roadmap: restore Motrix as a tensor-only Manager HOST_BRIDGE backend for canonical G1 training

Roadmap: restore Motrix as a tensor-only Manager HOST_BRIDGE backend for canonical G1 training

Follow-up to #2045 and #2048 with the same governance and acceptance discipline. The enabled physics backend is Motrix; the reference runtime boundary is MuJoCo/mjbatch. Other shelved adapters remain out of scope.

Goal

Restore Motrix as an active tensor-only Manager-Based HOST_BRIDGE backend for exactly two canonical training workloads:

SAC / g1_walk_flat / Motrix
FlashSAC / g1_motion_tracking / Motrix

MotrixSim remains CPU-authoritative. This roadmap does not attempt to make it DEVICE_RESIDENT. The D2H/H2D host bridge is a legitimate public boundary, but every semantic crossing must be explicit, packed, countable, and bounded.

Development and validation occur on the local workstation. Server deployment is out of scope.

Development governance (non-negotiable)

Manager-Based API is the sole training entry point.
- Do not restore direct mode.
- Do not restore task-owned direct runtime registration.
Do not restore the generic NumPy Manager.
- No generic NumPy owner/composer fallback.
Do not restore the NumPy RNG owner.
- Manager-owned Torch RNG remains authoritative.
Do not restore env.tensor_runtime / env.tensor_runtime_device.
- Placement derives from public env carriers and declared backend capabilities.
No backend-private probing from task/env/training code.
- Consume only declared unisim.backend.base.SimBackend APIs/capabilities.
- Add a public unisim contract first when a capability is absent.
Motrix is a CPU-authoritative packed HOST_BRIDGE.
- No hidden NumPy state mirror outside the backend-owned host bridge.
- No hidden task-side host reset composer.
- Each control/reset/read phase owns a bounded, countable device crossing.
- Do not masquerade HOST_BRIDGE as DEVICE_RESIDENT.
Unsupported capabilities fail closed.
- Tensor reset randomization and fixed variants remain disabled unless a public contract is added.
- No silent host fallback to satisfy an owner.
Historical Motrix owners are reference evidence only.
- Do not restore the old reward-retuned SAC Motrix owner.
- Do not restore the task-owned FlashSAC Motrix host-bridge runtime.
Sibling changes require their own reviewable PRs and evidence.
Do not modify main; do not publish PyPI artifacts.
Do not weaken the perf: end-to-end MJWarp training throughput gate and tensor-native closeout #2023 measurement contract.
- run_summary.json:tail_env_steps_per_sec remains decisive.
- Transfer counters and benchmarks remain diagnostic-only.

Current audit

Runtime scope

Current active tensor Manager scope:

mujoco, mjwarp, genesis, newton

Motrix remains shelved:

motrix, drake, isaacgym, isaacsim, superdex

Evidence:

src/unilab/base/registry.py keeps motrix in _SHELVED_SIM_BACKENDS.
src/unilab/cli.py keeps Motrix out of SUPPORTED_SIMS.
register_env(..., sim_backend="motrix") currently fails closed.
--sim motrix is rejected before dependency detection.

Task registration and owners

No canonical Motrix factory is registered:

G1WalkFlat: mujoco, mjwarp, genesis, newton
G1MotionTrackingSAC: mujoco, mjwarp, genesis, newton

No Motrix owner YAML exists under src/unilab/conf.

The historical owners removed in 402aed69 are not directly restorable:

SAC Motrix retuned reward shaping and predates fused tensor reward/reset owners.
FlashSAC Motrix was a MuJoCo-checkpoint sim2sim/eval owner, not a canonical training owner.
An older task-owned FlashSAC host-bridge factory violates current governance.

Local runtime

After uv sync --extra motrix:

motrixsim-core 0.10.1
motrixsim       installed

Focused existing Motrix backend smokes pass, including construction, stepping, state reads, body state, and quaternion convention. Existing unisim packed host-bridge adapter tests pass.

