cpu: o3: replace issueLatency with bool pipelined

Currently, each op class has a parameter issueLat that denotes the cycles after
which another op of the same class can be issued.  As of now, this latency can
either be one cycle (fully pipelined) or same as execution latency of the op
(not at all pipelined).  The fact that issueLat is a parameter of type Cycles
makes one believe that it can be set to any value.  To avoid the confusion, the
parameter is being renamed as 'pipelined' with type boolean.  If set to true,
the op would execute in a fully pipelined fashion. Otherwise, it would execute
in an unpipelined fashion.

configs/common/O3_ARM_v7a.py

@@ -36,9 +36,9 @@ class O3_ARM_v7a_Simple_Int(FUDesc):
 
 # Complex ALU instructions have a variable latencies
 class O3_ARM_v7a_Complex_Int(FUDesc):
-    opList = [ OpDesc(opClass='IntMult', opLat=3, issueLat=1),
+    opList = [ OpDesc(opClass='IntMult', opLat=3, pipelined=True),
-               OpDesc(opClass='IntDiv', opLat=12, issueLat=12),
+               OpDesc(opClass='IntDiv', opLat=12, pipelined=False),
-               OpDesc(opClass='IprAccess', opLat=3, issueLat=1) ]
+               OpDesc(opClass='IprAccess', opLat=3, pipelined=True) ]
     count = 1
 
 
@@ -67,8 +67,8 @@ class O3_ARM_v7a_FP(FUDesc):
                OpDesc(opClass='FloatAdd', opLat=5),
                OpDesc(opClass='FloatCmp', opLat=5),
                OpDesc(opClass='FloatCvt', opLat=5),
-               OpDesc(opClass='FloatDiv', opLat=9, issueLat=9),
+               OpDesc(opClass='FloatDiv', opLat=9, pipelined=False),
-               OpDesc(opClass='FloatSqrt', opLat=33, issueLat=33),
+               OpDesc(opClass='FloatSqrt', opLat=33, pipelined=False),
                OpDesc(opClass='FloatMult', opLat=4) ]
     count = 2
 

src/cpu/FuncUnit.py

@@ -54,9 +54,10 @@ class OpClass(Enum):
 class OpDesc(SimObject):
     type = 'OpDesc'
     cxx_header = "cpu/func_unit.hh"
-    issueLat = Param.Cycles(1, "cycles until another can be issued")
     opClass = Param.OpClass("type of operation")
     opLat = Param.Cycles(1, "cycles until result is available")
+    pipelined = Param.Bool(True, "set to true when the functional unit for"
+        "this op is fully pipelined. False means not pipelined at all.")
 
 class FUDesc(SimObject):
     type = 'FUDesc'

src/cpu/func_unit.cc

@@ -52,7 +52,7 @@ FuncUnit::FuncUnit(const FuncUnit &fu)
 
     for (int i = 0; i < Num_OpClasses; ++i) {
         opLatencies[i] = fu.opLatencies[i];
-        issueLatencies[i] = fu.issueLatencies[i];
+        pipelined[i] = fu.pipelined[i];
     }
 
     capabilityList = fu.capabilityList;
@@ -60,15 +60,15 @@ FuncUnit::FuncUnit(const FuncUnit &fu)
 
 
 void
-FuncUnit::addCapability(OpClass cap, unsigned oplat, unsigned issuelat)
+FuncUnit::addCapability(OpClass cap, unsigned oplat, bool pipeline)
 {
-    if (issuelat == 0 || oplat == 0)
+    if (oplat == 0)
         panic("FuncUnit:  you don't really want a zero-cycle latency do you?");
 
     capabilityList.set(cap);
 
     opLatencies[cap] = oplat;
-    issueLatencies[cap] = issuelat;
+    pipelined[cap] = pipeline;
 }
 
 bool
@@ -89,10 +89,10 @@ FuncUnit::opLatency(OpClass cap)
     return opLatencies[cap];
 }
 
-unsigned
+bool
-FuncUnit::issueLatency(OpClass capability)
+FuncUnit::isPipelined(OpClass capability)
 {
-    return issueLatencies[capability];
+    return pipelined[capability];
 }
 
 ////////////////////////////////////////////////////////////////////////////

src/cpu/func_unit.hh

@@ -52,11 +52,11 @@ class OpDesc : public SimObject
   public:
     OpClass opClass;
     Cycles opLat;
-    Cycles issueLat;
+    bool pipelined;
 
     OpDesc(const OpDescParams *p)
         : SimObject(p), opClass(p->opClass), opLat(p->opLat),
-          issueLat(p->issueLat) {};
+          pipelined(p->pipelined) {};
 };
 
 class FUDesc : public SimObject
@@ -85,7 +85,7 @@ class FuncUnit
 {
   private:
     unsigned opLatencies[Num_OpClasses];
-    unsigned issueLatencies[Num_OpClasses];
+    bool pipelined[Num_OpClasses];
     std::bitset<Num_OpClasses> capabilityList;
 
   public:
@@ -94,13 +94,13 @@ class FuncUnit
 
     std::string name;
 
