Files
gem5/src/mem/hbm_ctrl.cc
Ayaz Akram 60317f99fe mem: Add an HBM memory controller
This change adds an hbm memory controller in gem5,
which is capable of controlling two hbm memory interfaces
(two pseudo channels). HBMCtrl inherits from MemCtrl
and tries to reuse most of the MemCtrl functions for two
different dram interfaces. Morever, a notion of pseudo
channel is added in the memory interface itself, to make
sure that the scheduling decisions in any interface are
based on the pkts for that pseudo channel only. Also, the
command bandwidth checks are divided into row and column
commands, which are shared by both pseudo channels.

Change-Id: Ie2ee8183d0f7f744aff2ed05cabc75fec3ea2171
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/59732
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Maintainer: Jason Lowe-Power <power.jg@gmail.com>
Reviewed-by: Wendy Elsasser <welsasser@rambus.com>
Tested-by: kokoro <noreply+kokoro@google.com>
2022-06-06 18:31:06 +00:00

493 lines
16 KiB
C++

/*
* Copyright (c) 2022 The Regents of the University of California
* 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.
*/
#include "mem/hbm_ctrl.hh"
#include "base/trace.hh"
#include "debug/DRAM.hh"
#include "debug/Drain.hh"
#include "debug/MemCtrl.hh"
#include "debug/QOS.hh"
#include "mem/dram_interface.hh"
#include "mem/mem_interface.hh"
#include "sim/system.hh"
namespace gem5
{
namespace memory
{
HBMCtrl::HBMCtrl(const HBMCtrlParams &p) :
MemCtrl(p),
retryRdReqPC1(false), retryWrReqPC1(false),
nextReqEventPC1([this] {processNextReqEvent(pc1Int, respQueuePC1,
respondEventPC1, nextReqEventPC1, retryWrReqPC1);},
name()),
respondEventPC1([this] {processRespondEvent(pc1Int, respQueuePC1,
respondEventPC1, retryRdReqPC1); }, name()),
pc1Int(p.dram_2),
partitionedQ(p.partitioned_q)
{
DPRINTF(MemCtrl, "Setting up HBM controller\n");
pc0Int = dynamic_cast<DRAMInterface*>(dram);
assert(dynamic_cast<DRAMInterface*>(p.dram_2) != nullptr);
readBufferSize = pc0Int->readBufferSize + pc1Int->readBufferSize;
writeBufferSize = pc0Int->writeBufferSize + pc1Int->writeBufferSize;
fatal_if(!pc0Int, "Memory controller must have pc0 interface");
fatal_if(!pc1Int, "Memory controller must have pc1 interface");
pc0Int->setCtrl(this, commandWindow, 0);
pc1Int->setCtrl(this, commandWindow, 1);
if (partitionedQ) {
writeHighThreshold = (writeBufferSize * (p.write_high_thresh_perc/2)
/ 100.0);
writeLowThreshold = (writeBufferSize * (p.write_low_thresh_perc/2)
/ 100.0);
} else {
writeHighThreshold = (writeBufferSize * p.write_high_thresh_perc
/ 100.0);
writeLowThreshold = (writeBufferSize * p.write_low_thresh_perc
/ 100.0);
}
}
void
HBMCtrl::init()
{
MemCtrl::init();
}
void
HBMCtrl::startup()
{
MemCtrl::startup();
isTimingMode = system()->isTimingMode();
if (isTimingMode) {
// shift the bus busy time sufficiently far ahead that we never
// have to worry about negative values when computing the time for
// the next request, this will add an insignificant bubble at the
// start of simulation
pc1Int->nextBurstAt = curTick() + pc1Int->commandOffset();
}
}
Tick
HBMCtrl::recvAtomic(PacketPtr pkt)
{
Tick latency = 0;
