src/mem/cache/prefetch/bop.hh

/**
 * Copyright (c) 2018 Metempsy Technology Consulting
 * 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.
 */

/**
 * Implementation of the 'A Best-Offset Prefetcher'
 * Reference:
 *   Michaud, P. (2015, June). A best-offset prefetcher.
 *   In 2nd Data Prefetching Championship.
 */

#ifndef __MEM_CACHE_PREFETCH_BOP_HH__
#define __MEM_CACHE_PREFETCH_BOP_HH__

#include <queue>

#include "mem/cache/prefetch/queued.hh"
#include "mem/packet.hh"

namespace gem5
{

struct BOPPrefetcherParams;

namespace prefetch
{

class BOP : public Queued
{
    private:

        enum RRWay
        {
            Left,
            Right
        };

        /** Learning phase parameters */
        const unsigned int scoreMax;
        const unsigned int roundMax;
        const unsigned int badScore;
        /** Recent requests table parameteres */
        const unsigned int rrEntries;
        const unsigned int tagMask;
        /** Delay queue parameters */
        const bool         delayQueueEnabled;
        const unsigned int delayQueueSize;
        const unsigned int delayTicks;

        std::vector<Addr> rrLeft;
        std::vector<Addr> rrRight;

        /** Structure to save the offset and the score */
        typedef std::pair<int16_t, uint8_t> OffsetListEntry;
        std::vector<OffsetListEntry> offsetsList;

        /** In a first implementation of the BO prefetcher, both banks of the
         *  RR were written simultaneously when a prefetched line is inserted
         *  into the cache. Adding the delay queue tries to avoid always
         *  striving for timeless prefetches, which has been found to not
         *  always being optimal.
         */
        struct DelayQueueEntry
        {
            Addr baseAddr;
            Tick processTick;

            DelayQueueEntry(Addr x, Tick t) : baseAddr(x), processTick(t)
            {}
        };

        std::deque<DelayQueueEntry> delayQueue;

        /** Event to handle the delay queue processing */
        void delayQueueEventWrapper();
        EventFunctionWrapper delayQueueEvent;

        /** Hardware prefetcher enabled */
        bool issuePrefetchRequests;
        /** Current best offset to issue prefetches */
        Addr bestOffset;
        /** Current best offset found in the learning phase */
        Addr phaseBestOffset;
        /** Current test offset index */
        std::vector<OffsetListEntry>::iterator offsetsListIterator;
        /** Max score found so far */
        unsigned int bestScore;
        /** Current round */
        unsigned int round;

        /** Generate a hash for the specified address to index the RR table
         *  @param addr: address to hash
         *  @param way:  RR table to which is addressed (left/right)
         */
        unsigned int hash(Addr addr, unsigned int way) const;

        /** Insert the specified address into the RR table
         *  @param addr: address to insert
         *  @param way: RR table to which the address will be inserted
         */
        void insertIntoRR(Addr addr, unsigned int way);

        /** Insert the specified address into the delay queue. This will
         *  trigger an event after the delay cycles pass
         *  @param addr: address to insert into the delay queue
         */
        void insertIntoDelayQueue(Addr addr);

        /** Reset all the scores from the offset list */
        void resetScores();

        /** Generate the tag for the specified address based on the tag bits
         *  and the block size
         *  @param addr: address to get the tag from
        */
        Addr tag(Addr addr) const;

        /** Test if @X-O is hitting in the RR table to update the
            offset score */
        bool testRR(Addr) const;

        /** Learning phase of the BOP. Update the intermediate values of the
            round and update the best offset if found */
        void bestOffsetLearning(Addr);

        /** Update the RR right table after a prefetch fill */
        void notifyFill(const PacketPtr& pkt) override;

    public:

        BOP(const BOPPrefetcherParams &p);
        ~BOP() = default;

        void calculatePrefetch(const PrefetchInfo &pfi,
                               std::vector<AddrPriority> &addresses) override;
};

} // namespace prefetch
} // namespace gem5

#endif /* __MEM_CACHE_PREFETCH_BOP_HH__ */