blob: b78796f8453ab1aa775c6ab7df5e2fd7bf1ad0e7 [file] [log] [blame]
/*
* Copyright (c) 2012-2020 ARM Limited
* All rights reserved
*
* The license below extends only to copyright in the software and shall
* not be construed as granting a license to any other intellectual
* property including but not limited to intellectual property relating
* to a hardware implementation of the functionality of the software
* licensed hereunder. You may use the software subject to the license
* terms below provided that you ensure that this notice is replicated
* unmodified and in its entirety in all distributions of the software,
* modified or unmodified, in source code or in binary form.
*
* Copyright (c) 2013 Amin Farmahini-Farahani
* 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.
*/
/**
* @file
* MemCtrl declaration
*/
#ifndef __MEM_CTRL_HH__
#define __MEM_CTRL_HH__
#include <deque>
#include <string>
#include <unordered_set>
#include <utility>
#include <vector>
#include "base/callback.hh"
#include "base/statistics.hh"
#include "enums/MemSched.hh"
#include "mem/qos/mem_ctrl.hh"
#include "mem/qport.hh"
#include "params/MemCtrl.hh"
#include "sim/eventq.hh"
namespace gem5
{
class DRAMInterface;
class NVMInterface;
/**
* A burst helper helps organize and manage a packet that is larger than
* the memory burst size. A system packet that is larger than the burst size
* is split into multiple packets and all those packets point to
* a single burst helper such that we know when the whole packet is served.
*/
class BurstHelper
{
public:
/** Number of bursts requred for a system packet **/
const unsigned int burstCount;
/** Number of bursts serviced so far for a system packet **/
unsigned int burstsServiced;
BurstHelper(unsigned int _burstCount)
: burstCount(_burstCount), burstsServiced(0)
{ }
};
/**
* A memory packet stores packets along with the timestamp of when
* the packet entered the queue, and also the decoded address.
*/
class MemPacket
{
public:
/** When did request enter the controller */
const Tick entryTime;
/** When will request leave the controller */
Tick readyTime;
/** This comes from the outside world */
const PacketPtr pkt;
/** RequestorID associated with the packet */
const RequestorID _requestorId;
const bool read;
/** Does this packet access DRAM?*/
const bool dram;
/** Will be populated by address decoder */
const uint8_t rank;
const uint8_t bank;
const uint32_t row;
/**
* Bank id is calculated considering banks in all the ranks
* eg: 2 ranks each with 8 banks, then bankId = 0 --> rank0, bank0 and
* bankId = 8 --> rank1, bank0
*/
const uint16_t bankId;
/**
* The starting address of the packet.
* This address could be unaligned to burst size boundaries. The
* reason is to keep the address offset so we can accurately check
* incoming read packets with packets in the write queue.
*/
Addr addr;
/**
* The size of this dram packet in bytes
* It is always equal or smaller than the burst size
*/
unsigned int size;
/**
* A pointer to the BurstHelper if this MemPacket is a split packet
* If not a split packet (common case), this is set to NULL
*/
BurstHelper* burstHelper;
/**
* QoS value of the encapsulated packet read at queuing time
*/
uint8_t _qosValue;
/**
* Set the packet QoS value
* (interface compatibility with Packet)
*/
inline void qosValue(const uint8_t qv) { _qosValue = qv; }
/**
* Get the packet QoS value
* (interface compatibility with Packet)
*/
inline uint8_t qosValue() const { return _qosValue; }
/**
* Get the packet RequestorID
* (interface compatibility with Packet)
*/
inline RequestorID requestorId() const { return _requestorId; }
/**
* Get the packet size
* (interface compatibility with Packet)
*/
inline unsigned int getSize() const { return size; }
/**
* Get the packet address
* (interface compatibility with Packet)
*/
inline Addr getAddr() const { return addr; }
/**
* Return true if its a read packet
* (interface compatibility with Packet)
*/
inline bool isRead() const { return read; }
/**
* Return true if its a write packet
* (interface compatibility with Packet)
*/
inline bool isWrite() const { return !read; }
/**
* Return true if its a DRAM access
*/
inline bool isDram() const { return dram; }
MemPacket(PacketPtr _pkt, bool is_read, bool is_dram, uint8_t _rank,
uint8_t _bank, uint32_t _row, uint16_t bank_id, Addr _addr,
unsigned int _size)
: entryTime(curTick()), readyTime(curTick()), pkt(_pkt),
_requestorId(pkt->requestorId()),
read(is_read), dram(is_dram), rank(_rank), bank(_bank), row(_row),
bankId(bank_id), addr(_addr), size(_size), burstHelper(NULL),
_qosValue(_pkt->qosValue())
{ }
};
// The memory packets are store in a multiple dequeue structure,
// based on their QoS priority
typedef std::deque<MemPacket*> MemPacketQueue;
/**
* The memory controller is a single-channel memory controller capturing
* the most important timing constraints associated with a
* contemporary controller. For multi-channel memory systems, the controller
* is combined with a crossbar model, with the channel address
* interleaving taking part in the crossbar.