Existing public tensor lifecycle

The unisim Motrix backend already declares:

execution             HOST_BRIDGE
state_views           true
state_fields          {qpos, qvel, ctrl}
sensor_views          true
stepping              true
selected_reset        true
reset_randomization   false
fixed_variants        false
host_pre_step_control false
packed_host_bridge    true
process_topology      IN_PROCESS
data_plane            HOST_BRIDGE
torch_devices         {cpu, cuda}

Public APIs already present:

compile_host_bridge_io
step_tensor
set_state_tensor
get_state_views
get_sensor_view
get_tensor_capabilities

The packed plan already owns stable host/device buffers and semantic transfer counters:

control: one packed D2H;
selected reset: one packed D2H;
full state/sensor read: one packed H2D;
post-reset selected publication: one packed H2D.

Existing adapter coverage proves selected-row preservation, bounded validation, no hidden CUDA-to-CPU detours, portable parity, and fail-closed fixed variants/randomization.

Gap 1 — Public state-width contract

Motrix does not implement get_public_state_widths():

MuJoCo/mjbatch exposes PublicStateWidths(nq, nv).
Motrix tensor layout currently infers widths by reading full public state.
Canonical tensor reset must consume an explicit public width contract.

This must be repaired in unisim before owner enablement.

Gap 2 — Selected-reset publication semantics

MuJoCo/mjbatch declares:

selected_reset_publication=AUTHORITATIVE_VIEWS

Motrix cannot truthfully declare that today because:

set_state_tensor commits selected rows through the CPU bridge;
device views become authoritative only after the plan's explicit selected-row H2D publication.

Do not add a false capability declaration. The initial owner must preserve the paired reset/read boundary. A stronger public reset transaction/readiness declaration is optional follow-up work and requires its own contract tests.

Gap 3 — Selected reset host commit is not a compiled transaction

The non-portable Motrix set_state path currently:

constructs a full-width boolean mask;
slices native data;
resets defaults;
clears forces;
writes qvel/qpos/control through multiple Python/native calls;
runs FK;
refreshes caches.

The portable path uses disjoint indices but still makes multiple native calls rather than one compiled selected-row reset program.

MotrixLab provides a public reference pattern:

model.compile_write(..., reset=True, forward_kinematic=True)
program.execute(data, env_ids)

The equivalent semantics must live inside the unisim Motrix host boundary; UniLab must not call MotrixLab APIs directly.

Gap 4 — Selected read computes full-batch sources

Current selected packed reads:

read full qpos/qvel;
compute full tracked-body state;
read full native sensor batches;
select rows on the host;
publish selected packet H2D.

This preserves semantics but can dominate motion tracking with many tracked bodies. MotrixLab's compiled query program supports selected-row execution and stable views:

query_program.execute(source, env_ids=selected_rows)

The backend should adopt the same public-shaped pattern internally without leaking MotrixLab manager architecture into UniLab.

Gap 5 — Reset randomization and fixed variants

Motrix tensor reset rejects randomization. The canonical SAC base still declares pd_gains; the owner must explicitly set it to null. Fixed variants must remain absent. Any future packed randomization payload is a separate unisim public contract, not a host fallback.

Gap 6 — Canonical owner tests and support evidence

Existing tests cover the adapter boundary, not canonical Manager training. Needed:

owner materialization;
unsupported reset randomization disabled explicitly;
real construction/reset/rollout;
selected reset/read row identity;
fixed boundary counts;
no task-side backend-private access;
support matrix/docs/CI updates.

Gap 7 — Diagnostic benchmark scope

The G1 FlashSAC backend benchmark currently rejects Motrix as shelved. It should accept Motrix as diagnostic-only and report boundary timing/counters. End-to-end training remains decisive.

Reference implementation boundary

/home/user/ws/simulator/MotrixLab is useful evidence for:

compiled reset/write programs with selected rows;
compiled read query programs with selected rows;
stable program-owned buffers/views;
task-level G1 walk and WBT semantics.

It must not be imported by UniLab task/runtime code. Its direct/NumPy manager architecture is not a governance model for this roadmap.

Proposed phases

Phase 0 — Contract and boundary baseline

Freeze the canonical Motrix D2H/H2D measurement contract.
Record current packed control/reset/full-read/selected-read counters and timing on the canonical G1 scenes.
Establish MuJoCo/mjbatch and Motrix interleaved diagnostic baselines.
Create the integration branch from develop/tensor-runtime.

Exit: every canonical requirement maps to an existing public API or a concrete unisim gap.

Phase 1 — Public unisim contract parity

Minimum changes:

Implement Motrix get_public_state_widths().
Add explicit tests proving public widths match canonical tensor reset layout.
Document the initial selected-reset publication boundary: commit + explicit selected H2D read, not AUTHORITATIVE_VIEWS.
Strengthen fail-closed tests for randomization/fixed variants.

