inorder: stage width as a python parameter

allow the user to specify how many instructions a pipeline stage can process
on any given cycle (stageWidth...i.e.bandwidth) by setting the parameter through
the python interface rather than compile the code after changing the *.cc file.
(we always had the parameter there, but still used the static 'ThePipeline::StageWidth'
instead)
-
Since StageWidth is now dynamically defined, change the interstage communication
structure to use a vector and get rid of array and array handling index (toNextStageIndex)
since we can just make calls to the list for the same information

7 files changed, 35 insertions, 67 deletions

diff --git a/src/cpu/inorder/InOrderCPU.py b/src/cpu/inorder/InOrderCPU.py
index 8e25891e7..5d24ae4fd 100644
--- a/src/cpu/inorder/InOrderCPU.py
+++ b/src/cpu/inorder/InOrderCPU.py
@@ -40,7 +40,7 @@ class InOrderCPU(BaseCPU):
     threadModel = Param.ThreadModel('SMT', "Multithreading model (SE-MODE only)")
     
     cachePorts = Param.Unsigned(2, "Cache Ports")
-    stageWidth = Param.Unsigned(1, "Stage width")
+    stageWidth = Param.Unsigned(4, "Stage width")
 
     fetchMemPort = Param.String("icache_port" , "Name of Memory Port to get instructions from")
     dataMemPort = Param.String("dcache_port" , "Name of Memory Port to get data from")
diff --git a/src/cpu/inorder/comm.hh b/src/cpu/inorder/comm.hh
index 386046d1c..ba9322079 100644
--- a/src/cpu/inorder/comm.hh
+++ b/src/cpu/inorder/comm.hh
@@ -44,8 +44,7 @@
 /** Struct that defines the information passed from in between stages */
 /** This information mainly goes forward through the pipeline. */
 struct InterStageStruct {
-    int size;
-    ThePipeline::DynInstPtr insts[ThePipeline::StageWidth];
+    std::vector<ThePipeline::DynInstPtr> insts;
     bool squash;
     bool branchMispredict;
     bool branchTaken;
@@ -55,10 +54,10 @@ struct InterStageStruct {
     bool includeSquashInst;
 
     InterStageStruct()
-        :size(0),  squash(false),
-         branchMispredict(false), branchTaken(false),
-         mispredPC(0), nextPC(0),
-         squashedSeqNum(0), includeSquashInst(false)
+        : squash(false),
+          branchMispredict(false), branchTaken(false),
+          mispredPC(0), nextPC(0),
+          squashedSeqNum(0), includeSquashInst(false)
     { }
 
 };
diff --git a/src/cpu/inorder/cpu.cc b/src/cpu/inorder/cpu.cc
index 5a742cb5d..ffdcae7df 100644
--- a/src/cpu/inorder/cpu.cc
+++ b/src/cpu/inorder/cpu.cc
@@ -174,6 +174,7 @@ InOrderCPU::InOrderCPU(Params *params)
       coreType("default"),
       _status(Idle),
       tickEvent(this),
+      stageWidth(params->stageWidth),
       timeBuffer(2 , 2),
       removeInstsThisCycle(false),
       activityRec(params->name, NumStages, 10, params->activity),
diff --git a/src/cpu/inorder/cpu.hh b/src/cpu/inorder/cpu.hh
index c3658373a..9ff0f12ce 100644
--- a/src/cpu/inorder/cpu.hh
+++ b/src/cpu/inorder/cpu.hh
@@ -268,6 +268,9 @@ class InOrderCPU : public BaseCPU
     /** The Pipeline Stages for the CPU */
     PipelineStage *pipelineStage[ThePipeline::NumStages];
 
+    /** Width (processing bandwidth) of each stage */
+    int stageWidth;
+
     /** Program Counters */
     TheISA::PCState pc[ThePipeline::MaxThreads];
 
diff --git a/src/cpu/inorder/pipeline_stage.cc b/src/cpu/inorder/pipeline_stage.cc
index b4b9e497f..e10ceb326 100644
--- a/src/cpu/inorder/pipeline_stage.cc
+++ b/src/cpu/inorder/pipeline_stage.cc
@@ -39,9 +39,9 @@ using namespace std;
 using namespace ThePipeline;
 
 PipelineStage::PipelineStage(Params *params, unsigned stage_num)
-    : stageNum(stage_num), stageWidth(ThePipeline::StageWidth),
+    : stageNum(stage_num), stageWidth(params->stageWidth),
       numThreads(ThePipeline::MaxThreads), _status(Inactive),
-      stageBufferMax(ThePipeline::interStageBuffSize[stage_num]),
+      stageBufferMax(params->stageWidth),
       prevStageValid(false), nextStageValid(false), idle(false)
 {
     switchedOutBuffer.resize(ThePipeline::MaxThreads);
@@ -143,7 +143,6 @@ PipelineStage::setNextStageQueue(TimeBuffer<InterStageStruct> *next_stage_ptr)
 
