norse.benchmark.benchmark module

class norse.benchmark.benchmark.BenchmarkConfig(batch_size: int, device: str, dt: float, label: str, runs: int, profile: bool, sequence_length: int, start: int, stop: int, step: int)[source]

Bases: tuple

Benchmark configurations for the benchmarking setup

Create new instance of BenchmarkConfig(batch_size, device, dt, label, runs, profile, sequence_length, start, stop, step)

batch_size: int

Alias for field number 0

device: str

Alias for field number 1

dt: float

Alias for field number 2

label: str

Alias for field number 3

profile: bool

Alias for field number 5

runs: int

Alias for field number 4

sequence_length: int

Alias for field number 6

start: int

Alias for field number 7

step: int

Alias for field number 9

stop: int

Alias for field number 8

class norse.benchmark.benchmark.BenchmarkData(config: norse.benchmark.benchmark.BenchmarkConfig, durations: numpy.ndarray, parameters: norse.benchmark.benchmark.BenchmarkParameters)[source]

Bases: tuple

Result from a benchmark run N number of times

Create new instance of BenchmarkData(config, durations, parameters)

config: norse.benchmark.benchmark.BenchmarkConfig

Alias for field number 0

durations: numpy.ndarray

Alias for field number 1

parameters: norse.benchmark.benchmark.BenchmarkParameters

Alias for field number 2

class norse.benchmark.benchmark.BenchmarkParameters(device: str, dt: float, features: int, batch_size: int, sequence_length: int)[source]

Bases: tuple

Benchmark parameters used as input for networks

Create new instance of BenchmarkParameters(device, dt, features, batch_size, sequence_length)

batch_size: int

Alias for field number 3

device: str

Alias for field number 0

dt: float

Alias for field number 1

features: int

Alias for field number 2

sequence_length: int

Alias for field number 4