Contact warm start

26 · Convergence

The eight-box stack again, at 4 iterations per step. contact_warmstart starts each step from the matched contacts' impulses of the previous step, so the solve continues where it left off. The collision pipeline must be created with contact matching to find last step's contacts.

drag to orbit · scroll to zoom

0.00 s

Cost

Time per step, CUDA graph replay, median of 7 interleaved rounds on one NVIDIA RTX PRO 6000 Blackwell Server Edition; ratios are against the first row. Solver allocation is the GPU memory the solver allocates at construction, before warm-up and graph capture. Scene: the 8-box stack in each world, 8 steps of 1/240 s per frame.

Option	1 world	64 worlds	1024 worlds	solver allocation, 1024 worlds	result in this tutorial
iterations=4	0.55 ms	0.72 ms	1.09 ms	215 MB	sag 7.0 mm
iterations=4 + warm start	0.69 ms 1.26×	1.03 ms 1.42×	1.51 ms 1.39×	230 MB	sag 0.5 mm mean, chatters (sd 1.1 mm)
iterations=8 + warm start	0.91 ms 1.65×	1.26 ms 1.75×	1.76 ms 1.62×	230 MB	separate run: sag 0.04 mm mean, sd 0.45 mm
iterations=12	1.01 ms 1.84×	1.21 ms 1.68×	1.64 ms 1.51×	215 MB	sag 2.1 mm

What to look at

Cold start at 4 iterations: about 7 mm sag.
Warm start at 4 iterations: about 0.5 mm on average, better than 12 cold iterations, but the top box chatters by about ±1.5 mm.
With contact_warmstart_decay=0.9 the chatter falls to about a third (standard deviation 0.33 vs 1.1 mm) at about 1 mm mean sag, still below 12 cold iterations and cheaper per step (see Cost).

Key parameter

contact_warmstart

Needs CollisionPipeline(model, contact_matching="latest" or "sticky"). contact_warmstart_decay scales the carried impulses.

Try changing

Use contact_matching="sticky".
Set contact_warmstart_decay=0.5.

Concept

Warm start

What it is

Contacts that persist from step to step usually need similar impulses. With contact_warmstart the solve starts from last step's impulses, matched by contact identity, scaled by the step ratio and clamped to the current friction cone.

In this scene

At 4 iterations the sag drops from 7.0 mm to 0.5 mm on average, better than 12 cold iterations (2.1 mm), but the top box chatters (standard deviation 1.1 mm). A decay of 0.9 cuts the chatter to 0.33 mm at 1.0 mm mean sag.

Considerations

Cost: +26% per step in one world and +39% in 1024 worlds (1.09 → 1.51 ms), plus 15 MB of solver memory for 1024 worlds; the pipeline's contact matching adds its own memory and about 3% time (separate run).
Interactions: Needs pipeline contact_matching; not with sleeping, contact compliance or torsion.
Failure modes: Stale impulses can cause chatter at low iteration counts; contact_warmstart_decay below 1 trades some sag for less chatter.
Use it when: Persistent stacks or grasps where 4 warm iterations beat 12 cold ones in both sag and time, as here.

The script

cd tutorials && python 26_contact_warmstart.py prints each variant's final metric; --record out.json also writes the replay. Every tutorial imports replay.py from the same folder.

26_contact_warmstart.py
replay.py

# SPDX-FileCopyrightText: Copyright (c) 2026 Skild AI and the Steel contributors
# SPDX-License-Identifier: Apache-2.0

"""contact_warmstart: start each step from the matched contacts' previous impulses."""

import newton
import warp as wp
from newton.solvers import SolverSteel

from replay import Recorder

DT = 1.0 / 240.0
SUBSTEPS = 8
N, H = 8, 0.05


def build():
    builder = newton.ModelBuilder()
    builder.add_ground_plane(cfg=newton.ModelBuilder.ShapeConfig(mu=0.6))
    for i in range(N):
        box = builder.add_body(
            xform=wp.transform((0.0, 0.0, H + 2 * H * i + 0.0002 * i), wp.quat_identity())
        )
        builder.add_shape_box(box, hx=H, hy=H, hz=H, cfg=newton.ModelBuilder.ShapeConfig(mu=0.6))
    return builder.finalize()


