diff --git a/src/cpu/o3/commit_impl.hh b/src/cpu/o3/commit_impl.hh
index 8384dbead..021d3ef90 100644
--- a/src/cpu/o3/commit_impl.hh
+++ b/src/cpu/o3/commit_impl.hh
@@ -204,19 +204,6 @@ DefaultCommit<Impl>::regStats()
         .flags(total)
         ;
 
-    //
-    //  Commit-Eligible instructions...
-    //
-    //  -> The number of instructions eligible to commit in those
-    //  cycles where we reached our commit BW limit (less the number
-    //  actually committed)
-    //
-    //  -> The average value is computed over ALL CYCLES... not just
-    //  the BW limited cycles
-    //
-    //  -> The standard deviation is computed only over cycles where
-    //  we reached the BW limit
-    //
     commitEligible
         .init(cpu->number_of_threads)
         .name(name() + ".COM:bw_limited")
diff --git a/src/cpu/o3/fetch.hh b/src/cpu/o3/fetch.hh
index 76b32de68..962d46437 100644
--- a/src/cpu/o3/fetch.hh
+++ b/src/cpu/o3/fetch.hh
@@ -421,6 +421,7 @@ class DefaultFetch
     Stats::Scalar<> icacheStallCycles;
     /** Stat for total number of fetched instructions. */
     Stats::Scalar<> fetchedInsts;
+    /** Total number of fetched branches. */
     Stats::Scalar<> fetchedBranches;
     /** Stat for total number of predicted branches. */
     Stats::Scalar<> predictedBranches;
diff --git a/src/cpu/o3/fetch_impl.hh b/src/cpu/o3/fetch_impl.hh
index af2aadf09..477a1469c 100644
--- a/src/cpu/o3/fetch_impl.hh
+++ b/src/cpu/o3/fetch_impl.hh
@@ -915,7 +915,11 @@ DefaultFetch<Impl>::fetch(bool &status_change)
 
         bool fetch_success = fetchCacheLine(fetch_PC, fault, tid);
         if (!fetch_success) {
-            ++fetchMiscStallCycles;
+            if (cacheBlocked) {
+                ++icacheStallCycles;
+            } else {
+                ++fetchMiscStallCycles;
+            }
             return;
         }
     } else {
diff --git a/src/cpu/o3/iew.hh b/src/cpu/o3/iew.hh
index 455de7c3f..615022dc9 100644
--- a/src/cpu/o3/iew.hh
+++ b/src/cpu/o3/iew.hh
@@ -437,14 +437,6 @@ class DefaultIEW
     Stats::Scalar<> iewIQFullEvents;
     /** Stat for number of times the LSQ becomes full. */
     Stats::Scalar<> iewLSQFullEvents;
-    /** Stat for total number of executed instructions. */
-    Stats::Scalar<> iewExecutedInsts;
-    /** Stat for total number of executed load instructions. */
-    Stats::Vector<> iewExecLoadInsts;
-    /** Stat for total number of executed store instructions. */
-//    Stats::Scalar<> iewExecStoreInsts;
-    /** Stat for total number of squashed instructions skipped at execute. */
-    Stats::Scalar<> iewExecSquashedInsts;
     /** Stat for total number of memory ordering violation events. */
     Stats::Scalar<> memOrderViolationEvents;
     /** Stat for total number of incorrect predicted taken branches. */
@@ -454,28 +446,25 @@ class DefaultIEW
     /** Stat for total number of mispredicted branches detected at execute. */
     Stats::Formula branchMispredicts;
 