*
* As a basic design principle, this controller
* model is not cycle callable, but instead uses events to: 1) decide
* when new decisions can be made, 2) when resources become available,
* 3) when things are to be considered done, and 4) when to send
* things back. The controller interfaces to media specific interfaces
* to enable flexible topoloties.
* Through these simple principles, the model delivers
* high performance, and lots of flexibility, allowing users to
* evaluate the system impact of a wide range of memory technologies.
*
* For more details, please see Hansson et al, "Simulating DRAM
* controllers for future system architecture exploration",
* Proc. ISPASS, 2014. If you use this model as part of your research
* please cite the paper.
*
*/
class MemCtrl : public qos::MemCtrl
{
private:
// For now, make use of a queued response port to avoid dealing with
// flow control for the responses being sent back
class MemoryPort : public QueuedResponsePort
{
RespPacketQueue queue;
MemCtrl& ctrl;
public:
MemoryPort(const std::string& name, MemCtrl& _ctrl);
protected:
Tick recvAtomic(PacketPtr pkt) override;
Tick recvAtomicBackdoor(
PacketPtr pkt, MemBackdoorPtr &backdoor) override;
void recvFunctional(PacketPtr pkt) override;
bool recvTimingReq(PacketPtr) override;
AddrRangeList getAddrRanges() const override;
};
/**
* Our incoming port, for a multi-ported controller add a crossbar
* in front of it
*/
MemoryPort port;
/**
* Remember if the memory system is in timing mode
*/
bool isTimingMode;
/**
* Remember if we have to retry a request when available.
*/
bool retryRdReq;
bool retryWrReq;
/**
* Bunch of things requires to setup "events" in gem5
* When event "respondEvent" occurs for example, the method
* processRespondEvent is called; no parameters are allowed
* in these methods
*/
void processNextReqEvent();
EventFunctionWrapper nextReqEvent;
void processRespondEvent();
EventFunctionWrapper respondEvent;
/**
* Check if the read queue has room for more entries
*
* @param pkt_count The number of entries needed in the read queue
* @return true if read queue is full, false otherwise
*/
bool readQueueFull(unsigned int pkt_count) const;
/**
* Check if the write queue has room for more entries
*
* @param pkt_count The number of entries needed in the write queue
* @return true if write queue is full, false otherwise
*/
bool writeQueueFull(unsigned int pkt_count) const;
/**
* When a new read comes in, first check if the write q has a
* pending request to the same address.\ If not, decode the
* address to populate rank/bank/row, create one or mutliple
* "mem_pkt", and push them to the back of the read queue.\
* If this is the only
* read request in the system, schedule an event to start
* servicing it.
*
* @param pkt The request packet from the outside world
* @param pkt_count The number of memory bursts the pkt
* @param is_dram Does this packet access DRAM?
* translate to. If pkt size is larger then one full burst,
* then pkt_count is greater than one.
*/
void addToReadQueue(PacketPtr pkt, unsigned int pkt_count, bool is_dram);
/**
* Decode the incoming pkt, create a mem_pkt and push to the
* back of the write queue. \If the write q length is more than
* the threshold specified by the user, ie the queue is beginning
* to get full, stop reads, and start draining writes.
*
* @param pkt The request packet from the outside world
* @param pkt_count The number of memory bursts the pkt
* @param is_dram Does this packet access DRAM?
* translate to. If pkt size is larger then one full burst,
* then pkt_count is greater than one.
*/
void addToWriteQueue(PacketPtr pkt, unsigned int pkt_count, bool is_dram);
/**
* Actually do the burst based on media specific access function.
* Update bus statistics when complete.
*
* @param mem_pkt The memory packet created from the outside world pkt
*/
void doBurstAccess(MemPacket* mem_pkt);
/**
* When a packet reaches its "readyTime" in the response Q,
* use the "access()" method in AbstractMemory to actually
* create the response packet, and send it back to the outside
* world requestor.
*
* @param pkt The packet from the outside world
* @param static_latency Static latency to add before sending the packet
*/
void accessAndRespond(PacketPtr pkt, Tick static_latency);
/**
* Determine if there is a packet that can issue.
*
* @param pkt The packet to evaluate
*/
bool packetReady(MemPacket* pkt);
/**
* Calculate the minimum delay used when scheduling a read-to-write
* transision.