     // Setup wire to write information to proper place in stage queue.
     nextStage = nextStageQueue->getWire(0);
-    nextStage->size = 0;
     nextStageValid = true;
 }
 
@@ -257,7 +256,7 @@ PipelineStage::removeStalls(ThreadID tid)
 inline bool
 PipelineStage::prevStageInstsValid()
 {
-    return prevStage->size > 0;
+    return prevStage->insts.size() > 0;
 }
 
 bool
@@ -382,7 +381,8 @@ PipelineStage::squashPrevStageInsts(InstSeqNum squash_seq_num, ThreadID tid)
     DPRINTF(InOrderStage, "[tid:%i]: Removing instructions from "
             "incoming stage queue.\n", tid);
 
-    for (int i=0; i < prevStage->size; i++) {
+    int insts_from_prev_stage = prevStage->insts.size();
+    for (int i=0; i < insts_from_prev_stage; i++) {
         if (prevStage->insts[i]->threadNumber == tid &&
             prevStage->insts[i]->seqNum > squash_seq_num) {
             // Change Comment to Annulling previous instruction
@@ -441,16 +441,8 @@ PipelineStage::stageBufferAvail()
         total += skidBuffer[i].size();
     }
 
-    int incoming_insts = (prevStageValid) ?
-        cpu->pipelineStage[stageNum]->prevStage->size :
-        0;
-
     int avail = stageBufferMax - total;
-
-    if (avail < 0)
-        fatal("stageNum %i:stageBufferAvail() < 0..."
-              "stBMax=%i,total=%i,incoming=%i=>%i",
-              stageNum, stageBufferMax, total, incoming_insts, avail);
+    assert(avail >= 0);
 
     return avail;
 }
@@ -462,7 +454,7 @@ PipelineStage::canSendInstToStage(unsigned stage_num)
 
     if (cpu->pipelineStage[stage_num]->prevStageValid) {
         buffer_avail = cpu->pipelineStage[stage_num]->stageBufferAvail() -
-            cpu->pipelineStage[stage_num-1]->nextStage->size >= 1;
+            cpu->pipelineStage[stage_num-1]->nextStage->insts.size() >= 1;
     }
 
     if (!buffer_avail && nextStageQueueValid(stage_num)) {
@@ -576,7 +568,9 @@ void
 PipelineStage::sortInsts()
 {
     if (prevStageValid) {
-        int insts_from_prev_stage = prevStage->size;
+        assert(prevStage->insts.size() <= stageWidth);
+
+        int insts_from_prev_stage = prevStage->insts.size();
         int insts_from_cur_stage = skidSize();
         DPRINTF(InOrderStage, "%i insts available from stage buffer %i. Stage "
                 "currently has %i insts from last cycle.\n",
@@ -591,7 +585,6 @@ PipelineStage::sortInsts()
                         "not inserting into stage buffer.\n",
                     prevStage->insts[i]->readTid(),
                     prevStage->insts[i]->seqNum);
-                prevStage->size--;
                 continue;
             }
 
@@ -619,12 +612,8 @@ PipelineStage::sortInsts()
 
             prevStage->insts[i] = cpu->dummyBufferInst;
 
-            prevStage->size--;
-
             inserted_insts++;
         }
-
-        assert(prevStage->size == 0);
     }
 }
 
@@ -728,11 +717,6 @@ PipelineStage::tick()
     wroteToTimeBuffer = false;
 
     bool status_change = false;
-
-    if (nextStageValid)
-        nextStage->size = 0;
-
-    toNextStageIndex = 0;
     
     sortInsts();
 
@@ -807,7 +791,7 @@ PipelineStage::processStage(bool &status_change)
 
     if (nextStageValid) {
         DPRINTF(InOrderStage, "%i insts now available for stage %i.\n",
-                nextStage->size, stageNum + 1);
+                nextStage->insts.size(), stageNum + 1);
     }
 
     if (instsProcessed > 0) {
@@ -1083,20 +1067,13 @@ PipelineStage::sendInstToNextStage(DynInstPtr inst)
 
             DPRINTF(InOrderStage, "[tid:%u]: [sn:%i]: being placed into  "
                     "index %i of stage buffer %i queue.\n",
-                    tid, inst->seqNum, toNextStageIndex,
+                    tid, inst->seqNum,
+                    cpu->pipelineStage[prev_stage]->nextStage->insts.size(),
                     cpu->pipelineStage[prev_stage]->nextStageQueue->id());
 
