Expand description
Modeled resident tensor compute provider.
The modeled provider is a deterministic TensorExecutor for tests,
recipes, and future hardware-provider conformance. It executes through the
published sim-lib-numbers-tensor CPU semantics, then wraps successful
results in resident storage owned by site/compute/model.
Structs§
- Compute
Model Lib - Loadable library that registers the modeled compute site.
- Modeled
Compute Profile - Configuration for a modeled compute site.
- Modeled
Compute Snapshot - Snapshot of modeled executor counters.
- Modeled
Resident Segment - One resident segment in the modeled storage arena.
- Modeled
Resident Storage - Resident storage that models readback into host tensor cells.
- Modeled
Tensor Executor - Deterministic tensor executor that models resident compute placement.
- Resident
Handle - Opaque resident allocation handle owned by the modeled site.
- Resident
OdeExecution - Result plus provider counters for a resident ODE execution.
- Resident
OdeExecutor - Executes checked tensor ODE plans through an active tensor executor.
- Resident
OdePlan - Checked resident ODE plan.
Enums§
- Modeled
Compute Fault - Fault injected into the modeled executor or resident storage.
- Resident
OdeKind - ODE execution family requested by the resident adapter.
- Resident
RhsLowering - Provider-side lowering classification for a tensor RHS.
Statics§
- RECIPES
- Cookbook recipes for this lib, embedded at build time.
Functions§
- compute_
model_ lib_ symbol - Runtime library symbol for the modeled compute provider.
- compute_
model_ site_ symbol - Site symbol exported by the modeled compute provider.
- modeled_
executor_ symbol - Stable symbol for the modeled tensor executor.