Exit: canonical reset composition no longer depends on backend-private state-shape inference.

Phase 2 — HOST_BRIDGE optimization

Maintain the existing path as the A/B baseline.

Then optimize only where evidence attributes cost:

Path B — compiled selected reset: replace the multi-call selected reset with one compiled reset/write transaction when MotrixSim supports the required semantics.
Path C — compiled selected read: refresh only selected rows and preserve stable packet views instead of computing full body/sensor sources for selected publication.
Path D — plan ownership audit: inspect direct/reset and scene read plans for duplicate buffers or synchronization. Unify only if measured evidence shows hot-path cost or split reset state.

Acceptance for every optimization:

selected rows update correctly;
untouched rows remain unchanged;
semantic D2H/H2D counts do not increase;
no hidden CPU detour on a CUDA hot path;
portable and non-portable parity where applicable;
diagnostic timing/counters improve without semantic loss.

Phase 3 — UniLab canonical owner enablement

Only after Phase 1:

Move Motrix into active registry/CLI scope.
Add fail-closed motrixsim dependency diagnostics.
Register only the two canonical factories through the current generic Manager factory.
Add canonical owner YAMLs:
- SAC g1_walk_flat/motrix.yaml
- FlashSAC g1_motion_tracking/motrix.yaml
SAC owner explicitly disables pd_gains.
Both owners avoid fixed variants and unsupported randomization.
Restore Motrix to support matrix, benchmark diagnostic scope, docs, ADR, and gated CI.
Add real CUDA/CPU manager lifecycle tests for both canonical owners.

Exit: both canonical train commands construct, reset, train, and shut down normally through Manager-Based API only.

Phase 4 — Local performance convergence

Commands:

uv run --extra motrix train \
  --algo sac \
  --task g1_walk_flat \
  --sim motrix \
  algo.max_iterations=160 \
  algo.num_envs=2048 \
  training.no_play=true \
  training.export_onnx=false

uv run --extra motrix train \
  --algo flashsac \
  --task g1_motion_tracking \
  --sim motrix \
  algo.max_iterations=160 \
  algo.num_envs=2048 \
  training.no_play=true \
  training.export_onnx=false

Reference:

uv run --extra mujoco train --algo sac --task g1_walk_flat --sim mujoco ...
uv run --extra mujoco train --algo flashsac --task g1_motion_tracking --sim mujoco ...

Measure:

run_summary.json:tail_env_steps_per_sec

At least three interleaved runs per workload; report every run.

Initial gates, subject to the first honest Phase 4 baseline:

Workload	Initial Motrix gate
SAC / `g1_walk_flat`	≥35% of MuJoCo/mjbatch tail throughput
FlashSAC / `g1_motion_tracking`	≥35% of MuJoCo/mjbatch tail throughput

Optimization target: ≥50%. Stretch: ≥60%.

Motrix is not required to match mjbatch exactly, but acceptance requires attribution showing that the remaining gap is dominated by MotrixSim native CPU physics rather than avoidable Python, reset, selected-read, or transfer-boundary work. If the first baseline materially exceeds the initial gate, raise the floor before optimization.

Every run must:

complete 160/160;
report normal shutdown;
leave no residual collector/worker process;
produce finite rewards/observations;
preserve observation/action dimensions;
write tail_iteration_count=20;
report stable boundary counts.

Phase 5 — Closeout

make check
make test
make test-all
unisim and UniLab PRs with focused evidence;
final interleaved local evidence table;
support matrix/docs/CI update;
final merge into develop/tensor-runtime;
explicit unsupported-capability statement.

Explicit non-goals

Full mjbatch throughput parity.
Motrix DEVICE_RESIDENT execution.
Tensor reset randomization parity.
Fixed-variant support.
Other Motrix tasks/owners.
Server deployment.
Playback/rendering parity beyond safe record/native validation.
Restoring direct mode, task-owned runtimes, generic NumPy Manager, NumPy RNG, or tensor_runtime switches.

Phase 0 integration branch created: dev/issue-2054-motrix-tensor-runtime from develop/tensor-runtime @ 90068b82.

feat(motrix): publish public tensor state widths (#339)

Phase 0 complete; Phase 1 public-width contract merged.

