def wrap(x):
_sample, num_proc, use_deadline = x
return heu.OPA(_sample, num_proc, None, use_deadline)
res_map = defaultdict(lambda: defaultdict(lambda: 0))
for num_tasks in [args.num_task]:
if num_proc >= num_tasks:
continue
util_range = get_util_range(num_proc)
for util in util_range:
print(num_tasks, util)
res = res_map[(num_proc, num_tasks, util)]
with open("eval/%d-%d/%s" % (num_proc, num_tasks, util), 'rb') as f:
train_dataset = pickle.load(f)
i = 0
for x, y in train_dataset:
if i == 10000:
break
opa_da_lc_res_ = cy_heu.OPA(y, num_proc, heu.test_DA_LC,
use_deadline)
#res['OPA_DA'] += opa_da_res
res['OPA_DA_LC'] += opa_da_lc_res_
p = scores_to_priority(
heu.get_DkC_scores(y, num_proc, use_deadline=use_deadline))
res['DkC_RTA_LC'] += (heu.test_RTA_LC(y, num_proc, p,
use_deadline))
res['DkC_DA_LC'] += (heu.test_DA_LC(y, num_proc, p, use_deadline))
p = scores_to_priority(
heu.get_DM_scores(y, num_proc, use_deadline=use_deadline))
res['DM_RTA_LC'] += (heu.test_RTA_LC(y, num_proc, p, use_deadline))
res['DM_DA_LC'] += (heu.test_DA_LC(y, num_proc, p, use_deadline))
p = scores_to_priority(
def wrap(x):
_sample, num_proc, use_deadline = x
return heu.OPA(
_sample, num_proc, None,
use_deadline) # 여기서 OPA에 테스트 안넘겼는데 까고 들어가보면 DA어쩌구 테스트를 사용함.
# Calculating heuristics
model = Solver(args.num_procs,
args.embedding_size,
args.hidden_size,
args.num_tasks,
use_deadline=use_deadline
nimp=bool(args.model))
if args.use_cuda:
model = model.cuda()
res_opa = []
for i, sample in tqdm(test_dataset):
ret = heu.OPA(sample, args.num_procs, heu.test_DA_LC, use_deadline)
res_opa.append(ret)
opares = np.sum(res_opa)
print("[before training][OPA generates %d]" % opares)
model.train()
optimizer = optim.Adam(model.parameters(), lr=args.lr)
scheduler = torch.optim.lr_scheduler.StepLR(optimizer, args.lr_decay_step, gamma=0.9, last_epoch=-1)
last_rl_model_sum = -1
updates = 0
noupdateinarow = 0
_max = -1
for epoch in range(args.num_epochs):
loss_ = 0
avg_hit = []
for batch_idx, (_, sample_batch) in enumerate(train_data_loader):
if args.use_cuda: