Profiler.hh revision 8164
1/* 2 * Copyright (c) 1999-2008 Mark D. Hill and David A. Wood 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions are 7 * met: redistributions of source code must retain the above copyright 8 * notice, this list of conditions and the following disclaimer; 9 * redistributions in binary form must reproduce the above copyright 10 * notice, this list of conditions and the following disclaimer in the 11 * documentation and/or other materials provided with the distribution; 12 * neither the name of the copyright holders nor the names of its 13 * contributors may be used to endorse or promote products derived from 14 * this software without specific prior written permission. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 17 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 18 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 19 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 20 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 21 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 22 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 26 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 27 */ 28 29/* 30 This file has been modified by Kevin Moore and Dan Nussbaum of the 31 Scalable Systems Research Group at Sun Microsystems Laboratories 32 (http://research.sun.com/scalable/) to support the Adaptive 33 Transactional Memory Test Platform (ATMTP). 34 35 Please send email to atmtp-interest@sun.com with feedback, questions, or 36 to request future announcements about ATMTP. 37 38 ---------------------------------------------------------------------- 39 40 File modification date: 2008-02-23 41 42 ---------------------------------------------------------------------- 43*/ 44 45#ifndef __MEM_RUBY_PROFILER_PROFILER_HH__ 46#define __MEM_RUBY_PROFILER_PROFILER_HH__ 47 48#include <iostream> 49#include <map> 50#include <string> 51#include <vector> 52 53#include "base/hashmap.hh" 54#include "mem/protocol/RubyAccessMode.hh" 55#include "mem/protocol/AccessType.hh" 56#include "mem/protocol/CacheRequestType.hh" 57#include "mem/protocol/GenericMachineType.hh" 58#include "mem/protocol/GenericRequestType.hh" 59#include "mem/protocol/PrefetchBit.hh" 60#include "mem/ruby/common/Address.hh" 61#include "mem/ruby/common/Consumer.hh" 62#include "mem/ruby/common/Global.hh" 63#include "mem/ruby/common/Histogram.hh" 64#include "mem/ruby/common/Set.hh" 65#include "mem/ruby/system/MachineID.hh" 66#include "mem/ruby/system/MemoryControl.hh" 67#include "mem/ruby/system/NodeID.hh" 68#include "params/RubyProfiler.hh" 69#include "sim/sim_object.hh" 70 71class CacheMsg; 72class AddressProfiler; 73 74class Profiler : public SimObject, public Consumer 75{ 76 public: 77 typedef RubyProfilerParams Params; 78 Profiler(const Params *); 79 ~Profiler(); 80 81 void wakeup(); 82 83 void setPeriodicStatsFile(const std::string& filename); 84 void setPeriodicStatsInterval(integer_t period); 85 86 void printStats(std::ostream& out, bool short_stats=false); 87 void printShortStats(std::ostream& out) { printStats(out, true); } 88 void printTraceStats(std::ostream& out) const; 89 void clearStats(); 90 void printConfig(std::ostream& out) const; 91 void printResourceUsage(std::ostream& out) const; 92 93 AddressProfiler* getAddressProfiler() { return m_address_profiler_ptr; } 94 AddressProfiler* getInstructionProfiler() { return m_inst_profiler_ptr; } 95 96 void addAddressTraceSample(const CacheMsg& msg, NodeID id); 97 98 void profileRequest(const std::string& requestStr); 99 void profileSharing(const Address& addr, AccessType type, 100 NodeID requestor, const Set& sharers, 101 const Set& owner); 102 103 void profileMulticastRetry(const Address& addr, int count); 104 105 void profileFilterAction(int action); 106 107 void profileConflictingRequests(const Address& addr); 108 109 void 110 profileOutstandingRequest(int outstanding) 111 { 112 m_outstanding_requests.add(outstanding); 113 } 114 115 void 116 profileOutstandingPersistentRequest(int outstanding) 117 { 118 m_outstanding_persistent_requests.add(outstanding); 119 } 120 