mirror of
https://github.com/harvard-edge/cs249r_book.git
synced 2026-07-22 15:11:24 -05:00
Vijay Janapa Reddi
cb0ae4082f
Run quartodoc build against the refactored package (engine/ extracted from core/, infra->infrastructure): regenerate the API reference so every stub points at the new paths. Adds engine.*/infrastructure.* stubs, removes the 41 stale core.solver.* / core.engine.* / core.scenarios.* / core.config.* / core.evaluation.* / infra*.qmd orphans, and refreshes the package pages (core now primitives-only) + index. Verified: all 47 documented symbols resolve in the new package; every index link resolves; zero stale mlsysim.core.<engine-mod> / mlsysim.infra references anywhere in docs/. (objects.json inventory is gitignored -- regenerated at build.)
37 lines
1.0 KiB
Plaintext
37 lines
1.0 KiB
Plaintext
# engine.solver.NetworkRooflineModel { #mlsysim.engine.solver.NetworkRooflineModel }
|
|
|
|
```python
|
|
engine.solver.NetworkRooflineModel()
|
|
```
|
|
|
|
Analyzes the Distributed Performance Bounds (The Network Wall).
|
|
|
|
This model elevates the single-node Roofline analysis to the fleet scale.
|
|
It calculates the Communication Intensity (CI) of a workload and identifies
|
|
if the fleet is Compute-Bound (Wall 1) or Network-Bound (Wall 2).
|
|
|
|
Literature Source:
|
|
1. Reddi et al. (2025), "Machine Learning Systems," Volume 2, Chapter 1.
|
|
2. Williams et al. (2009), "Roofline Model." (Theoretical basis)
|
|
3. Ghose et al. (2019), "A Survey of Communication-Efficient Distributed
|
|
Training."
|
|
|
|
## Methods
|
|
|
|
| Name | Description |
|
|
| --- | --- |
|
|
| [solve](#mlsysim.engine.solver.NetworkRooflineModel.solve) | Solves for the distributed performance bound. |
|
|
|
|
### solve { #mlsysim.engine.solver.NetworkRooflineModel.solve }
|
|
|
|
```python
|
|
engine.solver.NetworkRooflineModel.solve(
|
|
model,
|
|
fleet,
|
|
precision='fp16',
|
|
efficiency=0.5,
|
|
)
|
|
```
|
|
|
|
Solves for the distributed performance bound.
|