gem5/src/cpu/testers/garnet_synthetic_traffic/GarnetSyntheticTraffic.cc

359 lines
11 KiB
C++
Raw Normal View History

/*
* Copyright (c) 2016 Georgia Institute of Technology
* 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.
*
* Authors: Tushar Krishna
*/
2011-04-15 19:44:06 +02:00
#include <cmath>
#include <iomanip>
#include <set>
#include <string>
#include <vector>
#include "base/misc.hh"
#include "base/random.hh"
#include "base/statistics.hh"
#include "cpu/testers/garnet_synthetic_traffic/GarnetSyntheticTraffic.hh"
#include "debug/GarnetSyntheticTraffic.hh"
#include "mem/mem_object.hh"
#include "mem/packet.hh"
2011-04-15 19:44:06 +02:00
#include "mem/port.hh"
#include "mem/request.hh"
#include "sim/sim_events.hh"
#include "sim/stats.hh"
#include "sim/system.hh"
using namespace std;
int TESTER_NETWORK=0;
bool
GarnetSyntheticTraffic::CpuPort::recvTimingResp(PacketPtr pkt)
{
tester->completeRequest(pkt);
return true;
}
void
GarnetSyntheticTraffic::CpuPort::recvReqRetry()
{
tester->doRetry();
}
void
GarnetSyntheticTraffic::sendPkt(PacketPtr pkt)
{
MEM: Separate requests and responses for timing accesses This patch moves send/recvTiming and send/recvTimingSnoop from the Port base class to the MasterPort and SlavePort, and also splits them into separate member functions for requests and responses: send/recvTimingReq, send/recvTimingResp, and send/recvTimingSnoopReq, send/recvTimingSnoopResp. A master port sends requests and receives responses, and also receives snoop requests and sends snoop responses. A slave port has the reciprocal behaviour as it receives requests and sends responses, and sends snoop requests and receives snoop responses. For all MemObjects that have only master ports or slave ports (but not both), e.g. a CPU, or a PIO device, this patch merely adds more clarity to what kind of access is taking place. For example, a CPU port used to call sendTiming, and will now call sendTimingReq. Similarly, a response previously came back through recvTiming, which is now recvTimingResp. For the modules that have both master and slave ports, e.g. the bus, the behaviour was previously relying on branches based on pkt->isRequest(), and this is now replaced with a direct call to the apprioriate member function depending on the type of access. Please note that send/recvRetry is still shared by all the timing accessors and remains in the Port base class for now (to maintain the current bus functionality and avoid changing the statistics of all regressions). The packet queue is split into a MasterPort and SlavePort version to facilitate the use of the new timing accessors. All uses of the PacketQueue are updated accordingly. With this patch, the type of packet (request or response) is now well defined for each type of access, and asserts on pkt->isRequest() and pkt->isResponse() are now moved to the appropriate send member functions. It is also worth noting that sendTimingSnoopReq no longer returns a boolean, as the semantics do not alow snoop requests to be rejected or stalled. All these assumptions are now excplicitly part of the port interface itself.
2012-05-01 19:40:42 +02:00
if (!cachePort.sendTimingReq(pkt)) {
retryPkt = pkt; // RubyPort will retry sending
}
numPacketsSent++;
}
GarnetSyntheticTraffic::GarnetSyntheticTraffic(const Params *p)
: MemObject(p),
tickEvent(this),
cachePort("GarnetSyntheticTraffic", this),
retryPkt(NULL),
size(p->memory_size),
blockSizeBits(p->block_offset),
numDestinations(p->num_dest),
simCycles(p->sim_cycles),
numPacketsMax(p->num_packets_max),
numPacketsSent(0),
singleSender(p->single_sender),
singleDest(p->single_dest),
trafficType(p->traffic_type),
injRate(p->inj_rate),
injVnet(p->inj_vnet),
precision(p->precision),
responseLimit(p->response_limit),
masterId(p->system->getMasterId(name()))
{
// set up counters
noResponseCycles = 0;
schedule(tickEvent, 0);
initTrafficType();
if (trafficStringToEnum.count(trafficType) == 0) {
fatal("Unknown Traffic Type: %s!\n", traffic);
}
traffic = trafficStringToEnum[trafficType];
id = TESTER_NETWORK++;
DPRINTF(GarnetSyntheticTraffic,"Config Created: Name = %s , and id = %d\n",
name(), id);
}
BaseMasterPort &
GarnetSyntheticTraffic::getMasterPort(const std::string &if_name, PortID idx)
{
if (if_name == "test")
return cachePort;
else
return MemObject::getMasterPort(if_name, idx);
}
void
GarnetSyntheticTraffic::init()
{
numPacketsSent = 0;
}
void