* @param return minimum delay
*/
Tick minReadToWriteDataGap();
/**
* Calculate the minimum delay used when scheduling a write-to-read
* transision.
* @param return minimum delay
*/
Tick minWriteToReadDataGap();
/**
* The memory schduler/arbiter - picks which request needs to
* go next, based on the specified policy such as FCFS or FR-FCFS
* and moves it to the head of the queue.
* Prioritizes accesses to the same rank as previous burst unless
* controller is switching command type.
*
* @param queue Queued requests to consider
* @param extra_col_delay Any extra delay due to a read/write switch
* @return an iterator to the selected packet, else queue.end()
*/
MemPacketQueue::iterator chooseNext(MemPacketQueue& queue,
Tick extra_col_delay);
/**
* For FR-FCFS policy reorder the read/write queue depending on row buffer
* hits and earliest bursts available in memory
*
* @param queue Queued requests to consider
* @param extra_col_delay Any extra delay due to a read/write switch
* @return an iterator to the selected packet, else queue.end()
*/
MemPacketQueue::iterator chooseNextFRFCFS(MemPacketQueue& queue,
Tick extra_col_delay);
/**
* Calculate burst window aligned tick
*
* @param cmd_tick Initial tick of command
* @return burst window aligned tick
*/
Tick getBurstWindow(Tick cmd_tick);
/**
* Used for debugging to observe the contents of the queues.
*/
void printQs() const;
/**
* Burst-align an address.
*
* @param addr The potentially unaligned address
* @param is_dram Does this packet access DRAM?
*
* @return An address aligned to a memory burst
*/
Addr burstAlign(Addr addr, bool is_dram) const;
/**
* The controller's main read and write queues,
* with support for QoS reordering
*/
std::vector<MemPacketQueue> readQueue;
std::vector<MemPacketQueue> writeQueue;
/**
* To avoid iterating over the write queue to check for
* overlapping transactions, maintain a set of burst addresses
* that are currently queued. Since we merge writes to the same
* location we never have more than one address to the same burst
* address.
*/
std::unordered_set<Addr> isInWriteQueue;
/**
* Response queue where read packets wait after we're done working
* with them, but it's not time to send the response yet. The
* responses are stored separately mostly to keep the code clean
* and help with events scheduling. For all logical purposes such
* as sizing the read queue, this and the main read queue need to
* be added together.
*/
std::deque<MemPacket*> respQueue;
/**
* Holds count of commands issued in burst window starting at
* defined Tick. This is used to ensure that the command bandwidth
* does not exceed the allowable media constraints.
*/
std::unordered_multiset<Tick> burstTicks;
/**
* Create pointer to interface of the actual dram media when connected
*/
DRAMInterface* const dram;
/**
* Create pointer to interface of the actual nvm media when connected
*/
NVMInterface* const nvm;
/**
* The following are basic design parameters of the memory
* controller, and are initialized based on parameter values.
* The rowsPerBank is determined based on the capacity, number of
* ranks and banks, the burst size, and the row buffer size.
*/
const uint32_t readBufferSize;
const uint32_t writeBufferSize;
const uint32_t writeHighThreshold;
const uint32_t writeLowThreshold;
const uint32_t minWritesPerSwitch;
uint32_t writesThisTime;
uint32_t readsThisTime;
/**
* Memory controller configuration initialized based on parameter
* values.
*/
enums::MemSched memSchedPolicy;
/**
* Pipeline latency of the controller frontend. The frontend
* contribution is added to writes (that complete when they are in
* the write buffer) and reads that are serviced the write buffer.
*/
const Tick frontendLatency;
/**
* Pipeline latency of the backend and PHY. Along with the
* frontend contribution, this latency is added to reads serviced
* by the memory.
*/
const Tick backendLatency;
/**
* Length of a command window, used to check
* command bandwidth
*/
const Tick commandWindow;
/**
* Till when must we wait before issuing next RD/WR burst?
*/
Tick nextBurstAt;
Tick prevArrival;
/**
* The soonest you have to start thinking about the next request
* is the longest access time that can occur before
* nextBurstAt. Assuming you need to precharge, open a new row,
* and access, it is tRP + tRCD + tCL.
