from msprof_analyze.prof_common.logger import get_logger
from typing import List, Dict, Any
from msprof_analyze.advisor.analyzer.base_analyzer import BaseAnalyzer
from msprof_analyze.advisor.result.result import OptimizeResult
from msprof_analyze.advisor.analyzer.dataloader.dataloader_checker import DataloaderChecker
from msprof_analyze.advisor.display.html.priority_background_color import PriorityBackgroundColor
from msprof_analyze.advisor.display.html.render import HTMLRender
from msprof_analyze.advisor.dataset.timeline_event_dataset import ScheduleAnalysisDataset
logger = get_logger()
class DataloaderAnalyzer(BaseAnalyzer):
dataset_cls_list = [ScheduleAnalysisDataset]
def __init__(self, collection_path, n_processes: int = 1, **kwargs) -> None:
super().__init__(collection_path, n_processes, **kwargs)
key = ScheduleAnalysisDataset.get_key()
self.dataset = self.get_first_data_by_key(self.dataset_list, key)
self.result = OptimizeResult()
self.html_render = HTMLRender()
@BaseAnalyzer.check_data((ScheduleAnalysisDataset.get_key(),))
def optimize(self, **kwargs):
dataloader_checker = DataloaderChecker()
dataloader_checker.check_slow_dataloader(self.dataset)
dataloader_checker.make_record(self.result)
dataloader_checker.make_render(self.html_render, priority=self.get_priority(), rank=kwargs.get("rank"))
return self.result
def get_priority(self, max_mem_op_dur=None):
return PriorityBackgroundColor.high