121 void 122 profileAverageLatencyEstimate(int latency) 123 { 124 m_average_latency_estimate.add(latency); 125 } 126 127 void recordPrediction(bool wasGood, bool wasPredicted); 128 129 void startTransaction(int cpu); 130 void endTransaction(int cpu); 131 void profilePFWait(Time waitTime); 132 133 void controllerBusy(MachineID machID); 134 void bankBusy(); 135 136 void missLatency(Time t, 137 RubyRequestType type, 138 const GenericMachineType respondingMach); 139 140 void missLatencyWcc(Time issuedTime, 141 Time initialRequestTime, 142 Time forwardRequestTime, 143 Time firstResponseTime, 144 Time completionTime); 145 146 void missLatencyDir(Time issuedTime, 147 Time initialRequestTime, 148 Time forwardRequestTime, 149 Time firstResponseTime, 150 Time completionTime); 151 152 void swPrefetchLatency(Time t, 153 CacheRequestType type, 154 const GenericMachineType respondingMach); 155 156 void sequencerRequests(int num) { m_sequencer_requests.add(num); } 157 158 void profileMsgDelay(int virtualNetwork, int delayCycles); 159 160 void print(std::ostream& out) const; 161 162 void rubyWatch(int proc); 163 bool watchAddress(Address addr); 164 165 // return Ruby's start time 166 Time 167 getRubyStartTime() 168 { 169 return m_ruby_start; 170 } 171 172 // added by SS 173 bool getHotLines() { return m_hot_lines; } 174 bool getAllInstructions() { return m_all_instructions; } 175 176 private: 177 // Private copy constructor and assignment operator 178 Profiler(const Profiler& obj); 179 Profiler& operator=(const Profiler& obj); 180 181 AddressProfiler* m_address_profiler_ptr; 182 AddressProfiler* m_inst_profiler_ptr; 183 184 std::vector<int64> m_instructions_executed_at_start; 185 std::vector<int64> m_cycles_executed_at_start; 186 187 std::ostream* m_periodic_output_file_ptr; 188 integer_t m_stats_period; 189 190 Time m_ruby_start; 191 time_t m_real_time_start_time; 192 193 std::vector<std::vector<integer_t> > m_busyControllerCount; 194 integer_t m_busyBankCount; 195 Histogram m_multicast_retry_histogram; 196 197 Histogram m_filter_action_histogram; 198 Histogram m_tbeProfile; 199 200 Histogram m_sequencer_requests; 201 Histogram m_read_sharing_histogram; 202 Histogram m_write_sharing_histogram; 203 Histogram m_all_sharing_histogram; 204 int64 m_cache_to_cache; 205 int64 m_memory_to_cache; 206 207 Histogram m_prefetchWaitHistogram; 208 209 std::vector<Histogram> m_missLatencyHistograms; 210 std::vector<Histogram> m_machLatencyHistograms; 211 std::vector< std::vector<Histogram> > m_missMachLatencyHistograms; 212 Histogram m_wCCIssueToInitialRequestHistogram; 213 Histogram m_wCCInitialRequestToForwardRequestHistogram; 214 Histogram m_wCCForwardRequestToFirstResponseHistogram; 215 Histogram m_wCCFirstResponseToCompleteHistogram; 216 int64 m_wCCIncompleteTimes; 217 Histogram m_dirIssueToInitialRequestHistogram; 218 Histogram m_dirInitialRequestToForwardRequestHistogram; 219 Histogram m_dirForwardRequestToFirstResponseHistogram; 220 Histogram m_dirFirstResponseToCompleteHistogram; 221 int64 m_dirIncompleteTimes; 222 223 Histogram m_allMissLatencyHistogram; 224 225 Histogram m_allSWPrefetchLatencyHistogram; 226 Histogram m_SWPrefetchL2MissLatencyHistogram; 227 std::vector<Histogram> m_SWPrefetchLatencyHistograms; 228 std::vector<Histogram> m_SWPrefetchMachLatencyHistograms; 229 230 Histogram m_delayedCyclesHistogram; 231 Histogram m_delayedCyclesNonPFHistogram; 232 std::vector<Histogram> m_delayedCyclesVCHistograms; 233 234 Histogram m_outstanding_requests; 235 Histogram m_outstanding_persistent_requests; 236 237 Histogram m_average_latency_estimate; 238 239 m5::hash_set<Address> m_watch_address_set; 240 // counts all initiated cache request including PUTs 241 int m_requests; 242 std::map<std::string, int> m_requestProfileMap; 243 244 //added by SS 245 bool m_hot_lines; 246 bool m_all_instructions; 247 248 int m_num_of_sequencers; 249}; 250 251inline std::ostream& 252operator<<(std::ostream& out, const Profiler& obj) 253{ 254 obj.print(out); 255 out << std::flush; 256 return out; 257} 258 259#endif // __MEM_RUBY_PROFILER_PROFILER_HH__ 260 261 262