Multiple Results

Shows how a single task can produce multiple results in ArmoniK.

What it does

The client creates a session, pre-allocates N result slots (controlled by --nResultsPerTask), uploads an input string as the payload, submits one task that is expected to fill all N results, waits for all of them, then prints each one.

The worker iterates over every result ID it is expected to produce and writes "<input> World_ <resultId>" to each one.

Project structure

MultipleResults/
├── Client/
│   └── Program.cs              # Session creation, multi-result task submission, result download loop
└── Worker/
    ├── Program.cs              # Worker entry point
    └── MultipleResultsWorker.cs # WorkerStreamWrapper iterating over ExpectedResults

Key concepts

Pre-allocating multiple result IDs before task submission with CreateResultsMetaData
Passing multiple IDs in ExpectedOutputKeys when submitting a single task
Worker iterating over taskHandler.ExpectedResults to produce N outputs
Waiting for a list of results with EventsClient.WaitForResultsAsync

Running

Client

dotnet run --project Client -- \
  --endpoint http://localhost:5001 \
  --partition default \
  --input "Hello" \
  --nResultsPerTask 10

Option	Default	Description
`--endpoint`	`http://localhost:5001`	ArmoniK control plane URL
`--partition`	`default`	Partition to submit tasks to
`--input`	`Hello`	Input string sent as payload
`--nResultsPerTask`	`10`	Number of results the single task must produce

Worker

dotnet run --project Worker

The worker reads its gRPC endpoint from environment variables (standard ArmoniK worker configuration).