+    /** Stat for total number of executed instructions. */
+    Stats::Scalar<> iewExecutedInsts;
+    /** Stat for total number of executed load instructions. */
+    Stats::Vector<> iewExecLoadInsts;
+    /** Stat for total number of squashed instructions skipped at execute. */
+    Stats::Scalar<> iewExecSquashedInsts;
     /** Number of executed software prefetches. */
-    Stats::Vector<> exeSwp;
+    Stats::Vector<> iewExecutedSwp;
     /** Number of executed nops. */
-    Stats::Vector<> exeNop;
+    Stats::Vector<> iewExecutedNop;
     /** Number of executed meomory references. */
-    Stats::Vector<> exeRefs;
+    Stats::Vector<> iewExecutedRefs;
     /** Number of executed branches. */
-    Stats::Vector<> exeBranches;
-
-//    Stats::Vector<> issued_ops;
-/*
-    Stats::Vector<> stat_fu_busy;
-    Stats::Vector2d<> stat_fuBusy;
-    Stats::Vector<> dist_unissued;
-    Stats::Vector2d<> stat_issued_inst_type;
-*/
-    /** Number of instructions issued per cycle. */
-    Stats::Formula issueRate;
+    Stats::Vector<> iewExecutedBranches;
     /** Number of executed store instructions. */
     Stats::Formula iewExecStoreInsts;
-//    Stats::Formula issue_op_rate;
-//    Stats::Formula fu_busy_rate;
+    /** Number of instructions executed per cycle. */
+    Stats::Formula iewExecRate;
+
     /** Number of instructions sent to commit. */
     Stats::Vector<> iewInstsToCommit;
     /** Number of instructions that writeback. */
@@ -488,7 +477,6 @@ class DefaultIEW
      * to resource contention.
      */
     Stats::Vector<> wbPenalized;
-
     /** Number of instructions per cycle written back. */
     Stats::Formula wbRate;
     /** Average number of woken instructions per writeback. */
diff --git a/src/cpu/o3/iew_impl.hh b/src/cpu/o3/iew_impl.hh
index 0649f10ec..b02ee8555 100644
--- a/src/cpu/o3/iew_impl.hh
+++ b/src/cpu/o3/iew_impl.hh
@@ -93,6 +93,7 @@ DefaultIEW<Impl>::regStats()
     using namespace Stats;
 
     instQueue.regStats();
+    ldstQueue.regStats();
 
     iewIdleCycles
         .name(name() + ".iewIdleCycles")
@@ -138,20 +139,6 @@ DefaultIEW<Impl>::regStats()
         .name(name() + ".iewLSQFullEvents")
         .desc("Number of times the LSQ has become full, causing a stall");
 
-    iewExecutedInsts
-        .name(name() + ".iewExecutedInsts")
-        .desc("Number of executed instructions");
-
-    iewExecLoadInsts
-        .init(cpu->number_of_threads)
-        .name(name() + ".iewExecLoadInsts")
-        .desc("Number of load instructions executed")
-        .flags(total);
-
-    iewExecSquashedInsts
-        .name(name() + ".iewExecSquashedInsts")
-        .desc("Number of squashed instructions skipped in execute");
-
     memOrderViolationEvents
         .name(name() + ".memOrderViolationEvents")
         .desc("Number of memory order violations");
@@ -170,114 +157,105 @@ DefaultIEW<Impl>::regStats()
 
     branchMispredicts = predictedTakenIncorrect + predictedNotTakenIncorrect;
 
-    exeSwp
+    iewExecutedInsts
+        .name(name() + ".EXEC:insts")
+        .desc("Number of executed instructions");
+
+    iewExecLoadInsts
+        .init(cpu->number_of_threads)
+        .name(name() + ".EXEC:loads")
+        .desc("Number of load instructions executed")
+        .flags(total);
+
+    iewExecSquashedInsts
+        .name(name() + ".EXEC:squashedInsts")
+        .desc("Number of squashed instructions skipped in execute");
+
+    iewExecutedSwp
         .init(cpu->number_of_threads)
         .name(name() + ".EXEC:swp")
         .desc("number of swp insts executed")
-        .flags(total)
-        ;
+        .flags(total);
 
-    exeNop
+    iewExecutedNop
         .init(cpu->number_of_threads)
         .name(name() + ".EXEC:nop")
         .desc("number of nop insts executed")
-        .flags(total)
-        ;
+        .flags(total);
 
-    exeRefs
+    iewExecutedRefs
         .init(cpu->number_of_threads)
         .name(name() + ".EXEC:refs")
         .desc("number of memory reference insts executed")
-        .flags(total)
-        ;
+        .flags(total);
 