if (pc0Int->getAddrRange().contains(pkt->getAddr())) {
latency = MemCtrl::recvAtomicLogic(pkt, pc0Int);
} else if (pc1Int->getAddrRange().contains(pkt->getAddr())) {
latency = MemCtrl::recvAtomicLogic(pkt, pc1Int);
} else {
panic("Can't handle address range for packet %s\n", pkt->print());
}
return latency;
}
void
HBMCtrl::recvFunctional(PacketPtr pkt)
{
bool found = MemCtrl::recvFunctionalLogic(pkt, pc0Int);
if (!found) {
found = MemCtrl::recvFunctionalLogic(pkt, pc1Int);
}
if (!found) {
panic("Can't handle address range for packet %s\n", pkt->print());
}
}
Tick
HBMCtrl::recvAtomicBackdoor(PacketPtr pkt, MemBackdoorPtr &backdoor)
{
Tick latency = recvAtomic(pkt);
if (pc0Int && pc0Int->getAddrRange().contains(pkt->getAddr())) {
pc0Int->getBackdoor(backdoor);
} else if (pc1Int && pc1Int->getAddrRange().contains(pkt->getAddr())) {
pc1Int->getBackdoor(backdoor);
}
else {
panic("Can't handle address range for packet %s\n",
pkt->print());
}
return latency;
}
bool
HBMCtrl::writeQueueFullPC0(unsigned int neededEntries) const
{
DPRINTF(MemCtrl,
"Write queue limit %d, PC0 size %d, entries needed %d\n",
writeBufferSize, writeQueueSizePC0, neededEntries);
unsigned int wrsize_new = (writeQueueSizePC0 + neededEntries);
return wrsize_new > (writeBufferSize/2);
}
bool
HBMCtrl::writeQueueFullPC1(unsigned int neededEntries) const
{
DPRINTF(MemCtrl,
"Write queue limit %d, PC1 size %d, entries needed %d\n",
writeBufferSize, writeQueueSizePC1, neededEntries);
unsigned int wrsize_new = (writeQueueSizePC1 + neededEntries);
return wrsize_new > (writeBufferSize/2);
}
bool
HBMCtrl::readQueueFullPC0(unsigned int neededEntries) const
{
DPRINTF(MemCtrl,
"Read queue limit %d, PC0 size %d, entries needed %d\n",
readBufferSize, readQueueSizePC0 + respQueue.size(),
neededEntries);
unsigned int rdsize_new = readQueueSizePC0 + respQueue.size()
+ neededEntries;
return rdsize_new > (readBufferSize/2);
}
bool
HBMCtrl::readQueueFullPC1(unsigned int neededEntries) const
{
DPRINTF(MemCtrl,
"Read queue limit %d, PC1 size %d, entries needed %d\n",
readBufferSize, readQueueSizePC1 + respQueuePC1.size(),
neededEntries);
unsigned int rdsize_new = readQueueSizePC1 + respQueuePC1.size()
+ neededEntries;
return rdsize_new > (readBufferSize/2);
}
bool
HBMCtrl::readQueueFull(unsigned int neededEntries) const
{
DPRINTF(MemCtrl,
"HBMCtrl: Read queue limit %d, entries needed %d\n",
readBufferSize, neededEntries);
unsigned int rdsize_new = totalReadQueueSize + respQueue.size() +
respQueuePC1.size() + neededEntries;
return rdsize_new > readBufferSize;
}
bool
HBMCtrl::recvTimingReq(PacketPtr pkt)
{
// This is where we enter from the outside world
DPRINTF(MemCtrl, "recvTimingReq: request %s addr %#x size %d\n",
pkt->cmdString(), pkt->getAddr(), pkt->getSize());
panic_if(pkt->cacheResponding(), "Should not see packets where cache "
"is responding");
panic_if(!(pkt->isRead() || pkt->isWrite()),
"Should only see read and writes at memory controller\n");
// Calc avg gap between requests
if (prevArrival != 0) {
stats.totGap += curTick() - prevArrival;
}
prevArrival = curTick();
// What type of media does this packet access?
