triton/tune/pysrc/optimize.py

import array, random, itertools
import deap.tools
import numpy as np

from genetic import GeneticOperators

def genetic(symbolic, Template, compute_perf, perf_metric, out):
    GA = GeneticOperators(symbolic, Template, out)
    return GA.optimize(maxtime='5m0s', maxgen=10000, compute_perf=compute_perf, perf_metric=perf_metric)
More packaging ... 2014-10-16 06:57:38 -04:00			`import array, random, itertools`
Added exhaustive search backend 2014-09-11 16:13:46 -04:00			`import deap.tools`
More packaging ... 2014-10-16 06:57:38 -04:00			`import numpy as np`
Auto-tuner : Initial push 2014-09-02 22:03:20 -04:00
Added exhaustive search backend 2014-09-11 16:13:46 -04:00			`from genetic import GeneticOperators`

Now ATIDLAS is standalone. Everything dynamic.... 2015-01-12 13:20:53 -05:00			`def genetic(symbolic, Template, compute_perf, perf_metric, out):`
			`GA = GeneticOperators(symbolic, Template, out)`
Squashed feature branch: * Added CUDA support * Performance improvements * API improvements * Added "depth" parameter to GEMM * Android cross-compilation 2015-04-29 15:50:57 -04:00			`return GA.optimize(maxtime='5m0s', maxgen=10000, compute_perf=compute_perf, perf_metric=perf_metric)`