-    exeBranches
+    iewExecutedBranches
         .init(cpu->number_of_threads)
         .name(name() + ".EXEC:branches")
         .desc("Number of branches executed")
-        .flags(total)
-        ;
-
-    issueRate
-        .name(name() + ".EXEC:rate")
-        .desc("Inst execution rate")
-        .flags(total)
-        ;
-    issueRate = iewExecutedInsts / cpu->numCycles;
+        .flags(total);
 
     iewExecStoreInsts
         .name(name() + ".EXEC:stores")
         .desc("Number of stores executed")
-        .flags(total)
-        ;
-    iewExecStoreInsts = exeRefs - iewExecLoadInsts;
-/*
-    for (int i=0; i<Num_OpClasses; ++i) {
-        stringstream subname;
-        subname << opClassStrings[i] << "_delay";
-        issue_delay_dist.subname(i, subname.str());
-    }
-*/
-    //
-    //  Other stats
-    //
+        .flags(total);
+    iewExecStoreInsts = iewExecutedRefs - iewExecLoadInsts;
+
+    iewExecRate
+        .name(name() + ".EXEC:rate")
+        .desc("Inst execution rate")
+        .flags(total);
+
+    iewExecRate = iewExecutedInsts / cpu->numCycles;
 
     iewInstsToCommit
         .init(cpu->number_of_threads)
         .name(name() + ".WB:sent")
         .desc("cumulative count of insts sent to commit")
-        .flags(total)
-        ;
+        .flags(total);
 
     writebackCount
         .init(cpu->number_of_threads)
         .name(name() + ".WB:count")
         .desc("cumulative count of insts written-back")
-        .flags(total)
-        ;
+        .flags(total);
 
     producerInst
         .init(cpu->number_of_threads)
         .name(name() + ".WB:producers")
         .desc("num instructions producing a value")
-        .flags(total)
-        ;
+        .flags(total);
 
     consumerInst
         .init(cpu->number_of_threads)
         .name(name() + ".WB:consumers")
         .desc("num instructions consuming a value")
-        .flags(total)
-        ;
+        .flags(total);
 
     wbPenalized
         .init(cpu->number_of_threads)
         .name(name() + ".WB:penalized")
         .desc("number of instrctions required to write to 'other' IQ")
-        .flags(total)
-        ;
+        .flags(total);
 
     wbPenalizedRate
         .name(name() + ".WB:penalized_rate")
         .desc ("fraction of instructions written-back that wrote to 'other' IQ")
-        .flags(total)
-        ;
+        .flags(total);
 
     wbPenalizedRate = wbPenalized / writebackCount;
 
     wbFanout
         .name(name() + ".WB:fanout")
         .desc("average fanout of values written-back")
-        .flags(total)
-        ;
+        .flags(total);
 
     wbFanout = producerInst / consumerInst;
 
     wbRate
         .name(name() + ".WB:rate")
         .desc("insts written-back per cycle")
-        .flags(total)
-        ;
+        .flags(total);
     wbRate = writebackCount / cpu->numCycles;
 }
 
@@ -1098,7 +1076,7 @@ DefaultIEW<Impl>::dispatchInsts(unsigned tid)
 
             instQueue.recordProducer(inst);
 
-            exeNop[tid]++;
+            iewExecutedNop[tid]++;
 
             add_to_iq = false;
         } else if (inst->isExecuted()) {
@@ -1509,9 +1487,9 @@ DefaultIEW<Impl>::updateExeInstStats(DynInstPtr &inst)
     //
 #ifdef TARGET_ALPHA
     if (inst->isDataPrefetch())
-        exeSwp[thread_number]++;
+        iewExecutedSwp[thread_number]++;
     else
-        iewExecutedInsts++;
+        iewIewExecutedcutedInsts++;
 #else
     iewExecutedInsts++;
 #endif
@@ -1520,13 +1498,13 @@ DefaultIEW<Impl>::updateExeInstStats(DynInstPtr &inst)
     //  Control operations
     //
     if (inst->isControl())
-        exeBranches[thread_number]++;
+        iewExecutedBranches[thread_number]++;
 