GarnetSyntheticTraffic::completeRequest(PacketPtr pkt)
{
Request *req = pkt->req;
DPRINTF(GarnetSyntheticTraffic,
"Completed injection of %s packet for address %x\n",
pkt->isWrite() ? "write" : "read\n",
req->getPaddr());
assert(pkt->isResponse());
noResponseCycles = 0;
delete req;
delete pkt;
}
void
GarnetSyntheticTraffic::tick()
{
if (++noResponseCycles >= responseLimit) {
fatal("%s deadlocked at cycle %d\n", name(), curTick());
}
// make new request based on injection rate
// (injection rate's range depends on precision)
// - generate a random number between 0 and 10^precision
// - send pkt if this number is < injRate*(10^precision)
bool sendAllowedThisCycle;
double injRange = pow((double) 10, (double) precision);
unsigned trySending = random_mt.random<unsigned>(0, (int) injRange);
if (trySending < injRate*injRange)
sendAllowedThisCycle = true;
else
sendAllowedThisCycle = false;
// always generatePkt unless fixedPkts or singleSender is enabled
if (sendAllowedThisCycle) {
bool senderEnable = true;
if (numPacketsMax >= 0 && numPacketsSent >= numPacketsMax)
senderEnable = false;
if (singleSender >= 0 && id != singleSender)
senderEnable = false;
if (senderEnable)
generatePkt();
}
// Schedule wakeup
if (curTick() >= simCycles)
exitSimLoop("Network Tester completed simCycles");
else {
if (!tickEvent.scheduled())
schedule(tickEvent, clockEdge(Cycles(1)));
}
}
void
GarnetSyntheticTraffic::generatePkt()
{
int num_destinations = numDestinations;
int radix = (int) sqrt(num_destinations);
unsigned destination = id;
int dest_x = -1;
int dest_y = -1;
int source = id;
int src_x = id%radix;
int src_y = id/radix;
if (singleDest >= 0)
{
destination = singleDest;
} else if (traffic == UNIFORM_RANDOM_) {
destination = random_mt.random<unsigned>(0, num_destinations - 1);
} else if (traffic == BIT_COMPLEMENT_) {
dest_x = radix - src_x - 1;
dest_y = radix - src_y - 1;
destination = dest_y*radix + dest_x;
} else if (traffic == BIT_REVERSE_) {
unsigned int straight = source;
unsigned int reverse = source & 1; // LSB
int num_bits = (int) log2(num_destinations);
for (int i = 1; i < num_bits; i++)
{
reverse <<= 1;
straight >>= 1;
reverse |= (straight & 1); // LSB
}
destination = reverse;
} else if (traffic == BIT_ROTATION_) {
if (source%2 == 0)
destination = source/2;
else // (source%2 == 1)
destination = ((source/2) + (num_destinations/2));
} else if (traffic == NEIGHBOR_) {
dest_x = (src_x + 1) % radix;
dest_y = src_y;
destination = dest_y*radix + dest_x;
} else if (traffic == SHUFFLE_) {
if (source < num_destinations/2)
destination = source*2;
else
destination = (source*2 - num_destinations + 1);
} else if (traffic == TRANSPOSE_) {
dest_x = src_y;
dest_y = src_x;
destination = dest_y*radix + dest_x;
} else if (traffic == TORNADO_) {
dest_x = (src_x + (int) ceil(radix/2) - 1) % radix;
dest_y = src_y;
destination = dest_y*radix + dest_x;
}
else {
fatal("Unknown Traffic Type: %s!\n", traffic);
}
// The source of the packets is a cache.
// The destination of the packets is a directory.
// The destination bits are embedded in the address after byte-offset.
Addr paddr = destination;
paddr <<= blockSizeBits;
unsigned access_size = 1; // Does not affect Ruby simulation
// Modeling different coherence msg types over different msg classes.
//
// GarnetSyntheticTraffic assumes the Garnet_standalone coherence protocol
// which models three message classes/virtual networks.
// These are: request, forward, response.
// requests and forwards are "control" packets (typically 8 bytes),
// while responses are "data" packets (typically 72 bytes).
//
// Life of a packet from the tester into the network:
// (1) This function generatePkt() generates packets of one of the
// following 3 types (randomly) : ReadReq, INST_FETCH, WriteReq
// (2) mem/ruby/system/RubyPort.cc converts these to RubyRequestType_LD,
// RubyRequestType_IFETCH, RubyRequestType_ST respectively
// (3) mem/ruby/system/Sequencer.cc sends these to the cache controllers
// in the coherence protocol.
// (4) Network_test-cache.sm tags RubyRequestType:LD,
// RubyRequestType:IFETCH and RubyRequestType:ST as
// Request, Forward, and Response events respectively;
// and injects them into virtual networks 0, 1 and 2 respectively.
// It immediately calls back the sequencer.
// (5) The packet traverses the network (simple/garnet) and reaches its
// destination (Directory), and network stats are updated.
// (6) Network_test-dir.sm simply drops the packet.
