Clock: Rework clocks to avoid tick-to-cycle transformations

This patch introduces the notion of a clock update function that aims
to avoid costly divisions when turning the current tick into a
cycle. Each clocked object advances a private (hidden) cycle member
and a tick member and uses these to implement functions for getting
the tick of the next cycle, or the tick of a cycle some time in the
future.

In the different modules using the clocks, changes are made to avoid
counting in ticks only to later translate to cycles. There are a few
oddities in how the O3 and inorder CPU count idle cycles, as seen by a
few locations where a cycle is subtracted in the calculation. This is
done such that the regression does not change any stats, but should be
revisited in a future patch.

Another, much needed, change that is not done as part of this patch is
to introduce a new typedef uint64_t Cycle to be able to at least hint
at the unit of the variables counting Ticks vs Cycles. This will be
done as a follow-up patch.

As an additional follow up, the thread context still uses ticks for
the book keeping of last activate and last suspend and this should
probably also be changed into cycles as well.

3 files changed, 25 insertions, 29 deletions

diff --git a/src/cpu/simple/atomic.cc b/src/cpu/simple/atomic.cc
index fc6724939..d1b0391fc 100644
--- a/src/cpu/simple/atomic.cc
+++ b/src/cpu/simple/atomic.cc
@@ -208,10 +208,10 @@ AtomicSimpleCPU::activateContext(ThreadID thread_num, int delay)
     assert(!tickEvent.scheduled());
 
     notIdleFraction++;
-    numCycles += tickToCycles(thread->lastActivate - thread->lastSuspend);
+    numCycles += tickToCycle(thread->lastActivate - thread->lastSuspend);
 
     //Make sure ticks are still on multiples of cycles
-    schedule(tickEvent, nextCycle(curTick() + ticks(delay)));
+    schedule(tickEvent, clockEdge(delay));
     _status = Running;
 }
 
@@ -518,7 +518,7 @@ AtomicSimpleCPU::tick()
                 stall_ticks += dcache_latency;
 
             if (stall_ticks) {
-                Tick stall_cycles = stall_ticks / ticks(1);
+                Tick stall_cycles = stall_ticks / clockPeriod();
                 Tick aligned_stall_ticks = ticks(stall_cycles);
 
                 if (aligned_stall_ticks < stall_ticks)
@@ -533,8 +533,8 @@ AtomicSimpleCPU::tick()
     }
 
     // instruction takes at least one cycle
-    if (latency < ticks(1))
-        latency = ticks(1);
+    if (latency < clockPeriod())
+        latency = clockPeriod();
 
     if (_status != Idle)
         schedule(tickEvent, curTick() + latency);
diff --git a/src/cpu/simple/timing.cc b/src/cpu/simple/timing.cc
index 9022845ce..5437e77aa 100644
--- a/src/cpu/simple/timing.cc
+++ b/src/cpu/simple/timing.cc
@@ -87,13 +87,11 @@ TimingSimpleCPU::TimingCPUPort::TickEvent::schedule(PacketPtr _pkt, Tick t)
 
 TimingSimpleCPU::TimingSimpleCPU(TimingSimpleCPUParams *p)
     : BaseSimpleCPU(p), fetchTranslation(this), icachePort(this),
-    dcachePort(this), fetchEvent(this)
+      dcachePort(this), ifetch_pkt(NULL), dcache_pkt(NULL), previousCycle(0),
+      fetchEvent(this)
 {
     _status = Idle;
 
-    ifetch_pkt = dcache_pkt = NULL;
-    drainEvent = NULL;
-    previousTick = 0;
     changeState(SimObject::Running);
     system->totalNumInsts = 0;
 }
@@ -156,7 +154,7 @@ TimingSimpleCPU::switchOut()
 {
     assert(_status == Running || _status == Idle);
     _status = SwitchedOut;
-    numCycles += tickToCycles(curTick() - previousTick);
+    numCycles += curCycle() - previousCycle;
 
     // If we've been scheduled to resume but are then told to switch out,
     // we'll need to cancel it.
@@ -184,7 +182,7 @@ TimingSimpleCPU::takeOverFrom(BaseCPU *oldCPU)
         _status = Idle;
     }
     assert(threadContexts.size() == 1);
-    previousTick = curTick();
+    previousCycle = curCycle();
 }
 