     //
     //  Memory operations
     //
     if (inst->isMemRef()) {
-        exeRefs[thread_number]++;
+        iewExecutedRefs[thread_number]++;
 
         if (inst->isLoad()) {
             iewExecLoadInsts[thread_number]++;
diff --git a/src/cpu/o3/inst_queue.hh b/src/cpu/o3/inst_queue.hh
index 60a713020..6fd3c6d0b 100644
--- a/src/cpu/o3/inst_queue.hh
+++ b/src/cpu/o3/inst_queue.hh
@@ -474,12 +474,17 @@ class InstructionQueue
     /** Stat for number of non-speculative instructions removed due to a squash.
      */
     Stats::Scalar<> iqSquashedNonSpecRemoved;
+    // Also include number of instructions rescheduled and replayed.
 
-    /** Distribution of number of instructions in the queue. */
+    /** Distribution of number of instructions in the queue.
+     * @todo: Need to create struct to track the entry time for each
+     * instruction. */
     Stats::VectorDistribution<> queueResDist;
     /** Distribution of the number of instructions issued. */
     Stats::Distribution<> numIssuedDist;
-    /** Distribution of the cycles it takes to issue an instruction. */
+    /** Distribution of the cycles it takes to issue an instruction.
+     * @todo: Need to create struct to track the ready time for each
+     * instruction. */
     Stats::VectorDistribution<> issueDelayDist;
 
     /** Number of times an instruction could not be issued because a
@@ -492,8 +497,7 @@ class InstructionQueue
 
     /** Number of instructions issued per cycle. */
     Stats::Formula issueRate;
-//    Stats::Formula issue_stores;
-//    Stats::Formula issue_op_rate;
+
     /** Number of times the FU was busy. */
     Stats::Vector<> fuBusy;
     /** Number of times the FU was busy per instruction issued. */
diff --git a/src/cpu/o3/inst_queue_impl.hh b/src/cpu/o3/inst_queue_impl.hh
index 06a052c6f..66d4a54c6 100644
--- a/src/cpu/o3/inst_queue_impl.hh
+++ b/src/cpu/o3/inst_queue_impl.hh
@@ -289,22 +289,7 @@ InstructionQueue<Impl>::regStats()
         .flags(total)
         ;
     issueRate = iqInstsIssued / cpu->numCycles;
-/*
-    issue_stores
-        .name(name() + ".ISSUE:stores")
-        .desc("Number of stores issued")
-        .flags(total)
-        ;
-    issue_stores = exe_refs - exe_loads;
-*/
-/*
-    issue_op_rate
-        .name(name() + ".ISSUE:op_rate")
-        .desc("Operation issue rate")
-        .flags(total)
-        ;
-    issue_op_rate = issued_ops / numCycles;
-*/
+
     statFuBusy
         .init(Num_OpClasses)
         .name(name() + ".ISSUE:fu_full")
diff --git a/src/cpu/o3/lsq.hh b/src/cpu/o3/lsq.hh
index bc4154c85..1dbd46b8e 100644
--- a/src/cpu/o3/lsq.hh
+++ b/src/cpu/o3/lsq.hh
@@ -62,6 +62,9 @@ class LSQ {
     /** Returns the name of the LSQ. */
     std::string name() const;
 
+    /** Registers statistics of each LSQ unit. */
+    void regStats();
+
     /** Sets the pointer to the list of active threads. */
     void setActiveThreads(std::list<unsigned> *at_ptr);
     /** Sets the CPU pointer. */
diff --git a/src/cpu/o3/lsq_impl.hh b/src/cpu/o3/lsq_impl.hh
index 27aa0dc3c..0b6c6f542 100644
--- a/src/cpu/o3/lsq_impl.hh
+++ b/src/cpu/o3/lsq_impl.hh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2004-2006 The Regents of The University of Michigan
+ * Copyright (c) 2005-2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -106,6 +106,16 @@ LSQ<Impl>::name() const
     return iewStage->name() + ".lsq";
 }
 
