Bubble-Scheduling/processor_graph.py at master · Midren/Bubble-Scheduling · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
from collections import defaultdict
from graph import Node

from pprint import pprint


class ProcessorsGraph:

    def __init__(self, my_dict):
        self.inc_dct = my_dict

    def bfs(self, start):
        visited, queue = list(), [start]
        while queue:
            vertex = queue.pop(0)
            if vertex not in visited:
                visited.append(vertex)
                queue.extend([element for element in self.inc_dct[vertex] if
                              element not in visited])
        return visited

    def max_degree_node(self):
        node = None
        for i in self.inc_dct.keys():
            if node is None:
                node = i
            elif len(self.inc_dct[i]) > len(self.inc_dct[node]):
                node = i
        return i


class Tasks:
    def __init__(self, tasks, tasks_node, graph, pivot_pe):
        self.graph = graph
        self.comm_dict = dict()
        tasks[0].proc = pivot_pe
        for num, task in enumerate(tasks[1:]):
            task.proc = pivot_pe

            for t in graph.get_parents(tasks_node[num + 1]):
                for tsk in tasks:
                    if tsk.index == t.index:
                        task.parents.append(tsk)

            task.dat = tasks[1 + num - 1].fn
            task.st = tasks[1 + num - 1].fn
            task.fn = task.st + task.computation
        self.tasks = defaultdict(list)
        self.tasks[pivot_pe] = tasks
        self.links = defaultdict(list)
        for nd in graph.get_nodes():
            for node in graph.get_children_nodes(nd):
                self.comm_dict[(nd.index, node.index)] = graph.get_edge(nd,
                                                                        node)

    def st_if_migrate_comp(self, task, proc):
        if not self.tasks[proc]:
            return task.computation
        for num, t in enumerate(self.tasks[proc][1:]):
            new_st = max(self.tasks[proc][1 + num - 1].fn, task.dat)
            if (t.st - new_st >= task.computation):  # and new_st < task.st:
                return new_st
        return self.tasks[proc][len(self.tasks[proc]) - 1].fn

    def st_if_migrate_comm(self, message):
        if not self.links[message.parent.proc]:
            return self.comm_dict[(message.parent.index, message.child.index)]
        for num, m in enumerate(self.links[message.child.proc][1:]):
            new_st = max(self.links[message.child.proc][1 + num - 1].fn,
                         m.child.fn)
            if (m.st - new_st >= message.communication):
                return new_st
        return self.links[message.parent.proc][
            len(self.links[message.parent.proc]) - 1].fn

    def st_if_migrate(self, task, proc):
        new_st_2 = 0
        if task.parents:
            parent = self.vip(task)
            message = Message(parent, task,
                              self.comm_dict[(parent.index, task.index)])
            new_st_2 = self.st_if_migrate_comm(message)
        new_st_1 = self.st_if_migrate_comp(task, proc)
        return max(new_st_1, new_st_2)

    def migrate(self, task, proc):
        pivot_pe = task.proc
        if len(self.tasks[proc]) <= 1:
            self.tasks[proc].append(task)
            self.tasks[task.proc].remove(task)
            task.proc = proc
        else:
            for num, t in enumerate(self.tasks[proc][1:]):
                new_st = max(self.tasks[proc][num - 1].fn, task.dat)
                if (t.st - new_st >= task.computation):  # and new_st < task.st:
                    self.tasks[proc].insert(1 + num, task)
                    self.tasks[task.proc].remove(task)
                    task.proc = proc
                    break
            else:
                self.tasks[proc].append(task)
                self.tasks[task.proc].remove(task)
                task.proc = proc


        if task.parents:
            parent = self.vip(task)
            message = Message(parent, task,
                              self.comm_dict[(parent.index, task.index)])
            if len(self.links[message.parent.proc]) <= 1:
                self.links[message.parent.proc].append(message)
                task.communication += message.communication
            else:
                for num, m in enumerate(self.links[message.parent.proc][1:]):
                    new_st = max(
                        self.links[message.parent.proc][1 + num - 1].fn,
                        message.parent.fn)
                    if (m.st - new_st >= message.communication):
                        self.links[message.parent.proc].insert(num, message)
                        task.communication += message.communication
                        break
                else:
                    self.links[message.parent.proc].append(message)
                    task.communication += message.communication

        self.update(pivot_pe, proc)

    def update(self, proc, pivot_pe):
        for proc in [pivot_pe, proc]:
            self.tasks[proc][0].dat = self.tasks[proc][0].communication
            self.tasks[proc][0].st = max(self.tasks[proc][0].dat, 0)
            self.tasks[proc][0].fn = self.tasks[proc][0].st + self.tasks[proc][0].computation
            for num, task in enumerate(self.tasks[proc][1:]):
                task.dat = self.vip(task).fn + task.communication
                task.st = max(self.tasks[proc][1 + num - 1].fn, task.dat)
                task.fn = task.st + task.computation

    def vip(self, task):
        if task.parents:
            parent = max(
                [(self.comm_dict[(t.index, task.index)], t) for t in task.parents],
                key=lambda x: x[0])
        else:
            parent = (None, None)
        return parent[1]


class Message:
    def __init__(self, parent_node, node, communication):
        self.parent = parent_node
        self.child = node
        self.communication = communication
        self.st = 0
        self.fn = self.communication

    def __repr__(self):
        return self.__class__.__name__ + "( " + str(self.parent) + ", " + str(
            self.child) + " )"


class Task(Node):
    def __init__(self, node):
        super().__init__(node.index, node.computation, node.tranfering)
        self.st = 0
        self.fn = self.computation
        self.dat = 0
        self.proc = None
        self.parents = []
        self.communication = 0