-    void addCapability(OpClass cap, unsigned oplat, unsigned issuelat);
+    void addCapability(OpClass cap, unsigned oplat, bool pipelined);
 
     bool provides(OpClass capability);
     std::bitset<Num_OpClasses> capabilities();
 
     unsigned &opLatency(OpClass capability);
-    unsigned issueLatency(OpClass capability);
+    bool isPipelined(OpClass capability);
 };
 
 #endif // __FU_POOL_HH__

src/cpu/o3/FuncUnitConfig.py

@@ -49,7 +49,7 @@ class IntALU(FUDesc):
 
 class IntMultDiv(FUDesc):
     opList = [ OpDesc(opClass='IntMult', opLat=3),
-               OpDesc(opClass='IntDiv', opLat=20, issueLat=19) ]
+               OpDesc(opClass='IntDiv', opLat=20, pipelined=False) ]
 
     # DIV and IDIV instructions in x86 are implemented using a loop which
     # issues division microops.  The latency of these microops should really be
@@ -57,7 +57,6 @@ class IntMultDiv(FUDesc):
     # of the quotient.
     if buildEnv['TARGET_ISA'] in ('x86'):
         opList[1].opLat=1
-        opList[1].issueLat=1
 
     count=2
 
@@ -69,8 +68,8 @@ class FP_ALU(FUDesc):
 
 class FP_MultDiv(FUDesc):
     opList = [ OpDesc(opClass='FloatMult', opLat=4),
-               OpDesc(opClass='FloatDiv', opLat=12, issueLat=12),
+               OpDesc(opClass='FloatDiv', opLat=12, pipelined=False),
-               OpDesc(opClass='FloatSqrt', opLat=24, issueLat=24) ]
+               OpDesc(opClass='FloatSqrt', opLat=24, pipelined=False) ]
     count = 2
 
 class SIMD_Unit(FUDesc):
@@ -109,6 +108,6 @@ class RdWrPort(FUDesc):
     count = 4
 
 class IprPort(FUDesc):
-    opList = [ OpDesc(opClass='IprAccess', opLat = 3, issueLat = 3) ]
+    opList = [ OpDesc(opClass='IprAccess', opLat = 3, pipelined = False) ]
     count = 1
 

src/cpu/o3/fu_pool.cc

@@ -89,7 +89,7 @@ FUPool::FUPool(const Params *p)
 
     for (int i = 0; i < Num_OpClasses; ++i) {
         maxOpLatencies[i] = Cycles(0);
-        maxIssueLatencies[i] = Cycles(0);
+        pipelined[i] = true;
     }
 
     //
@@ -123,13 +123,13 @@ FUPool::FUPool(const Params *p)
                     fuPerCapList[(*j)->opClass].addFU(numFU + k);
 
                 // indicate that this FU has the capability
-                fu->addCapability((*j)->opClass, (*j)->opLat, (*j)->issueLat);
+                fu->addCapability((*j)->opClass, (*j)->opLat, (*j)->pipelined);
 
                 if ((*j)->opLat > maxOpLatencies[(*j)->opClass])
                     maxOpLatencies[(*j)->opClass] = (*j)->opLat;
 
-                if ((*j)->issueLat > maxIssueLatencies[(*j)->opClass])
+                if (!(*j)->pipelined)
-                    maxIssueLatencies[(*j)->opClass] = (*j)->issueLat;
+                    pipelined[(*j)->opClass] = false;
             }
 
             numFU++;

src/cpu/o3/fu_pool.hh

@@ -72,8 +72,8 @@ class FUPool : public SimObject
   private:
     /** Maximum op execution latencies, per op class. */
     Cycles maxOpLatencies[Num_OpClasses];
-    /** Maximum issue latencies, per op class. */
+    /** Whether op is pipelined or not. */
-    Cycles maxIssueLatencies[Num_OpClasses];
+    bool pipelined[Num_OpClasses];
 
     /** Bitvector listing capabilities of this FU pool. */
     std::bitset<Num_OpClasses> capabilityList;
@@ -160,8 +160,8 @@ class FUPool : public SimObject
     }
 
     /** Returns the issue latency of the given capability. */
-    Cycles getIssueLatency(OpClass capability) {
+    bool isPipelined(OpClass capability) {
-        return maxIssueLatencies[capability];
+        return pipelined[capability];
     }
 
     /** Have all the FUs drained? */

src/cpu/o3/inst_queue_impl.hh

@@ -825,7 +825,7 @@ InstructionQueue<Impl>::scheduleReadyInsts()
                 if (idx >= 0)
                     fuPool->freeUnitNextCycle(idx);
             } else {
-                Cycles issue_latency = fuPool->getIssueLatency(op_class);
+                bool pipelined = fuPool->isPipelined(op_class);
                 // Generate completion event for the FU
                 ++wbOutstanding;
                 FUCompletion *execution = new FUCompletion(issuing_inst,
@@ -834,8 +834,7 @@ InstructionQueue<Impl>::scheduleReadyInsts()
                 cpu->schedule(execution,
                               cpu->clockEdge(Cycles(op_latency - 1)));
 
-                // @todo: Enforce that issue_latency == 1 or op_latency
+                if (!pipelined) {
-                if (issue_latency > Cycles(1)) {
                     // If FU isn't pipelined, then it must be freed
                     // upon the execution completing.
                     execution->setFreeFU();