 
@@ -202,7 +200,7 @@ TimingSimpleCPU::activateContext(ThreadID thread_num, int delay)
     _status = Running;
 
     // kick things off by initiating the fetch of the next instruction
-    schedule(fetchEvent, nextCycle(curTick() + ticks(delay)));
+    schedule(fetchEvent, clockEdge(delay));
 }
 
 
@@ -231,9 +229,8 @@ TimingSimpleCPU::handleReadPacket(PacketPtr pkt)
 {
     RequestPtr req = pkt->req;
     if (req->isMmappedIpr()) {
-        Tick delay;
-        delay = TheISA::handleIprRead(thread->getTC(), pkt);
-        new IprEvent(pkt, this, nextCycle(curTick() + delay));
+        Tick delay = TheISA::handleIprRead(thread->getTC(), pkt);
+        new IprEvent(pkt, this, clockEdge(delay));
         _status = DcacheWaitResponse;
         dcache_pkt = NULL;
     } else if (!dcachePort.sendTimingReq(pkt)) {
@@ -322,8 +319,8 @@ TimingSimpleCPU::translationFault(Fault fault)
 {
     // fault may be NoFault in cases where a fault is suppressed,
     // for instance prefetches.
-    numCycles += tickToCycles(curTick() - previousTick);
-    previousTick = curTick();
+    numCycles += curCycle() - previousCycle;
+    previousCycle = curCycle();
 
     if (traceData) {
         // Since there was a fault, we shouldn't trace this instruction.
@@ -446,9 +443,8 @@ TimingSimpleCPU::handleWritePacket()
 {
     RequestPtr req = dcache_pkt->req;
     if (req->isMmappedIpr()) {
-        Tick delay;
-        delay = TheISA::handleIprWrite(thread->getTC(), dcache_pkt);
-        new IprEvent(dcache_pkt, this, nextCycle(curTick() + delay));
+        Tick delay = TheISA::handleIprWrite(thread->getTC(), dcache_pkt);
+        new IprEvent(dcache_pkt, this, clockEdge(delay));
         _status = DcacheWaitResponse;
         dcache_pkt = NULL;
     } else if (!dcachePort.sendTimingReq(dcache_pkt)) {
@@ -567,8 +563,8 @@ TimingSimpleCPU::fetch()
         _status = IcacheWaitResponse;
         completeIfetch(NULL);
 
-        numCycles += tickToCycles(curTick() - previousTick);
-        previousTick = curTick();
+        numCycles += curCycle() - previousCycle;
+        previousCycle = curCycle();
     }
 }
 
@@ -600,8 +596,8 @@ TimingSimpleCPU::sendFetch(Fault fault, RequestPtr req, ThreadContext *tc)
         advanceInst(fault);
     }
 
-    numCycles += tickToCycles(curTick() - previousTick);
-    previousTick = curTick();
+    numCycles += curCycle() - previousCycle;
+    previousCycle = curCycle();
 }
 
 
@@ -647,8 +643,8 @@ TimingSimpleCPU::completeIfetch(PacketPtr pkt)
 
     _status = Running;
 
-    numCycles += tickToCycles(curTick() - previousTick);
-    previousTick = curTick();
+    numCycles += curCycle() - previousCycle;
+    previousCycle = curCycle();
 
     if (getState() == SimObject::Draining) {
         if (pkt) {
@@ -754,8 +750,8 @@ TimingSimpleCPU::completeDataAccess(PacketPtr pkt)
     assert(_status == DcacheWaitResponse || _status == DTBWaitResponse ||
            pkt->req->getFlags().isSet(Request::NO_ACCESS));
 
-    numCycles += tickToCycles(curTick() - previousTick);
-    previousTick = curTick();
+    numCycles += curCycle() - previousCycle;
+    previousCycle = curCycle();
 
     if (pkt->senderState) {
         SplitFragmentSenderState * send_state =
diff --git a/src/cpu/simple/timing.hh b/src/cpu/simple/timing.hh
index b6b78c5db..c4d1573af 100644
--- a/src/cpu/simple/timing.hh
+++ b/src/cpu/simple/timing.hh
@@ -234,7 +234,7 @@ class TimingSimpleCPU : public BaseSimpleCPU
     PacketPtr ifetch_pkt;
     PacketPtr dcache_pkt;
 
-    Tick previousTick;
+    Tick previousCycle;
 
   protected: