summaryrefslogtreecommitdiff
path: root/src/mem/ruby/network/simple/Throttle.cc
blob: ceba4741128890a7ad7cdc5d6c343e4763f3a4d5 (plain)
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
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249

/*
 * Copyright (c) 1999-2008 Mark D. Hill and David A. Wood
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are
 * met: redistributions of source code must retain the above copyright
 * notice, this list of conditions and the following disclaimer;
 * redistributions in binary form must reproduce the above copyright
 * notice, this list of conditions and the following disclaimer in the
 * documentation and/or other materials provided with the distribution;
 * neither the name of the copyright holders nor the names of its
 * contributors may be used to endorse or promote products derived from
 * this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

/*
 * $Id$
 *
 * Description: see Throttle.hh
 *
 */

#include "mem/ruby/network/simple/Throttle.hh"
#include "mem/ruby/buffers/MessageBuffer.hh"
#include "mem/ruby/network/Network.hh"
#include "mem/ruby/system/System.hh"
#include "mem/ruby/slicc_interface/NetworkMessage.hh"
#include "mem/protocol/Protocol.hh"

const int HIGH_RANGE = 256;
const int ADJUST_INTERVAL = 50000;
const int MESSAGE_SIZE_MULTIPLIER = 1000;
//const int BROADCAST_SCALING = 4; // Have a 16p system act like a 64p systems
const int BROADCAST_SCALING = 1;
const int PRIORITY_SWITCH_LIMIT = 128;

static int network_message_to_size(NetworkMessage* net_msg_ptr);

extern std::ostream * debug_cout_ptr;

Throttle::Throttle(int sID, NodeID node, int link_latency, int link_bandwidth_multiplier)
{
  init(node, link_latency, link_bandwidth_multiplier);
  m_sID = sID;
}

Throttle::Throttle(NodeID node, int link_latency, int link_bandwidth_multiplier)
{
  init(node, link_latency, link_bandwidth_multiplier);
  m_sID = 0;
}

void Throttle::init(NodeID node, int link_latency, int link_bandwidth_multiplier)
{
  m_node = node;
  m_vnets = 0;

  ASSERT(link_bandwidth_multiplier > 0);
  m_link_bandwidth_multiplier = link_bandwidth_multiplier;
  m_link_latency = link_latency;

  m_wakeups_wo_switch = 0;
  clearStats();
}

void Throttle::clear()
{
  for (int counter = 0; counter < m_vnets; counter++) {
    m_in[counter]->clear();
    m_out[counter]->clear();
  }
}

void Throttle::addLinks(const Vector<MessageBuffer*>& in_vec, const Vector<MessageBuffer*>& out_vec)
{
  assert(in_vec.size() == out_vec.size());
  for (int i=0; i<in_vec.size(); i++) {
    addVirtualNetwork(in_vec[i], out_vec[i]);
  }

  m_message_counters.setSize(MessageSizeType_NUM);
  for (int i=0; i<MessageSizeType_NUM; i++) {
    m_message_counters[i].setSize(in_vec.size());
    for (int j=0; j<m_message_counters[i].size(); j++) {
      m_message_counters[i][j] = 0;
    }
  }
}

void Throttle::addVirtualNetwork(MessageBuffer* in_ptr, MessageBuffer* out_ptr)
{
  m_units_remaining.insertAtBottom(0);
  m_in.insertAtBottom(in_ptr);
  m_out.insertAtBottom(out_ptr);

  // Set consumer and description
  m_in[m_vnets]->setConsumer(this);
  string desc = "[Queue to Throttle " + NodeIDToString(m_sID) + " " + NodeIDToString(m_node) + "]";
  m_in[m_vnets]->setDescription(desc);
  m_vnets++;
}

void Throttle::wakeup()
{
  // Limits the number of message sent to a limited number of bytes/cycle.
  assert(getLinkBandwidth() > 0);
  int bw_remaining = getLinkBandwidth();

  // Give the highest numbered link priority most of the time
  m_wakeups_wo_switch++;
  int highest_prio_vnet = m_vnets-1;
  int lowest_prio_vnet = 0;
  int counter = 1;
  bool schedule_wakeup = false;

  // invert priorities to avoid starvation seen in the component network
  if (m_wakeups_wo_switch > PRIORITY_SWITCH_LIMIT) {
    m_wakeups_wo_switch = 0;
    highest_prio_vnet = 0;
    lowest_prio_vnet = m_vnets-1;
    counter = -1;
  }

  for (int vnet = highest_prio_vnet; (vnet*counter) >= (counter*lowest_prio_vnet); vnet -= counter) {

    assert(m_out[vnet] != NULL);
    assert(m_in[vnet] != NULL);
    assert(m_units_remaining[vnet] >= 0);

    while ((bw_remaining > 0) && ((m_in[vnet]->isReady()) || (m_units_remaining[vnet] > 0)) && m_out[vnet]->areNSlotsAvailable(1)) {

      // See if we are done transferring the previous message on this virtual network
      if (m_units_remaining[vnet] == 0 && m_in[vnet]->isReady()) {

        // Find the size of the message we are moving
        MsgPtr msg_ptr = m_in[vnet]->peekMsgPtr();
        NetworkMessage* net_msg_ptr = dynamic_cast<NetworkMessage*>(msg_ptr.ref());
        m_units_remaining[vnet] += network_message_to_size(net_msg_ptr);

        DEBUG_NEWLINE(NETWORK_COMP,HighPrio);
        DEBUG_MSG(NETWORK_COMP,HighPrio,"throttle: " + int_to_string(m_node)
                  + " my bw " + int_to_string(getLinkBandwidth())
                  + " bw spent enqueueing net msg " + int_to_string(m_units_remaining[vnet])
                  + " time: " + int_to_string(g_eventQueue_ptr->getTime()) + ".");

        // Move the message
        m_out[vnet]->enqueue(m_in[vnet]->peekMsgPtr(), m_link_latency);
        m_in[vnet]->pop();

        // Count the message
        m_message_counters[net_msg_ptr->getMessageSize()][vnet]++;

        DEBUG_MSG(NETWORK_COMP,LowPrio,*m_out[vnet]);
        DEBUG_NEWLINE(NETWORK_COMP,HighPrio);
      }

      // Calculate the amount of bandwidth we spent on this message
      int diff = m_units_remaining[vnet] - bw_remaining;
      m_units_remaining[vnet] = max(0, diff);
      bw_remaining = max(0, -diff);
    }

    if ((bw_remaining > 0) && ((m_in[vnet]->isReady()) || (m_units_remaining[vnet] > 0)) && !m_out[vnet]->areNSlotsAvailable(1)) {
      DEBUG_MSG(NETWORK_COMP,LowPrio,vnet);
      schedule_wakeup = true; // schedule me to wakeup again because I'm waiting for my output queue to become available
    }
  }

  // We should only wake up when we use the bandwidth
  //  assert(bw_remaining != getLinkBandwidth());  // This is only mostly true

  // Record that we used some or all of the link bandwidth this cycle
  double ratio = 1.0-(double(bw_remaining)/double(getLinkBandwidth()));
  // If ratio = 0, we used no bandwidth, if ratio = 1, we used all
  linkUtilized(ratio);

  if ((bw_remaining > 0) && !schedule_wakeup) {
    // We have extra bandwidth and our output buffer was available, so we must not have anything else to do until another message arrives.
    DEBUG_MSG(NETWORK_COMP,LowPrio,*this);
    DEBUG_MSG(NETWORK_COMP,LowPrio,"not scheduled again");
  } else {
    DEBUG_MSG(NETWORK_COMP,LowPrio,*this);
    DEBUG_MSG(NETWORK_COMP,LowPrio,"scheduled again");
    // We are out of bandwidth for this cycle, so wakeup next cycle and continue
    g_eventQueue_ptr->scheduleEvent(this, 1);
  }
}

void Throttle::printStats(ostream& out) const
{
  out << "utilized_percent: " << getUtilization() << endl;
}

void Throttle::clearStats()
{
  m_ruby_start = g_eventQueue_ptr->getTime();
  m_links_utilized = 0.0;

  for (int i=0; i<m_message_counters.size(); i++) {
    for (int j=0; j<m_message_counters[i].size(); j++) {
      m_message_counters[i][j] = 0;
    }
  }
}

void Throttle::printConfig(ostream& out) const
{

}

double Throttle::getUtilization() const
{
  return (100.0 * double(m_links_utilized)) / (double(g_eventQueue_ptr->getTime()-m_ruby_start));
}

void Throttle::print(ostream& out) const
{
  out << "[Throttle: " << m_sID << " " << m_node << " bw: " << getLinkBandwidth() << "]";
}

// Helper function

static
int network_message_to_size(NetworkMessage* net_msg_ptr)
{
  assert(net_msg_ptr != NULL);

  // Artificially increase the size of broadcast messages
  if (BROADCAST_SCALING > 1) {
    if (net_msg_ptr->getDestination().isBroadcast()) {
      return (RubySystem::getNetwork()->MessageSizeType_to_int(net_msg_ptr->getMessageSize()) * MESSAGE_SIZE_MULTIPLIER * BROADCAST_SCALING);
    }
  }
  return (RubySystem::getNetwork()->MessageSizeType_to_int(net_msg_ptr->getMessageSize()) * MESSAGE_SIZE_MULTIPLIER);
}