Integration branch: dev/issue-2054-motrix-tensor-runtime.

Phase 0 audit conclusions

Motrix is blocked at UniLab registry/CLI/owner layer.
unisim already has a usable packed HOST_BRIDGE lifecycle:
- one packed control D2H;
- one packed selected-reset D2H;
- one packed full state/sensor H2D;
- one packed post-reset selected-row H2D.
Existing unisim tests cover transfer counts, row preservation, bounded validation, and no hidden CUDA→CPU detours.
Historical Motrix owners are not restorable:
- SAC owner predates the fused tensor reward/reset owners and retunes rewards;
- FlashSAC owner was a sim2sim eval owner, not canonical training;
- the old task-owned host-bridge runtime violates current governance.
Main performance gaps are backend-internal, not task-layer:
- selected reset is a multi-call host transaction, not a compiled selected-row write;
- selected read computes full state/body/sensor sources before selecting rows;
- reset/read plans may duplicate ownership and buffers.
MotrixLab is useful as an implementation reference for compiled selected-row write/query programs, but its direct/NumPy manager architecture must not enter UniLab.

Phase 1 merged

unisim PR #339 → roadmap/issue-1811-tensor-manager, merge cbcb3c6.

Adds:

MotrixBackend.get_public_state_widths() -> PublicStateWidths(nq, nv)

and proves public widths match public state layout and packed reset validation.

All remote CI passed:

Python 3.10 Linux/macOS/Windows
mypy + pyright
source distribution

The selected-reset publication boundary remains explicitly paired commit + selected H2D read; Motrix does not falsely declare AUTHORITATIVE_VIEWS.

Current heads

Repo	Branch	Commit
UniLab	`dev/issue-2054-motrix-tensor-runtime`	`90068b82`
unisim	`roadmap/issue-1811-tensor-manager`	`cbcb3c6`

Next: Phase 2 boundary baseline and A/B attribution on the canonical G1 scenes.

perf(motrix): read only selected packed rows (#340)

Phase 2 Path C merged: selected packed publication no longer computes unselected rows.

unisim PR #340 → roadmap/issue-1811-tensor-manager, merge bc61643.

Before:

selected publication assembled full public qpos/qvel;
then full tracked-body state;
then full native sensor batches;
only afterwards selected reset rows.

After:

selected qpos/qvel assembled for reset rows only;
selected tracked-body state uses a new fused selected-row/selected-link read;
native sensors use the existing row-local API;
the same stable selected host/device packet and one H2D scatter remain.

Transfer counts and stable views are unchanged. This is a host-boundary compute reduction, not an additional or hidden transfer.

New contract tests:

full-batch get_state / get_body_state_w are rejected during selected publication;
selected world-frame body aliases match authoritative full world reads after reset.

Remote CI: Linux/macOS/Windows Python 3.10, type checks, and source distribution all passed.

Current unisim head: bc61643.

Phase 2 boundary attribution on canonical G1 flat scene (2048 envs).

Local diagnostic measurements after unisim #339/#340. Values are mean over 30 iterations after 5 warmups, unless noted.

Selected packed reset + selected publication

Rows	MuJoCo/mjbatch	Motrix	Motrix / MuJoCo
1	0.495 ms	0.668 ms	74%
128 (1/16 batch)	1.028 ms	1.017 ms	101%
2048	9.360 ms	3.530 ms	265%

At the canonical 1/16 reset cadence, Motrix selected reset/publication is already at parity with mjbatch. Full-batch reset is faster on Motrix, while sparse resets are comparable.

Control + 3 physics substeps + one packed full read

Backend	Mean	Median
MuJoCo/mjbatch	13.979 ms	13.885 ms
Motrix	26.177 ms	25.876 ms

Ratio: 53.4% of MuJoCo/mjbatch.

This cycle dominates every control step, while selected reset/publication is a small phase at the 1/16 cadence. Therefore the remaining Motrix gap is primarily in MotrixSim CPU physics/native full-read work, not D2H/H2D or selected reset orchestration.

Host reset sub-attribution (Motrix, 2048 envs, 128 reset rows)

Mean selected set_state wall time: 0.346 ms.