+template<class Impl>
+void
+LSQ<Impl>::regStats()
+{
+    //Initialize LSQs
+    for (int tid=0; tid < numThreads; tid++) {
+        thread[tid].regStats();
+    }
+}
+
 template<class Impl>
 void
 LSQ<Impl>::setActiveThreads(list<unsigned> *at_ptr)
diff --git a/src/cpu/o3/lsq_unit.hh b/src/cpu/o3/lsq_unit.hh
index ce0cdd36f..3de581519 100644
--- a/src/cpu/o3/lsq_unit.hh
+++ b/src/cpu/o3/lsq_unit.hh
@@ -77,6 +77,9 @@ class LSQUnit {
     /** Returns the name of the LSQ unit. */
     std::string name() const;
 
+    /** Registers statistics. */
+    void regStats();
+
     /** Sets the CPU pointer. */
     void setCPU(FullCPU *cpu_ptr);
 
@@ -127,9 +130,6 @@ class LSQUnit {
 
     void completeDataAccess(PacketPtr pkt);
 
-    // @todo: Include stats in the LSQ unit.
-    //void regStats();
-
     /** Clears all the entries in the LQ. */
     void clearLQ();
 
@@ -443,25 +443,35 @@ class LSQUnit {
     // Will also need how many read/write ports the Dcache has.  Or keep track
     // of that in stage that is one level up, and only call executeLoad/Store
     // the appropriate number of times.
-/*
-    // total number of loads forwaded from LSQ stores
-    Stats::Vector<> lsq_forw_loads;
 
-    // total number of loads ignored due to invalid addresses
-    Stats::Vector<> inv_addr_loads;
+    /** Total number of loads forwaded from LSQ stores. */
+    Stats::Scalar<> lsqForwLoads;
 
-    // total number of software prefetches ignored due to invalid addresses
-    Stats::Vector<> inv_addr_swpfs;
+    /** Total number of loads ignored due to invalid addresses. */
+    Stats::Scalar<> invAddrLoads;
 
-    // total non-speculative bogus addresses seen (debug var)
-    Counter sim_invalid_addrs;
-    Stats::Vector<> fu_busy;  //cumulative fu busy
+    /** Total number of squashed loads. */
+    Stats::Scalar<> lsqSquashedLoads;
 
-    // ready loads blocked due to memory disambiguation
-    Stats::Vector<> lsq_blocked_loads;
+    /** Total number of responses from the memory system that are
+     * ignored due to the instruction already being squashed. */
+    Stats::Scalar<> lsqIgnoredResponses;
+
+    /** Total number of squashed stores. */
+    Stats::Scalar<> lsqSquashedStores;
+
+    /** Total number of software prefetches ignored due to invalid addresses. */
+    Stats::Scalar<> invAddrSwpfs;
+
+    /** Ready loads blocked due to partial store-forwarding. */
+    Stats::Scalar<> lsqBlockedLoads;
+
+    /** Number of loads that were rescheduled. */
+    Stats::Scalar<> lsqRescheduledLoads;
+
+    /** Number of times the LSQ is blocked due to the cache. */
+    Stats::Scalar<> lsqCacheBlocked;
 
-    Stats::Scalar<> lsqInversion;
-*/
   public:
     /** Executes the load at the given index. */
     template <class T>
@@ -519,6 +529,7 @@ LSQUnit<Impl>::read(Request *req, T &data, int load_idx)
     if (req->getFlags() & UNCACHEABLE &&
         (load_idx != loadHead || !load_inst->reachedCommit)) {
         iewStage->rescheduleMemInst(load_inst);
+        ++lsqRescheduledLoads;
         return TheISA::genMachineCheckFault();
     }
 
@@ -598,7 +609,7 @@ LSQUnit<Impl>::read(Request *req, T &data, int load_idx)
             // @todo: Need to make this a parameter.
             wb->schedule(curTick);
 
-            // Should keep track of stat for forwarded data
+            ++lsqForwLoads;
             return NoFault;
         } else if ((store_has_lower_limit && lower_load_has_store_part) ||
                    (store_has_upper_limit && upper_load_has_store_part) ||
@@ -626,6 +637,7 @@ LSQUnit<Impl>::read(Request *req, T &data, int load_idx)
             // Tell IQ/mem dep unit that this instruction will need to be
             // rescheduled eventually
             iewStage->rescheduleMemInst(load_inst);
+            ++lsqRescheduledLoads;
 
             // Do not generate a writeback event as this instruction is not
             // complete.
@@ -633,6 +645,7 @@ LSQUnit<Impl>::read(Request *req, T &data, int load_idx)
                     "Store idx %i to load addr %#x\n",
                     store_idx, req->getVaddr());
 
+            ++lsqBlockedLoads;
             return NoFault;
         }
     }
@@ -660,6 +673,7 @@ LSQUnit<Impl>::read(Request *req, T &data, int load_idx)
 
     // if we have a cache, do cache access too
     if (!dcachePort->sendTiming(data_pkt)) {
+        ++lsqCacheBlocked;
         // There's an older load that's already going to squash.
         if (isLoadBlocked && blockedLoadSeqNum < load_inst->seqNum)
             return NoFault;
diff --git a/src/cpu/o3/lsq_unit_impl.hh b/src/cpu/o3/lsq_unit_impl.hh
index 4c01f29a2..a5c1eb12a 100644
--- a/src/cpu/o3/lsq_unit_impl.hh
+++ b/src/cpu/o3/lsq_unit_impl.hh
@@ -196,6 +196,47 @@ LSQUnit<Impl>::name() const
     }
 }
 
+template<class Impl>
+void
+LSQUnit<Impl>::regStats()
+{
+    lsqForwLoads
+        .name(name() + ".forwLoads")
+        .desc("Number of loads that had data forwarded from stores");
+
+    invAddrLoads
+        .name(name() + ".invAddrLoads")
+        .desc("Number of loads ignored due to an invalid address");
+
+    lsqSquashedLoads
+        .name(name() + ".squashedLoads")
+        .desc("Number of loads squashed");
+
+    lsqIgnoredResponses
+        .name(name() + ".ignoredResponses")
+        .desc("Number of memory responses ignored because the instruction is squashed");
+
+    lsqSquashedStores
+        .name(name() + ".squashedStores")
+        .desc("Number of stores squashed");
+
+    invAddrSwpfs
+        .name(name() + ".invAddrSwpfs")
+        .desc("Number of software prefetches ignored due to an invalid address");
+
+    lsqBlockedLoads
+        .name(name() + ".blockedLoads")
+        .desc("Number of blocked loads due to partial load-store forwarding");
+
+    lsqRescheduledLoads
+        .name(name() + ".rescheduledLoads")
+        .desc("Number of loads that were rescheduled");
+
+    lsqCacheBlocked
+        .name(name() + ".cacheBlocked")
+        .desc("Number of times an access to memory failed due to the cache being blocked");
+}
+
 template<class Impl>
 void
 LSQUnit<Impl>::clearLQ()
@@ -618,7 +659,7 @@ LSQUnit<Impl>::writebackStores()
         if (!dcachePort->sendTiming(data_pkt)) {
             // Need to handle becoming blocked on a store.
             isStoreBlocked = true;
-
+            ++lsqCacheBlocked;
             assert(retryPkt == NULL);
             retryPkt = data_pkt;
         } else {
@@ -677,6 +718,7 @@ LSQUnit<Impl>::squash(const InstSeqNum &squashed_num)
         loadTail = load_idx;
 
         decrLdIdx(load_idx);
+        ++lsqSquashedLoads;
     }
 
     if (isLoadBlocked) {
@@ -723,6 +765,7 @@ LSQUnit<Impl>::squash(const InstSeqNum &squashed_num)
         storeTail = store_idx;
 
         decrStIdx(store_idx);
+        ++lsqSquashedStores;
     }
 }
 
@@ -782,6 +825,7 @@ LSQUnit<Impl>::writeback(DynInstPtr &inst, PacketPtr pkt)
     // Squashed instructions do not need to complete their access.
     if (inst->isSquashed()) {
         assert(!inst->isStore());
+        ++lsqIgnoredResponses;
         return;
     }
 
@@ -858,6 +902,7 @@ LSQUnit<Impl>::recvRetry()
             isStoreBlocked = false;
         } else {
             // Still blocked!
+            ++lsqCacheBlocked;
         }
     } else if (isLoadBlocked) {
         DPRINTF(LSQUnit, "Loads squash themselves and all younger insts, "