//
MemCmd::Command requestType;
Request *req = nullptr;
Request::Flags flags;
// Inject in specific Vnet
// Vnet 0 and 1 are for control packets (1-flit)
// Vnet 2 is for data packets (5-flit)
int injReqType = injVnet;
if (injReqType < 0 || injReqType > 2)
{
// randomly inject in any vnet
injReqType = random_mt.random(0, 2);
}
if (injReqType == 0) {
// generate packet for virtual network 0
requestType = MemCmd::ReadReq;
req = new Request(paddr, access_size, flags, masterId);
} else if (injReqType == 1) {
// generate packet for virtual network 1
requestType = MemCmd::ReadReq;
flags.set(Request::INST_FETCH);
req = new Request(0, 0x0, access_size, flags, masterId, 0x0, 0);
req->setPaddr(paddr);
} else { // if (injReqType == 2)
// generate packet for virtual network 2
requestType = MemCmd::WriteReq;
req = new Request(paddr, access_size, flags, masterId);
}
req->setContext(id);
//No need to do functional simulation
//We just do timing simulation of the network
DPRINTF(GarnetSyntheticTraffic,
"Generated packet with destination %d, embedded in address %x\n",
destination, req->getPaddr());
MEM: Remove the Broadcast destination from the packet This patch simplifies the packet by removing the broadcast flag and instead more firmly relying on (and enforcing) the semantics of transactions in the classic memory system, i.e. request packets are routed from a master to a slave based on the address, and when they are created they have neither a valid source, nor destination. On their way to the slave, the request packet is updated with a source field for all modules that multiplex packets from multiple master (e.g. a bus). When a request packet is turned into a response packet (at the final slave), it moves the potentially populated source field to the destination field, and the response packet is routed through any multiplexing components back to the master based on the destination field. Modules that connect multiplexing components, such as caches and bridges store any existing source and destination field in the sender state as a stack (just as before). The packet constructor is simplified in that there is no longer a need to pass the Packet::Broadcast as the destination (this was always the case for the classic memory system). In the case of Ruby, rather than using the parameter to the constructor we now rely on setDest, as there is already another three-argument constructor in the packet class. In many places where the packet information was printed as part of DPRINTFs, request packets would be printed with a numeric "dest" that would always be -1 (Broadcast) and that field is now removed from the printing.
2012-04-14 11:45:55 +02:00
PacketPtr pkt = new Packet(req, requestType);
pkt->dataDynamic(new uint8_t[req->getSize()]);
pkt->senderState = NULL;
sendPkt(pkt);
}
void
GarnetSyntheticTraffic::initTrafficType()
{
trafficStringToEnum["bit_complement"] = BIT_COMPLEMENT_;
trafficStringToEnum["bit_reverse"] = BIT_REVERSE_;
trafficStringToEnum["bit_rotation"] = BIT_ROTATION_;
trafficStringToEnum["neighbor"] = NEIGHBOR_;
trafficStringToEnum["shuffle"] = SHUFFLE_;
trafficStringToEnum["tornado"] = TORNADO_;
trafficStringToEnum["transpose"] = TRANSPOSE_;
trafficStringToEnum["uniform_random"] = UNIFORM_RANDOM_;
}
void
GarnetSyntheticTraffic::doRetry()
{
MEM: Separate requests and responses for timing accesses This patch moves send/recvTiming and send/recvTimingSnoop from the Port base class to the MasterPort and SlavePort, and also splits them into separate member functions for requests and responses: send/recvTimingReq, send/recvTimingResp, and send/recvTimingSnoopReq, send/recvTimingSnoopResp. A master port sends requests and receives responses, and also receives snoop requests and sends snoop responses. A slave port has the reciprocal behaviour as it receives requests and sends responses, and sends snoop requests and receives snoop responses. For all MemObjects that have only master ports or slave ports (but not both), e.g. a CPU, or a PIO device, this patch merely adds more clarity to what kind of access is taking place. For example, a CPU port used to call sendTiming, and will now call sendTimingReq. Similarly, a response previously came back through recvTiming, which is now recvTimingResp. For the modules that have both master and slave ports, e.g. the bus, the behaviour was previously relying on branches based on pkt->isRequest(), and this is now replaced with a direct call to the apprioriate member function depending on the type of access. Please note that send/recvRetry is still shared by all the timing accessors and remains in the Port base class for now (to maintain the current bus functionality and avoid changing the statistics of all regressions). The packet queue is split into a MasterPort and SlavePort version to facilitate the use of the new timing accessors. All uses of the PacketQueue are updated accordingly. With this patch, the type of packet (request or response) is now well defined for each type of access, and asserts on pkt->isRequest() and pkt->isResponse() are now moved to the appropriate send member functions. It is also worth noting that sendTimingSnoopReq no longer returns a boolean, as the semantics do not alow snoop requests to be rejected or stalled. All these assumptions are now excplicitly part of the port interface itself.
2012-05-01 19:40:42 +02:00
if (cachePort.sendTimingReq(retryPkt)) {
retryPkt = NULL;
}
}
void
GarnetSyntheticTraffic::printAddr(Addr a)
{
cachePort.printAddr(a);
}
GarnetSyntheticTraffic *
GarnetSyntheticTrafficParams::create()
{
return new GarnetSyntheticTraffic(this);
}