Largest sub-phases:

Sub-phase	Mean
native data slice	0.129 ms
native selected data reset	0.064 ms
forward kinematics	0.050 ms
link pose cache refresh	0.036 ms
actuator control hold	0.021 ms
dof position write	0.016 ms
dof velocity write	0.015 ms
qpos conversion	0.007 ms
mask	0.002 ms

Interpretation

Selected read optimization (docs: add FlashSAC to support matrix and algorithms doc #340) removed full-batch host computation from selected publication.
Reset orchestration is not the dominant cost at canonical cadence.
The 46.6% cycle gap is concentrated in the control-step/full-read path:
- MotrixSim native 3-substep CPU physics;
- full state/sensor assembly and the packed H2D boundary.
A compiled reset transaction would target a phase already around 1 ms per canonical reset cycle and is therefore lower priority than control-step/full-read work.

Instrument the Motrix control/full-read cycle into write_control, native step_n, state/sensor pack, and H2D.
Determine whether full read duplicates native state/sensor reads already performed by MotrixSim and eliminate only proven duplication.
If the remaining cost is native MotrixSim physics, record that boundary and proceed to UniLab owner enablement; do not trade contract strength for speculative host optimization.

Phase 2 control-cycle attribution refinement.

The same canonical G1 flat scene, 2048 envs, three physics substeps, and one packed full read were decomposed after the aggregate measurement.

Motrix

Phase	Mean	Median
packed control D2H	0.046 ms	0.044 ms
control submission to native data	0.068 ms	0.063 ms
MotrixSim `step_n(..., 3)`	24.787 ms	24.335 ms
link pose refresh + velocity invalidation	0.111 ms	0.112 ms
state/sensor host pack	0.742 ms	0.657 ms
packed H2D	0.060 ms	0.052 ms

MuJoCo/mjbatch comparable boundary

Phase	Mean	Median
packed control D2H	0.044 ms	0.041 ms
backend step + native bridge	13.799 ms	13.726 ms
packed full read	0.129 ms	0.110 ms

Conclusion

Transfer boundaries are already negligible:

Motrix control D2H: ~0.046 ms
Motrix packed full H2D: ~0.060 ms

The aggregate Motrix/MuJoCo ratio of ~53.4% is overwhelmingly explained by native MotrixSim CPU physics: ~24.8 ms vs the complete MuJoCo backend step boundary of ~13.8 ms. Python orchestration and explicit D2H/H2D account for under 1 ms.

The remaining host-pack cost is ~0.74 ms. It is the only non-native optimization candidate, and even eliminating it entirely would change the ratio by roughly 2%, below the measured native physics gap.

Boundary decision: continue contract-first UniLab owner enablement rather than speculative host rewrite. Further Motrix throughput work belongs primarily in MotrixSim native physics, not the packed HOST_BRIDGE contract.

feat(motrix): restore canonical packed host-bridge owners (#2056)

Phase 3 complete: canonical Motrix owners merged.

PR #2056 → dev/issue-2054-motrix-tensor-runtime, merge 60ab8473.

Enabled surface

Exactly two canonical workloads:

uv run --extra motrix train --algo sac --task g1_walk_flat --sim motrix
uv run --extra motrix train --algo flashsac --task g1_motion_tracking --sim motrix

Other Motrix owners remain out of scope and stay Registered, not production claims.

Owner decisions

Active tensor Manager scope now includes Motrix alongside MuJoCo/MJWarp/Genesis/Newton.
CLI fails closed when motrixsim is absent.
Both factories use the current generic Manager-Based runtime.
SAC walk owner:
- tensor reset events;
- G1WalkResetOwner;
- fused walk reward pack;
- fused termination pack;
- events.pd_gains: null.
FlashSAC motion owner:
- TensorMotionCommandCfg;
- MotionResetOwner owning command/action/observation/metric reset;
- no unsupported reset randomization.
Neither owner uses fixed variants.
Packed HOST_BRIDGE reset selection now recognizes event-manager tensor terms, enabling scoped_device_event_tensor_with_host_commit for tensor reset events.

Validation

make check: pass.
Focused registry/CLI/config/support/benchmark tests: pass.
Real CUDA Manager rollouts:
- SAC walk: finite 5-step rollout; obs/critic 98/101.
- FlashSAC motion: finite 3-step rollout; obs/critic 160/289.
End-to-end 256-env × 3-iteration smokes:
- SAC walk: completed, normal shutdown, collector exit 0, finite rewards.
- FlashSAC motion: completed, normal shutdown, collector exit 0, finite rewards.
No residual training/Motrix process.
No direct mode, generic NumPy Manager, NumPy RNG owner, tensor-runtime switch, hidden reset composer, or backend-private task probing.

