Struct RunLog

Nested Relationships

struct RunLog

A collection of LogFrame’s related to a single model run The data available depends on the LoggingConfig used at runtime

Public Functions

inline RunLog(const ExitLogFrame &_exit, const std::list<StepLogFrame> &_step)

Constructs a RunLog from existing data frames

Parameters:

inline const ExitLogFrame &getExitLog() const

Return the exit LogFrame

Returns:: The logging information collected after completion of the model run

inline const std::list<StepLogFrame> &getStepLog() const

Return the ordered list of step LogFrames

See also

Note

If logging frequency was changed in the StepLoggingConfig, there may be less than 1 LogFrame per step.

inline uint64_t getRandomSeed() const: Returns the random seed used for this run

inline unsigned int getStepLogFrequency() const: Returns the frequency that steps were logged

Note

This value is configured via StepLoggingConfig::setFrequency()

inline PerformanceSpecs getPerformanceSpecs() const: Return a copy of a structure containing performance relevant device and software information

Friends

struct PerformanceSpecs

Public Members

std::string device_name: The name of the selected CUDA device, as returned by cudaGetDeviceProperties()

int device_cc_major: The selected CUDA device’s compute capability major version number

int device_cc_minor: The selected CUDA device’s compute capability minor version number

int cuda_version: The version number of the current CUDA Runtime instance. The version is returned as (1000 major + 10 minor). For example, CUDA 9.2 would be represented by 9020.

Note

Uses cudaRuntimeGetVersion()

bool seatbelts: True if FLAME GPU was built with FLAMEGPU_SEATBELTS enabled at CMake configure time For maximum performance, this value should be false

std::string flamegpu_version

Full FLAMEGPU version string