def simulate(rec, label, warmstart, iterations, decay=1.0):
    model = build()
    solver = SolverSteel(
        model, iterations=iterations, contact_warmstart=warmstart, contact_warmstart_decay=decay
    )
    # warm start finds last step's impulses through the pipeline's contact matching
    pipeline = newton.CollisionPipeline(
        model, contact_matching="latest" if warmstart else "disabled"
    )
    contacts = pipeline.contacts()
    state_0, state_1, control = model.state(), model.state(), model.control()

    run = rec.variant(label, model)
    for _ in range(rec.frames(4.0)):
        top = state_0.body_q.numpy()[N - 1, 2]
        run.frame(state_0, sag_mm=1000.0 * (H * (2 * N - 1) - top))
        for _ in range(SUBSTEPS):
            pipeline.collide(state_0, contacts)
            solver.step(state_0, state_1, control, contacts, DT)
            state_0, state_1 = state_1, state_0
    run.check(solver)


if __name__ == "__main__":
    rec = Recorder()
    simulate(rec, "iterations=4, cold start (default)", False, 4)
    simulate(rec, "iterations=4, contact_warmstart=True", True, 4)
    simulate(rec, "iterations=4, warm start, decay 0.9", True, 4, decay=0.9)
    simulate(rec, "iterations=12, cold start", False, 12)
    rec.save()

# SPDX-FileCopyrightText: Copyright (c) 2026 Skild AI and the Steel contributors
# SPDX-License-Identifier: Apache-2.0

"""Record Steel tutorial runs as replay JSON for the tutorials page, or print a summary.

Every tutorial imports this module. ``python NN_name.py`` runs the variants and prints
the final metric of each; ``--record out.json`` also writes body poses, shapes and metric
series that the page replays in 3D.
"""

import argparse
import base64
import json
import os
import subprocess
import sys
import time

import newton
import numpy as np
import warp as wp

from source_hash import source_sha256

_RENDERED_TYPES = {
    newton.GeoType.PLANE,
    newton.GeoType.SPHERE,
    newton.GeoType.CAPSULE,
    newton.GeoType.ELLIPSOID,
    newton.GeoType.CYLINDER,
    newton.GeoType.BOX,
    newton.GeoType.MESH,
    newton.GeoType.CONE,
    newton.GeoType.CONVEX_MESH,
}


class Recorder:
    """Collect the variants of one tutorial; ``fps`` is the replay frame rate."""

    def __init__(self, fps: int = 30):
        parser = argparse.ArgumentParser()
        parser.add_argument(
            "--record", metavar="OUT.json", help="write a replay file for the tutorials page"
        )
        parser.add_argument("--device", default=None, help="Warp device, for example cuda:0")
        self.args = parser.parse_args()
        if self.args.device:
            wp.set_device(self.args.device)
        self.fps = fps
        self.script = os.path.abspath(sys.argv[0])
        self.variants: list[Variant] = []

    def frames(self, seconds: float) -> int:
        """Number of replay frames in ``seconds``, including the first."""
        return int(round(seconds * self.fps)) + 1

    def variant(
        self,
        label: str,
        model: newton.Model,
        *,
        ghost: bool = False,
        slot: int | None = None,
        scene: str | None = None,
    ) -> "Variant":
        """Start one run; ``ghost`` overlays run ``slot``, ``scene`` groups runs for charts."""
        variant = Variant(label, model, self.fps, ghost, slot, scene)
        self.variants.append(variant)
        return variant

    def save(self) -> None:
        """Print a summary and, with ``--record``, write the replay file."""
        for variant in self.variants:
            last = variant.metrics[-1] if variant.metrics else {}
            flag = "  (rows dropped)" if variant.overflow else ""
            print(f"{variant.label:<48} {json.dumps(last)}  {variant.wall:.2f} s{flag}")
        if not self.args.record:
            return
        source_sha = source_sha256(self.script)
        doc = {
            "script": os.path.basename(self.script),
            "script_sha256": source_sha,
            "newton_commit": _newton_commit(),
            "newton_version": newton.__version__,
            "warp_version": wp.config.version,
            "gpu": wp.get_device().name,
            "fps": self.fps,
            "variants": [v.to_json() for v in self.variants],
        }
        with open(self.args.record, "w") as f:
            json.dump(doc, f, separators=(",", ":"))
        print(f"wrote {self.args.record}")


class Variant:
    """One recorded run: a pose per frame for every body, plus named metric values."""