Current heads

Repo	Branch	Commit
UniLab	`dev/issue-2054-motrix-tensor-runtime`	`60ab8473`
unisim	`roadmap/issue-1811-tensor-manager`	`bc61643`

Next: Phase 4 local 160×2048 ×3 interleaved throughput gate against MuJoCo/mjbatch.

Phase 4 first measured baseline.

Both Motrix runs completed 160/160 at 2048 envs on the local workstation, with no play/ONNX. MuJoCo/mjbatch reference runs used the same shape and measurement contract.

Workload	Backend	Status	Iter	Tail steps/s	Tail window	Shutdown	Collector	Ratio
SAC / `g1_walk_flat`	Motrix	completed	160/160	61,117.173	20	normal_completion	exit 0	64.76%
SAC / `g1_walk_flat`	MuJoCo/mjbatch	completed	160/160	94,377.964	20	normal_completion	exit 0	—
FlashSAC / `g1_motion_tracking`	Motrix	completed	160/160	56,626.750	20	normal_completion	exit 0	64.63%
FlashSAC / `g1_motion_tracking`	MuJoCo/mjbatch	completed	160/160	87,622.321	20	normal_completion	exit 0	—

No residual training/Motrix process remained after the runs.

Both Motrix workloads exceed the initial 35% floor and the 50% optimization target, and nearly reach the 60% stretch gate. This is consistent with the Phase 2 attribution: explicit D2H/H2D is ~0.1 ms/control step, while the remaining gap is dominated by MotrixSim native step_n CPU physics.

Final acceptance still requires three interleaved Motrix runs per workload, reporting every run. The MuJoCo references will also be interleaved with those acceptance runs.

Phase 4 interleaved acceptance evidence (3 Motrix runs per workload, MuJoCo/mjbatch references interleaved).

Interleaved order for each of rounds 1–3: SAC Motrix → SAC MuJoCo → FlashSAC Motrix → FlashSAC MuJoCo.
Every run: 160×2048, no play/ONNX, tail_iteration_count=20, normal shutdown, collector exit 0.

SAC / `g1_walk_flat`

Backend	Run	Tail steps/s
Motrix	`2026-10-04_14-21-50_motrix`	61,117.173
Motrix	`2026-10-04_14-25-15_motrix`	63,204.377
Motrix	`2026-10-04_14-26-33_motrix`	63,328.873
MuJoCo	`2026-10-04_14-23-25_mujoco`	94,377.964
MuJoCo	`2026-10-04_14-25-34_mujoco`	94,445.013
MuJoCo	`2026-10-04_14-26-53_mujoco`	90,992.450

Motrix min/median/max: 61,117.173 / 63,204.377 / 63,328.873
MuJoCo min/median/max: 90,992.450 / 94,377.964 / 94,445.013
Worst-run ratio: 67.17%

FlashSAC / `g1_motion_tracking`

Backend	Run	Tail steps/s
Motrix	`2026-10-04_14-22-36_motrix`	56,626.750
Motrix	`2026-10-04_14-25-50_motrix`	55,490.442
Motrix	`2026-10-04_14-27-09_motrix`	56,342.518
MuJoCo	`2026-10-04_14-23-54_mujoco`	87,622.321
MuJoCo	`2026-10-04_14-26-14_mujoco`	87,084.407
MuJoCo	`2026-10-04_14-27-32_mujoco`	90,448.674

Motrix min/median/max: 55,490.442 / 56,342.518 / 56,626.750
MuJoCo min/median/max: 87,084.407 / 87,622.321 / 90,448.674
Worst-run ratio: 63.72%

Gate result

Gate	SAC walk	FlashSAC motion
≥35% MuJoCo initial gate	pass	pass
≥50% optimization target	pass	pass
≥60% stretch	pass	pass

All runs completed 160/160 with finite training state and no residual collector/Motrix process after the sequence.