*/
Tick nextReqTime;
struct CtrlStats : public statistics::Group
{
CtrlStats(MemCtrl &ctrl);
void regStats() override;
MemCtrl &ctrl;
// All statistics that the model needs to capture
statistics::Scalar readReqs;
statistics::Scalar writeReqs;
statistics::Scalar readBursts;
statistics::Scalar writeBursts;
statistics::Scalar servicedByWrQ;
statistics::Scalar mergedWrBursts;
statistics::Scalar neitherReadNorWriteReqs;
// Average queue lengths
statistics::Average avgRdQLen;
statistics::Average avgWrQLen;
statistics::Scalar numRdRetry;
statistics::Scalar numWrRetry;
statistics::Vector readPktSize;
statistics::Vector writePktSize;
statistics::Vector rdQLenPdf;
statistics::Vector wrQLenPdf;
statistics::Histogram rdPerTurnAround;
statistics::Histogram wrPerTurnAround;
statistics::Scalar bytesReadWrQ;
statistics::Scalar bytesReadSys;
statistics::Scalar bytesWrittenSys;
// Average bandwidth
statistics::Formula avgRdBWSys;
statistics::Formula avgWrBWSys;
statistics::Scalar totGap;
statistics::Formula avgGap;
// per-requestor bytes read and written to memory
statistics::Vector requestorReadBytes;
statistics::Vector requestorWriteBytes;
// per-requestor bytes read and written to memory rate
statistics::Formula requestorReadRate;
statistics::Formula requestorWriteRate;
// per-requestor read and write serviced memory accesses
statistics::Vector requestorReadAccesses;
statistics::Vector requestorWriteAccesses;
// per-requestor read and write total memory access latency
statistics::Vector requestorReadTotalLat;
statistics::Vector requestorWriteTotalLat;
// per-requestor raed and write average memory access latency
statistics::Formula requestorReadAvgLat;
statistics::Formula requestorWriteAvgLat;
};
CtrlStats stats;
/**
* Upstream caches need this packet until true is returned, so
* hold it for deletion until a subsequent call
*/
std::unique_ptr<Packet> pendingDelete;
/**
* Select either the read or write queue
*
* @param is_read The current burst is a read, select read queue
* @return a reference to the appropriate queue
*/
std::vector<MemPacketQueue>& selQueue(bool is_read)
{
return (is_read ? readQueue : writeQueue);
};
/**
* Remove commands that have already issued from burstTicks
*/
void pruneBurstTick();
public:
MemCtrl(const MemCtrlParams &p);
/**
* Ensure that all interfaced have drained commands
*
* @return bool flag, set once drain complete
*/
bool allIntfDrained() const;
DrainState drain() override;
/**
* Check for command bus contention for single cycle command.
* If there is contention, shift command to next burst.
* Check verifies that the commands issued per burst is less
* than a defined max number, maxCommandsPerWindow.
* Therefore, contention per cycle is not verified and instead
* is done based on a burst window.
*
* @param cmd_tick Initial tick of command, to be verified
* @param max_cmds_per_burst Number of commands that can issue
* in a burst window
* @return tick for command issue without contention
*/
Tick verifySingleCmd(Tick cmd_tick, Tick max_cmds_per_burst);
/**
* Check for command bus contention for multi-cycle (2 currently)
* command. If there is contention, shift command(s) to next burst.
* Check verifies that the commands issued per burst is less
* than a defined max number, maxCommandsPerWindow.
* Therefore, contention per cycle is not verified and instead
* is done based on a burst window.
*
* @param cmd_tick Initial tick of command, to be verified
* @param max_multi_cmd_split Maximum delay between commands
* @param max_cmds_per_burst Number of commands that can issue
* in a burst window
* @return tick for command issue without contention
*/
Tick verifyMultiCmd(Tick cmd_tick, Tick max_cmds_per_burst,
Tick max_multi_cmd_split = 0);
/**
* Is there a respondEvent scheduled?
*
* @return true if event is scheduled
*/
bool respondEventScheduled() const { return respondEvent.scheduled(); }
/**
* Is there a read/write burst Event scheduled?
*
* @return true if event is scheduled
*/
bool requestEventScheduled() const { return nextReqEvent.scheduled(); }
/**
* restart the controller
* This can be used by interfaces to restart the
* scheduler after maintainence commands complete
*
* @param Tick to schedule next event
*/
void restartScheduler(Tick tick) { schedule(nextReqEvent, tick); }
/**
* Check the current direction of the memory channel
*
* @param next_state Check either the current or next bus state
* @return True when bus is currently in a read state
*/
bool inReadBusState(bool next_state) const;
/**
* Check the current direction of the memory channel
*
* @param next_state Check either the current or next bus state
* @return True when bus is currently in a write state
*/
bool inWriteBusState(bool next_state) const;
Port &getPort(const std::string &if_name,
PortID idx=InvalidPortID) override;
virtual void init() override;
virtual void startup() override;
virtual void drainResume() override;
protected:
Tick recvAtomic(PacketPtr pkt);
Tick recvAtomicBackdoor(PacketPtr pkt, MemBackdoorPtr &backdoor);
void recvFunctional(PacketPtr pkt);
bool recvTimingReq(PacketPtr pkt);
};
} // namespace gem5
#endif //__MEM_CTRL_HH__