    def __init__(self, label, model, fps, ghost, slot, scene):
        self.label, self.model, self.fps, self.ghost, self.slot = label, model, fps, ghost, slot
        self.scene = scene
        self.poses: list[np.ndarray] = []
        self.metrics: list[dict] = []
        self.awake: list[list[int]] = []
        self.particles: list[np.ndarray] = []
        self.overflow = False
        self.start = time.perf_counter()
        self.wall = 0.0

    def frame(self, state: newton.State, awake=None, particles=None, **metrics) -> None:
        """Record ``state.body_q``, optional awake flags, particle positions and metric values."""
        self.poses.append(state.body_q.numpy().astype(np.float32).copy())
        if particles is not None:
            self.particles.append(particles.numpy().astype(np.float32).copy())
        self.metrics.append({k: _round(v) for k, v in metrics.items()})
        if awake is not None:
            self.awake.append([int(x) for x in awake.numpy()])
        self.wall = time.perf_counter() - self.start

    def check(self, solver) -> None:
        """Note whether the solver dropped rows; call once after the run."""
        try:
            solver.check_constraint_capacity()
        except RuntimeError:
            self.overflow = True

    def to_json(self) -> dict:
        poses = (
            np.stack(self.poses)
            if self.poses
            else np.zeros((0, self.model.body_count, 7), np.float32)
        )
        return {
            "label": self.label,
            "fps": self.fps,
            "ghost": self.ghost,
            "slot": self.slot,
            "scene": self.scene,
            "bodies": int(self.model.body_count),
            "frames": int(poses.shape[0]),
            "poses": base64.b64encode(poses.astype(np.float32).tobytes()).decode(),
            "shapes": _shape_table(self.model),
            "metric": self.metrics,
            "awake": self.awake or None,
            "particles": self._particles_json(),
            "wall_s": round(self.wall, 3),
            "overflow": self.overflow,
        }

    def _particles_json(self):
        if not self.particles:
            return None
        frames = np.stack(self.particles)
        tris = self.model.tri_indices
        radius = self.model.particle_radius.numpy()
        return {
            "count": int(frames.shape[1]),
            "frames": base64.b64encode(frames.astype(np.float32).tobytes()).decode(),
            "radius": round(float(radius.mean()), 5) if radius.size else 0.0,
            "tris": [int(v) for v in tris.numpy().reshape(-1)]
            if tris is not None and tris.shape[0]
            else None,
        }


def _shape_table(model):
    types = model.shape_type.numpy()
    scales = model.shape_scale.numpy()
    xforms = model.shape_transform.numpy()
    bodies = model.shape_body.numpy()
    flags = model.shape_flags.numpy()
    shapes = []
    for i, t in enumerate(types):
        if int(t) not in _RENDERED_TYPES or not flags[i] & int(newton.ShapeFlags.VISIBLE):
            continue
        shape = {
            "b": int(bodies[i]),
            "t": int(t),
            "s": [round(float(v), 5) for v in scales[i]],
            "x": [round(float(v), 6) for v in xforms[i]],
        }
        if int(t) in (newton.GeoType.MESH, newton.GeoType.CONVEX_MESH):
            mesh = model.shape_source[i]
            shape["v"] = [round(float(v), 5) for v in np.asarray(mesh.vertices).reshape(-1)]
            shape["i"] = [int(v) for v in np.asarray(mesh.indices).reshape(-1)]
        shapes.append(shape)
    return shapes


def _round(value):
    value = float(value)
    return round(value, 5) if np.isfinite(value) else None


def _newton_commit():
    if os.environ.get("NEWTON_COMMIT"):
        return os.environ["NEWTON_COMMIT"]
    try:
        root = os.path.dirname(os.path.dirname(newton.__file__))
        return subprocess.check_output(["git", "-C", root, "rev-parse", "HEAD"], text=True).strip()
    except (OSError, subprocess.CalledProcessError):
        return None

Recorded with SolverSteel on Newton 1.9.0.dev0, Warp 1.19.0, NVIDIA RTX PRO 6000 Blackwell Server Edition. Replays read the state back every frame, so they are not a speed benchmark. Script · page text

Steel docs and tutorials are licensed under Apache-2.0. Third-party notices. "Skild", "Skild AI" and the Skild logo are trademarks of Skild AI and are not licensed under Apache-2.0.