bool is_pc0;
// TODO: make the interleaving bit across pseudo channels a parameter
if (bits(pkt->getAddr(), 6) == 0) {
is_pc0 = true;
} else {
is_pc0 = false;
}
// Find out how many memory packets a pkt translates to
// If the burst size is equal or larger than the pkt size, then a pkt
// translates to only one memory packet. Otherwise, a pkt translates to
// multiple memory packets
unsigned size = pkt->getSize();
uint32_t burst_size = pc0Int->bytesPerBurst();
unsigned offset = pkt->getAddr() & (burst_size - 1);
unsigned int pkt_count = divCeil(offset + size, burst_size);
// run the QoS scheduler and assign a QoS priority value to the packet
qosSchedule({&readQueue, &writeQueue}, burst_size, pkt);
// check local buffers and do not accept if full
if (pkt->isWrite()) {
if (is_pc0) {
if (partitionedQ ? writeQueueFullPC0(pkt_count) :
writeQueueFull(pkt_count))
{
DPRINTF(MemCtrl, "Write queue full, not accepting\n");
// remember that we have to retry this port
MemCtrl::retryWrReq = true;
stats.numWrRetry++;
return false;
} else {
addToWriteQueue(pkt, pkt_count, pc0Int);
stats.writeReqs++;
stats.bytesWrittenSys += size;
}
} else {
if (partitionedQ ? writeQueueFullPC1(pkt_count) :
writeQueueFull(pkt_count))
{
DPRINTF(MemCtrl, "Write queue full, not accepting\n");
// remember that we have to retry this port
retryWrReqPC1 = true;
stats.numWrRetry++;
return false;
} else {
addToWriteQueue(pkt, pkt_count, pc1Int);
stats.writeReqs++;
stats.bytesWrittenSys += size;
}
}
} else {
assert(pkt->isRead());
assert(size != 0);
if (is_pc0) {
if (partitionedQ ? readQueueFullPC0(pkt_count) :
HBMCtrl::readQueueFull(pkt_count)) {
DPRINTF(MemCtrl, "Read queue full, not accepting\n");
// remember that we have to retry this port
retryRdReqPC1 = true;
stats.numRdRetry++;
return false;
} else {
if (!addToReadQueue(pkt, pkt_count, pc0Int)) {
if (!nextReqEvent.scheduled()) {
DPRINTF(MemCtrl, "Request scheduled immediately\n");
schedule(nextReqEvent, curTick());
}
}
stats.readReqs++;
stats.bytesReadSys += size;
}
} else {
if (partitionedQ ? readQueueFullPC1(pkt_count) :
HBMCtrl::readQueueFull(pkt_count)) {
DPRINTF(MemCtrl, "Read queue full, not accepting\n");
// remember that we have to retry this port
retryRdReqPC1 = true;
stats.numRdRetry++;
return false;
} else {
if (!addToReadQueue(pkt, pkt_count, pc1Int)) {
if (!nextReqEventPC1.scheduled()) {
DPRINTF(MemCtrl, "Request scheduled immediately\n");
schedule(nextReqEventPC1, curTick());
}
}
stats.readReqs++;
stats.bytesReadSys += size;
}
}
}
return true;
}
void
HBMCtrl::pruneRowBurstTick()
{
auto it = rowBurstTicks.begin();
while (it != rowBurstTicks.end()) {
auto current_it = it++;
if (MemCtrl::getBurstWindow(curTick()) > *current_it) {
DPRINTF(MemCtrl, "Removing burstTick for %d\n", *current_it);
rowBurstTicks.erase(current_it);
}
}
}
void
HBMCtrl::pruneColBurstTick()
{
auto it = colBurstTicks.begin();
while (it != colBurstTicks.end()) {
auto current_it = it++;
if (MemCtrl::getBurstWindow(curTick()) > *current_it) {
DPRINTF(MemCtrl, "Removing burstTick for %d\n", *current_it);
colBurstTicks.erase(current_it);
}
}
}
void
HBMCtrl::pruneBurstTick()
{
pruneRowBurstTick();
pruneColBurstTick();
}
Tick
HBMCtrl::verifySingleCmd(Tick cmd_tick, Tick max_cmds_per_burst, bool row_cmd)
{
// start with assumption that there is no contention on command bus
Tick cmd_at = cmd_tick;
// get tick aligned to burst window
Tick burst_tick = MemCtrl::getBurstWindow(cmd_tick);
