can exceed practical accelerator memory without decimation or partitioning
  • Decimated runs are not peer-comparable with native-resolution scores on the same problem
  • Scalable variants (for example X-MeshGraphNet) should not be attributed to the base architecture without qualification
Compute
Memory grows with graph size and edge count. Example: DrivAerML open packs use native ~8M-node surfaces — a full MeshGraphNet graph there is impractical on typical GPUs without cutting to ~200k vertices (or partitioning). Prefer architectures that keep the native mesh when that is the published protocol.
Preprocessing
Graph construction, edge features, consistent topology handling, and often mesh partitioning or decimation on large surfaces.

Submissions (15)

ProblemSubmittedRun
ROEB-Stress: NAFEMS D6 (full LHS)Sep 12, 2026sig13
ROEB-Stress: NAFEMS D6 (full LHS)Sep 12, 2026sign7
ROEB-Stress: NAFEMS D6 (full LHS)Sep 12, 2026signt
ROEB-Stress: NAFEMS D2 (diameter + thickness)Sep 11, 2026var13
ROEB-Stress: NAFEMS D2 (diameter + thickness)Sep 11, 2026_vary

Showing 1–5 of 15

1 / 3

Attention-based surrogate

Transformer (general)

Attention over tokens derived from mesh points, patches, or physics-aware embeddings. Flexible, but geometry and scale choices dominate whether it works on industrial meshes.

Strengths

  • Flexible tokenization for many mesh and parameter layouts
  • Strong when global context and multi-field outputs matter
  • Large ecosystem of training tricks and scaling recipes

Limitations & caveats

  • Quadratic attention cost without specialized variants
  • Needs careful geometry encoding for unstructured industrial meshes
  • Compute and memory scale quickly on million-node surfaces

Submissions

No submissions on this architecture yet.

Graph neural network

Graph neural network (general)

Message passing on mesh or k-NN graphs. Excellent local physics prior; watch memory on dense industrial connectivity.

Strengths

  • Natural fit for mesh adjacency and local conservation-like patterns
  • Well suited to transient steps on fixed topology
  • Many published variants (MeshGraphNet, GraphCast-style stacks)

Limitations & caveats

  • Long-range coupling needs depth or hierarchical graphs
  • Edge count drives memory on fine industrial meshes
  • Decimated graphs change the problem. compare like with like

Submissions

No submissions on this architecture yet.

Neural operator

Neural operator (general)

Learns mappings between function spaces (FNO, DeepONet, GNO, etc.). Strong for parametric PDE families when training distribution matches deployment.

Strengths

  • Designed for family-of-PDEs and parameter sweeps
  • Can generalize across boundary conditions when trained broadly
  • Multiple public operator architectures to build on

Limitations & caveats

  • Often assumes structured grids or specific query patterns
  • Unstructured million-node CFD may need hybrid or point-cloud operators (e.g. DoMINO-style)
  • Preprocessing and normalization dominate transfer quality

Submissions

No submissions on this architecture yet.

Convolutional network

CNN (general)

Convolution on structured grids or voxelized fields. Fast and mature when the problem can be cast to a regular lattice.

Strengths

  • Efficient on structured or voxelized representations
  • Mature tooling and predictable memory
  • Good baseline for 2D fields and parametric sweeps on grids

Limitations & caveats

  • Unstructured industrial meshes need resampling or projection
  • Geometry fidelity can be lost in voxelization
  • Less natural for native CAD surface meshes at full resolution

Submissions

No submissions on this architecture yet.

Hybrid architecture

Hybrid (general)

Combines operators (e.g. GNN + transformer, FNO + CNN). Useful when no single inductive bias covers the whole mesh. Document which parts do what.

Strengths

  • Can mix local mesh bias with global attention or operators
  • Room to specialize submodules for parameters vs fields

Limitations & caveats

  • More moving parts. Harder to compare fairly across submissions
  • Preprocessing and compute cost sum across branches
  • Reviewers need clarity on which components are novel vs off-the-shelf

Submissions

No submissions on this architecture yet.

Problem fit (quick reference)

ProblemCommon architecturesWatch out for
Plate with holeTwo-Stream Transformer, MeshGraphNet, Transolver, DoMINOOne physics story, several datasets: Rescale Tutorial solids plus NAFEMS D1–D6 shells. Moderate meshes fit most families; compare train time fairly. NAFEMS scores use the peak-error suite (AEinPEAK, AE@PEAK, MAE, …) plus S1 / Kt where reported.
2D AirfoilTwo-Stream Transformer, MeshGraphNet, Transolver, DoMINOSteady 2D CFD testbed: surface Cp plus global Cl/Cd on a fixed airfoil mesh.
DrivAerMLDoMINO, Two-Stream Transformer, GeoTransolver (Open subset); MeshGraphNet (Decimated)Native ~8M-node surfaces. Vanilla MeshGraphNet usually needs heavy decimation; the ~200k MeshGraphNet run lives on the Decimated pack, not as a peer on Open subset.
OpenRadioss CarMeshGraphNet, GeoTransolver (more welcome)Transient FE displacement on Neon. Published rows today are MGN and GeoT. Mesh resolution can differ by dataset (some packs are decimated).
HiLiftAeroMLTwo-Stream Transformer (and other large-surface CFD arches welcome)High-lift aircraft on a cleaned ~8M CAD skin. Native surfaces favor transformers and DoMINO-style operators over vanilla MeshGraphNet.

See also the FAQ on DrivAerML and MeshGraphNet and the evaluation protocol.