Phase 2 attribution remains controlling: explicit D2H/H2D is approximately 0.1 ms/control step, selected reset is at parity at the canonical 1/16 cadence, and the remaining gap is dominated by MotrixSim native step_n CPU physics. No contract weakening or speculative host rewrite is warranted.

test(motrix): align flashsac owner scope with canonical runtime (#2057)

test(motrix): scope tensor event assertion to Motrix owner (#2058)

test(support): allow absent Motrix support cells (#2059)

Phase 5 closeout validation and integration state.

Final integration head: dev/issue-2054-motrix-tensor-runtime @ 44207bfd.

Commands and outcomes:

make check — pass.
make test — 1,863 passed / 20 skipped / 0 failed.
make test-all — pass, including benchmark import smoke (35/35 module mode, 36/36 script mode).
Focused slow canonical contract gate — 129 passed / 8 skipped.
The initial slow run exposed three stale test expectations from before Motrix enablement:
- Motrix must be present in the FlashSAC Manager registry;
- the tensor reset-event assertion applies to the Motrix owner, not MuJoCo owners;
- non-canonical Motrix support cells are MISSING rather than absent from the generated matrix.
Those corrections were merged in test(motrix): align flashsac owner scope with canonical runtime #2057, test(motrix): scope tensor event assertion to Motrix owner #2058, and test(support): allow absent Motrix support cells #2059, then the final validation above was rerun with MuJoCo/MJWarp/Motrix/Newton extras installed.

Merged UniLab PRs on the integration branch:

feat(motrix): restore canonical packed host-bridge owners #2056 feat(motrix): restore canonical packed host-bridge owners → 60ab8473
test(motrix): align flashsac owner scope with canonical runtime #2057 test(motrix): align flashsac owner scope with canonical runtime → 1a468e0d
test(motrix): scope tensor event assertion to Motrix owner #2058 test(motrix): scope tensor event assertion to Motrix owner → 84cb0bcb
test(support): allow absent Motrix support cells #2059 test(support): allow absent Motrix support cells → 44207bfd

No main mutation and no PyPI release.

roadmap: restore Motrix canonical packed host-bridge runtime (#2060)

Completed by #2060. Final integration includes #2056, #2057, #2058, and #2059; unisim prerequisites #339/#340 are included through the pinned roadmap workspace. All phase evidence and gates are recorded in this issue.

merge: promote tensor-manager contracts to tensor runtime (#343)

roadmap: restore Motrix as a tensor-only Manager HOST_BRIDGE backend for canonical G1 training #2054

Description

Roadmap: restore Motrix as a tensor-only Manager HOST_BRIDGE backend for canonical G1 training

Goal

Development governance (non-negotiable)

Current audit

Runtime scope

Task registration and owners

Local runtime

Existing public tensor lifecycle

Gap 1 — Public state-width contract

Gap 2 — Selected-reset publication semantics

Gap 3 — Selected reset host commit is not a compiled transaction

Gap 4 — Selected read computes full-batch sources

Gap 5 — Reset randomization and fixed variants

Gap 6 — Canonical owner tests and support evidence

Gap 7 — Diagnostic benchmark scope

Reference implementation boundary

Proposed phases

Phase 0 — Contract and boundary baseline

Phase 1 — Public unisim contract parity

Phase 2 — HOST_BRIDGE optimization

Phase 3 — UniLab canonical owner enablement

Phase 4 — Local performance convergence

Phase 5 — Closeout

Explicit non-goals

Activity

TATP-233 commented on Oct 4, 2026

TATP-233 commented on Oct 4, 2026

Phase 0 audit conclusions

Phase 1 merged

Current heads

TATP-233 commented on Oct 4, 2026

TATP-233 commented on Oct 4, 2026

Selected packed reset + selected publication

Control + 3 physics substeps + one packed full read

Host reset sub-attribution (Motrix, 2048 envs, 128 reset rows)

Interpretation

Next

TATP-233 commented on Oct 4, 2026

Motrix

MuJoCo/mjbatch comparable boundary

Conclusion

TATP-233 commented on Oct 4, 2026

Enabled surface

Owner decisions

Validation

Current heads

TATP-233 commented on Oct 4, 2026

TATP-233 commented on Oct 4, 2026

SAC / g1_walk_flat

FlashSAC / g1_motion_tracking

Gate result

TATP-233 commented on Oct 4, 2026

TATP-233 commented on Oct 4, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

SAC / `g1_walk_flat`

FlashSAC / `g1_motion_tracking`