-            int next_stage_idx = 
-                cpu->pipelineStage[prev_stage]->nextStage->size;
-
             // Place instructions in inter-stage communication struct for next
             // pipeline stage to read next cycle
-            cpu->pipelineStage[prev_stage]->nextStage->insts[next_stage_idx] 
-                = inst;
-
-            ++(cpu->pipelineStage[prev_stage]->nextStage->size);
-
-            ++toNextStageIndex;
+            cpu->pipelineStage[prev_stage]->nextStage->insts.push_back(inst);
 
             success = true;
 
diff --git a/src/cpu/inorder/pipeline_traits.hh b/src/cpu/inorder/pipeline_traits.hh
index 7abfc9a81..df964e254 100644
--- a/src/cpu/inorder/pipeline_traits.hh
+++ b/src/cpu/inorder/pipeline_traits.hh
@@ -52,7 +52,6 @@ namespace ThePipeline {
     // Pipeline Constants
     const unsigned NumStages = 5;
     const ThreadID MaxThreads = 8;
-    const unsigned StageWidth = 1;
     const unsigned BackEndStartStage = 2;
 
     // List of Resources The Pipeline Uses
@@ -71,19 +70,6 @@ namespace ThePipeline {
        FetchBuff2
     };
 
-    // Expand this as necessary for your inter stage buffer sizes
-    static const unsigned interStageBuffSize[] = {
-        StageWidth, /* Stage 0 - 1 */
-        StageWidth, /* Stage 1 - 2 */
-        StageWidth, /* Stage 2 - 3 */
-        StageWidth, /* Stage 3 - 4 */
-        StageWidth, /* Stage 4 - 5 */
-        StageWidth, /* Stage 5 - 6 */
-        StageWidth, /* Stage 6 - 7 */
-        StageWidth, /* Stage 7 - 8 */
-        StageWidth  /* Stage 8 - 9 */
-    };
-
     typedef InOrderCPUParams Params;
     typedef RefCountingPtr<InOrderDynInst> DynInstPtr;
 
diff --git a/src/cpu/inorder/resource_pool.cc b/src/cpu/inorder/resource_pool.cc
index e8400405a..a0ec910f5 100644
--- a/src/cpu/inorder/resource_pool.cc
+++ b/src/cpu/inorder/resource_pool.cc
@@ -45,46 +45,48 @@ ResourcePool::ResourcePool(InOrderCPU *_cpu, ThePipeline::Params *params)
     //This will help in the auto-generation of this pipeline model.
     //ThePipeline::addResources(resources, memObjects);
 
+    int stage_width = cpu->stageWidth;
+
     // Declare Resource Objects
     // name - id - bandwidth - latency - CPU - Parameters
     // --------------------------------------------------
     resources.push_back(new FetchSeqUnit("Fetch-Seq-Unit", FetchSeq, 
-                                         StageWidth * 2, 0, _cpu, params));
+                                         stage_width * 2, 0, _cpu, params));
 
     memObjects.push_back(ICache);
     resources.push_back(new CacheUnit("icache_port", ICache, 
-                                      StageWidth * MaxThreads, 0, _cpu, 
+                                      stage_width * MaxThreads, 0, _cpu,
                                       params));
 
     resources.push_back(new DecodeUnit("Decode-Unit", Decode, 
-                                       StageWidth, 0, _cpu, params));
+                                       stage_width, 0, _cpu, params));
 
     resources.push_back(new BranchPredictor("Branch-Predictor", BPred, 
-                                            StageWidth, 0, _cpu, params));
+                                            stage_width, 0, _cpu, params));
 
     resources.push_back(new InstBuffer("Fetch-Buffer-T0", FetchBuff, 4, 
                                        0, _cpu, params));
 
     resources.push_back(new UseDefUnit("RegFile-Manager", RegManager, 
-                                       StageWidth * MaxThreads, 0, _cpu, 
+                                       stage_width * MaxThreads, 0, _cpu,
                                        params));
 
     resources.push_back(new AGENUnit("AGEN-Unit", AGEN, 
-                                     StageWidth, 0, _cpu, params));
+                                     stage_width, 0, _cpu, params));
 
     resources.push_back(new ExecutionUnit("Execution-Unit", ExecUnit, 
-                                          StageWidth, 0, _cpu, params));
+                                          stage_width, 0, _cpu, params));
 
     resources.push_back(new MultDivUnit("Mult-Div-Unit", MDU, 5, 0, _cpu, 
                                         params));
 
     memObjects.push_back(DCache);
     resources.push_back(new CacheUnit("dcache_port", DCache, 
-                                      StageWidth * MaxThreads, 0, _cpu, 
+                                      stage_width * MaxThreads, 0, _cpu,
                                       params));
 
     resources.push_back(new GraduationUnit("Graduation-Unit", Grad, 
-                                           StageWidth * MaxThreads, 0, _cpu, 
+                                           stage_width * MaxThreads, 0, _cpu,
                                            params));
 
     resources.push_back(new InstBuffer("Fetch-Buffer-T1", FetchBuff2, 4,