Class SystematicAgentBenchmark

java.lang.Object
org.episteme.benchmarks.benchmark.benchmarks.SystematicAgentBenchmark
All Implemented Interfaces:
RunnableBenchmark
@AutoService(RunnableBenchmark.class) public class SystematicAgentBenchmark extends Object implements RunnableBenchmark
Systematic Benchmark for Agent Systems. Measures the scaling capability of Virtual Thread Agents.
Author:
Silvere Martin-Michiellot, Gemini AI (Google DeepMind)

  • Constructor Details

    • SystematicAgentBenchmark

      public SystematicAgentBenchmark()

  • Method Details

    • getId

      public String getId()
      Specified by:
      getId in interface RunnableBenchmark
      Returns:
      A unique stable identifier for this benchmark (e.g. "ga-tsp-50")

    • getName

      public String getName()
      Specified by:
      getName in interface RunnableBenchmark
      Returns:
      Unique human-readable name of the benchmark

    • getDescription

      public String getDescription()
      Specified by:
      getDescription in interface RunnableBenchmark
      Returns:
      A detailed description of what the benchmark measures

    • getDomain

      public String getDomain()
      Specified by:
      getDomain in interface RunnableBenchmark
      Returns:
      The domain of this benchmark (e.g. "Linear Algebra", "Physics")

    • getAlgorithmType

      public String getAlgorithmType()
      Specified by:
      getAlgorithmType in interface RunnableBenchmark
      Returns:
      The algorithm type (e.g. "inference", "linear algebra")

    • getAlgorithmProvider

      public String getAlgorithmProvider()
      Specified by:
      getAlgorithmProvider in interface RunnableBenchmark
      Returns:
      The algorithm provider name (e.g. "Episteme", "OpenCL", "NativeBLAS") or "Standard" if not applicable.

    • isAvailable

      public boolean isAvailable()
      Description copied from interface: RunnableBenchmark
      Checks if the benchmark requirements (e.g. native libraries, hardware) are met.
      Specified by:
      isAvailable in interface RunnableBenchmark
      Returns:
      true if the benchmark can be executed

    • setup

      public void setup()
      Description copied from interface: RunnableBenchmark
      Setup method called once before the benchmark runs. Use this to initialize data structures.
      Specified by:
      setup in interface RunnableBenchmark

    • run

      public void run()
      Description copied from interface: RunnableBenchmark
      The core action to measure. This method will be called repeatedly in a loop.
      Specified by:
      run in interface RunnableBenchmark

    • teardown

      public void teardown()
      Description copied from interface: RunnableBenchmark
      Cleanup method called once after the benchmark completes.
      Specified by:
      teardown in interface RunnableBenchmark

    • getSuggestedIterations

      public int getSuggestedIterations()
      Specified by:
      getSuggestedIterations in interface RunnableBenchmark
      Returns:
      Estimated number of iterations required for a stable result