Skip to content

Releases: arm/ai-ml-sdk-scenario-runner

Scenario Runner v0.7.0

22 Oct 14:05

Choose a tag to compare

Version 0.7.0 – Initial Public Release

Purpose

Executes ML workloads for functional validation and performance
exploration
.

Features

Execution

  • Workload Execution: Runs shader and neural network graph workloads through
    both Vulkan® core compute and Vulkan® ML extensions for comprehensive ML
    workload testing
  • Complete Workflow: Loads input stimuli, executes computational graphs, and
    writes output artifacts - providing end-to-end scenario execution
  • Validation & Benchmarking: Serves as a validation vehicle for drivers and
    devices, plus performance exploration and benchmarking capabilities

Test description

  • Declarative Scenarios: Test cases and scenarios defined in JSON format for
    easy configuration, repeatability, and programmatic test generation
  • Several types of resources: Supports tensors, images, buffers, etc. so
    scenarios can exercise varied input/output.

Platform Support

The following platform combinations are supported:

  • Linux - X86-64
  • Windows® - X86-64