fluid_benchmark_log_to_vdl_data.py

from __future__ import print_function
import random
from visualdl import LogWriter

logdir = "resnet50_v100"



def parse_logdata(logfile):
    loss_ret = {}
    acc1_ret = {}
    acc5_ret = {}

    with open(logfile, "r") as fn:
        for line in fn.readlines():
            if not (line.startswith("Pass") or line.startswith("Test")):
                continue
            sections = [s.strip() for s in line.strip().split(",")]
            if sections[0].startswith("Total") or sections[0].startswith("read images"):
                continue
            if sections[1].startswith("Test"):
                passid = int(sections[0].replace("Pass: ", ""))
                continue
                # acc1 = sections[1].replace("Test Accuracy: [", "")
                # acc5 = sections[2].replace("]", "")
                # dist_test_acc1[passid] = float(acc1)
                # dist_test_acc5[passid] = float(acc5)
            else:
                passid = int(sections[0].replace("Pass ", ""))
                loss = float(sections[2].replace("loss ", ""))
                acc1 = float(sections[3].replace("accucacys: [", ""))
                acc5 = float(sections[4].replace("]", ""))
                if not passid in loss_ret:
                    loss_ret[passid] = []
                if not passid in acc1_ret:
                    acc1_ret[passid] = []
                if not passid in acc5_ret:
                    acc5_ret[passid] = []
                loss_ret[passid].append(loss)
                acc1_ret[passid].append(acc1)
                acc5_ret[passid].append(acc5)
    # cal avg
    ret = [{}, {}, {}]
    for idx, data in enumerate([loss_ret, acc1_ret, acc5_ret]):
        for passid in data:
            ret[idx][passid] = sum(data[passid]) / len(data[passid])

    return ret


def write_vdldata(logfile, logger, name_prefix):
    data = parse_logdata(logfile)
    with logger.mode("%s-loss" % name_prefix):
        s0 = logger.scalar("%s/loss" % logdir)
    with logger.mode("%s-acc1" % name_prefix):
        s1 = logger.scalar("%s/acc1" % logdir)
    with logger.mode("%s-acc5" % name_prefix):
        s2 = logger.scalar("%s/acc1" % logdir)

    for k, v in data[0].iteritems():
            s0.add_record(k, v)
    for k, v in data[1].iteritems():
            s1.add_record(k, v)
    for k, v in data[2].iteritems():
            s2.add_record(k, v)

def main():
    logger = LogWriter(logdir, sync_cycle=10000)
    write_vdldata("local_8GPU_0.1_48_piecewise.log", logger, "local")
    write_vdldata("nccl2_32GPU_0.1_32_piecewise.log", logger, "nccl2")
    write_vdldata("ps_32GPU_0.1_48_piecewise.log", logger, "ps1")
    write_vdldata("ps_32GPU_0.4_48_piecewise.log", logger, "ps4")


if __name__ == "__main__":
    main()