diff --git a/src/gpu-compute/compute_unit.cc b/src/gpu-compute/compute_unit.cc
index e485aa6161..3e0b8070fd 100644
--- a/src/gpu-compute/compute_unit.cc
+++ b/src/gpu-compute/compute_unit.cc
@@ -2436,15 +2436,15 @@ ComputeUnit::ComputeUnitStats::ComputeUnitStats(statistics::Group *parent,
     instCyclesLdsPerSimd.init(cu->numVectorALUs);
 
     hitsPerTLBLevel.init(4);
-    execRateDist.init(0, 10, 2);
-    ldsBankConflictDist.init(0, cu->wfSize(), 2);
+    execRateDist.init(0, 10-1, 2);
+    ldsBankConflictDist.init(0, cu->wfSize()-1, 2);
 
     pageDivergenceDist.init(1, cu->wfSize(), 4);
     controlFlowDivergenceDist.init(1, cu->wfSize(), 4);
     activeLanesPerGMemInstrDist.init(1, cu->wfSize(), 4);
     activeLanesPerLMemInstrDist.init(1, cu->wfSize(), 4);
 
-    headTailLatency.init(0, 1000000, 10000).flags(statistics::pdf |
+    headTailLatency.init(0, 1000000-1, 10000).flags(statistics::pdf |
         statistics::oneline);
     waveLevelParallelism.init(0, n_wf * cu->numVectorALUs, 1);
     instInterleave.init(cu->numVectorALUs, 0, 20, 1);
diff --git a/src/gpu-compute/exec_stage.cc b/src/gpu-compute/exec_stage.cc
index bcba938cd8..f2b847c8a7 100644
--- a/src/gpu-compute/exec_stage.cc
+++ b/src/gpu-compute/exec_stage.cc
@@ -216,7 +216,7 @@ ExecStage::ExecStageStats::ExecStageStats(statistics::Group *parent)
     ComputeUnit *compute_unit = static_cast<ComputeUnit*>(parent);
 
     spc.init(0, compute_unit->numExeUnits(), 1);
-    idleDur.init(0, 75, 5);
+    idleDur.init(0, 75-1, 5);
     numCyclesWithInstrTypeIssued.init(compute_unit->numExeUnits());
     numCyclesWithNoInstrTypeIssued.init(compute_unit->numExeUnits());
 
diff --git a/src/gpu-compute/shader.cc b/src/gpu-compute/shader.cc
index 6e3d556026..792ecb6678 100644
--- a/src/gpu-compute/shader.cc
+++ b/src/gpu-compute/shader.cc
@@ -584,31 +584,31 @@ Shader::ShaderStats::ShaderStats(statistics::Group *parent, int wf_size)
                "vector instruction destination operand distribution")
 {
     allLatencyDist
-        .init(0, 1600000, 10000)
+        .init(0, 1600000-1, 10000)
         .flags(statistics::pdf | statistics::oneline);
 
     loadLatencyDist
-        .init(0, 1600000, 10000)
+        .init(0, 1600000-1, 10000)
         .flags(statistics::pdf | statistics::oneline);
 
     storeLatencyDist
-        .init(0, 1600000, 10000)
+        .init(0, 1600000-1, 10000)
         .flags(statistics::pdf | statistics::oneline);
 
     initToCoalesceLatency
-        .init(0, 1600000, 10000)
+        .init(0, 1600000-1, 10000)
         .flags(statistics::pdf | statistics::oneline);
 
     rubyNetworkLatency
-        .init(0, 1600000, 10000)
+        .init(0, 1600000-1, 10000)
         .flags(statistics::pdf | statistics::oneline);
 
     gmEnqueueLatency
-        .init(0, 1600000, 10000)
+        .init(0, 1600000-1, 10000)
         .flags(statistics::pdf | statistics::oneline);
 
     gmToCompleteLatency
-        .init(0, 1600000, 10000)
+        .init(0, 1600000-1, 10000)
         .flags(statistics::pdf | statistics::oneline);
 
     coalsrLineAddresses
@@ -624,7 +624,7 @@ Shader::ShaderStats::ShaderStats(statistics::Group *parent, int wf_size)
         ccprintf(namestr, "%s.cacheBlockRoundTrip%d",
                  static_cast<Shader*>(parent)->name(), idx);
         cacheBlockRoundTrip[idx]
-            .init(0, 1600000, 10000)
+            .init(0, 1600000-1, 10000)
             .name(namestr.str())
             .desc("Coalsr-to-coalsr time for the Nth cache block in an inst")
             .flags(statistics::pdf | statistics::oneline);