// verify that we have command bandwidth to issue the command
// if not, iterate over next window(s) until slot found
if (row_cmd) {
while (rowBurstTicks.count(burst_tick) >= max_cmds_per_burst) {
DPRINTF(MemCtrl, "Contention found on row command bus at %d\n",
burst_tick);
burst_tick += commandWindow;
cmd_at = burst_tick;
}
DPRINTF(MemCtrl, "Now can send a row cmd_at %d\n",
cmd_at);
rowBurstTicks.insert(burst_tick);
} else {
while (colBurstTicks.count(burst_tick) >= max_cmds_per_burst) {
DPRINTF(MemCtrl, "Contention found on col command bus at %d\n",
burst_tick);
burst_tick += commandWindow;
cmd_at = burst_tick;
}
DPRINTF(MemCtrl, "Now can send a col cmd_at %d\n",
cmd_at);
colBurstTicks.insert(burst_tick);
}
return cmd_at;
}
Tick
HBMCtrl::verifyMultiCmd(Tick cmd_tick, Tick max_cmds_per_burst,
Tick max_multi_cmd_split)
{
// start with assumption that there is no contention on command bus
Tick cmd_at = cmd_tick;
// get tick aligned to burst window
Tick burst_tick = MemCtrl::getBurstWindow(cmd_tick);
// Command timing requirements are from 2nd command
// Start with assumption that 2nd command will issue at cmd_at and
// find prior slot for 1st command to issue
// Given a maximum latency of max_multi_cmd_split between the commands,
// find the burst at the maximum latency prior to cmd_at
Tick burst_offset = 0;
Tick first_cmd_offset = cmd_tick % commandWindow;
while (max_multi_cmd_split > (first_cmd_offset + burst_offset)) {
burst_offset += commandWindow;
}
// get the earliest burst aligned address for first command
// ensure that the time does not go negative
Tick first_cmd_tick = burst_tick - std::min(burst_offset, burst_tick);
// Can required commands issue?
bool first_can_issue = false;
bool second_can_issue = false;
// verify that we have command bandwidth to issue the command(s)
while (!first_can_issue || !second_can_issue) {
bool same_burst = (burst_tick == first_cmd_tick);
auto first_cmd_count = rowBurstTicks.count(first_cmd_tick);
auto second_cmd_count = same_burst ?
first_cmd_count + 1 : rowBurstTicks.count(burst_tick);
first_can_issue = first_cmd_count < max_cmds_per_burst;
second_can_issue = second_cmd_count < max_cmds_per_burst;
if (!second_can_issue) {
DPRINTF(MemCtrl, "Contention (cmd2) found on command bus at %d\n",
burst_tick);
burst_tick += commandWindow;
cmd_at = burst_tick;
}
// Verify max_multi_cmd_split isn't violated when command 2 is shifted
// If commands initially were issued in same burst, they are
// now in consecutive bursts and can still issue B2B
bool gap_violated = !same_burst &&
((burst_tick - first_cmd_tick) > max_multi_cmd_split);
if (!first_can_issue || (!second_can_issue && gap_violated)) {
DPRINTF(MemCtrl, "Contention (cmd1) found on command bus at %d\n",
first_cmd_tick);
first_cmd_tick += commandWindow;
}
}
// Add command to burstTicks
rowBurstTicks.insert(burst_tick);
rowBurstTicks.insert(first_cmd_tick);
return cmd_at;
}
void
HBMCtrl::drainResume()
{
MemCtrl::drainResume();
if (!isTimingMode && system()->isTimingMode()) {
// if we switched to timing mode, kick things into action,
// and behave as if we restored from a checkpoint
startup();
pc1Int->startup();
} else if (isTimingMode && !system()->isTimingMode()) {
// if we switch from timing mode, stop the refresh events to
// not cause issues with KVM
if (pc1Int) {
pc1Int->drainRanks();
}
}
// update the mode
isTimingMode = system()->isTimingMode();
}
} // namespace memory
} // namespace gem5