Contribution Options

Configure resource limits, scheduling, and constraints for TealeSDK contributions.

Overview

ContributionOptions controls how much of the device's resources are made available and when contributions are active. All options can be changed at runtime.

Options

let options = ContributionOptions(
    ramLimit: .percent(0.5),
    schedule: .idle,
    requireWiFi: true,
    requirePluggedIn: false,
    maxConcurrentRequests: 2,
    allowedModelFamilies: ["Llama", "Qwen"]
)

ramLimit

Controls how much RAM can be used for model loading and inference.

Value	Description
.percent(Double)	Fraction of total RAM (e.g., .percent(0.5) = 50% of total RAM)
.absoluteGB(Double)	Fixed GB limit (e.g., .absoluteGB(8) = 8 GB maximum)

Default: .percent(0.5)

The SDK will not load models that would exceed this limit. The 4 GB OS reservation is applied on top of this limit.

schedule

Controls when the device contributes to the network.

Value	Description
.always	Contribute whenever the app is running
.idle	Contribute only when the device is idle (no user activity)
.afterHours	Contribute only between 10 PM and 6 AM local time
.custom(TimeRange)	Contribute during a custom time range

Default: .idle

requireWiFi

When true, contributions pause when the device is on cellular data.

Default: true

requirePluggedIn

When true, contributions pause when the device is on battery power.

Default: false

maxConcurrentRequests

Maximum number of inference requests to serve simultaneously.

Default: 2

Higher values increase throughput but also increase memory and CPU usage. For most devices, 1-3 is appropriate.

allowedModelFamilies

Restrict which model families can be loaded. An empty array means all families are allowed.

// Only serve Llama and Qwen models
allowedModelFamilies: ["Llama", "Qwen"]

// Allow all models
allowedModelFamilies: []

Default: [] (all families)

Runtime Updates

Options can be updated after initialization:

contributor.options.ramLimit = .absoluteGB(16)
contributor.options.schedule = .always
contributor.options.requirePluggedIn = true

Changes take effect immediately. If the new options are more restrictive (e.g., lower RAM limit), currently loaded models that exceed the limit will be unloaded.

Automatic Pausing

The SDK automatically pauses contributions when:

1. Thermal throttling -- device thermal state reaches serious or critical
2. Low battery -- battery drops below 20% (when requirePluggedIn is false)
3. Schedule -- outside the configured schedule window
4. Network -- on cellular when requireWiFi is true

When conditions improve, contributions resume automatically. The contributor state changes to paused during these periods.