TPC QC: Add GPUErrorQA task

Detectors/TPC/qc/CMakeLists.txt

@@ -19,6 +19,7 @@ o2_add_library(TPCQC
                        src/SACs.cxx
                        src/IDCsVsSACs.cxx
                        src/TrackClusters.cxx
+		       src/GPUErrorQA.cxx
                PUBLIC_LINK_LIBRARIES O2::TPCBase
                                      O2::DataFormatsTPC
                                      O2::GPUO2Interface
@@ -36,7 +37,8 @@ o2_target_root_dictionary(TPCQC
                                   include/TPCQC/DCSPTemperature.h
                                   include/TPCQC/SACs.h
                                   include/TPCQC/IDCsVsSACs.h
-                                  include/TPCQC/TrackClusters.h)
+                                  include/TPCQC/TrackClusters.h
+				  include/TPCQC/GPUErrorQA.h)
 
 o2_add_test(PID
             COMPONENT_NAME tpc

Detectors/TPC/qc/include/TPCQC/GPUErrorQA.h

@@ -0,0 +1,69 @@
+// Copyright 2019-2020 CERN and copyright holders of ALICE O2.
+// See https://alice-o2.web.cern.ch/copyright for details of the copyright holders.
+// All rights not expressly granted are reserved.
+//
+// This software is distributed under the terms of the GNU General Public
+// License v3 (GPL Version 3), copied verbatim in the file "COPYING".
+//
+// In applying this license CERN does not waive the privileges and immunities
+// granted to it by virtue of its status as an Intergovernmental Organization
+// or submit itself to any jurisdiction.
+
+///
+/// @file   GPUErrorQA.h
+/// @author Anton Riedel, anton.riedel@cern.ch
+///
+
+#ifndef AliceO2_TPC_QC_GPUERRORQA_H
+#define AliceO2_TPC_QC_GPUERRORQA_H
+
+#include <memory>
+#include <string>
+#include <vector>
+#include <unordered_map>
+
+// root includes
+
+// o2 includes
+// #include "DataFormatsTPC/Defs.h"
+
+class TH1;
+namespace o2::tpc::qc
+{
+
+/// @brief TPC QC task for errors from GPU reconstruction
+///
+/// This class is used to retrieve and visualize GPU errors
+/// according to corresponding error code and location.
+///
+/// origin: TPC
+/// @author Anton Riedel, anton.riedel@cern.ch
+class GPUErrorQA
+{
+ public:
+  /// \brief Constructor.
+  GPUErrorQA() = default;
+
+  /// process gpu error reported by the reconstruction workflow
+  void processErrors(std::vector<std::array<uint32_t, 4>> errors);
+
+  /// Initialize all histograms
+  void initializeHistograms();
+
+  /// Reset all histograms
+  void resetHistograms();
+
+  /// return histograms
+  const std::unordered_map<std::string, std::unique_ptr<TH1>>& getMapHist() const { return mMapHist; }
+
+  /// Dump results to a file
+  void dumpToFile(std::string filename);
+
+ private:
+  std::unordered_map<std::string, std::unique_ptr<TH1>> mMapHist;
+
+  ClassDefNV(GPUErrorQA, 2);
+};
+} // namespace o2::tpc::qc
+
+#endif // AliceO2_TPC_QC_GPUERRORQA_H

Detectors/TPC/qc/src/GPUErrorQA.cxx

@@ -0,0 +1,80 @@
+// Copyright 2019-2025 CERN and copyright holders of ALICE O2.
+// See https://alice-o2.web.cern.ch/copyright for details of the copyright holders.
+// All rights not expressly granted are reserved.
+//
+// This software is distributed under the terms of the GNU General Public
+// License v3 (GPL Version 3), copied verbatim in the file "COPYING".
+//
+// In applying this license CERN does not waive the privileges and immunities
+// granted to it by virtue of its status as an Intergovernmental Organization
+// or submit itself to any jurisdiction.
+
+#define _USE_MATH_DEFINES
+
+// root includes
+#include "TFile.h"
+#include "TH1I.h"
+
+// o2 includes
+#include "TPCQC/GPUErrorQA.h"
+#include "GPUErrors.h"
+
+ClassImp(o2::tpc::qc::GPUErrorQA);
+
+using namespace o2::tpc::qc;
+
+//______________________________________________________________________________
+void GPUErrorQA::initializeHistograms()
+{
+  TH1::AddDirectory(false);
+
+  auto const& errorNames = o2::gpu::GPUErrors::getErrorNames();
+
+  int maxErrorCode = 1;
+  for (const auto& [key, _] : errorNames) {
+    if (static_cast<int>(key) > maxErrorCode) {
+      maxErrorCode = key;
+    }
+  }
+
+  // 1D histogram counting all reported errors
+  mMapHist["ErrorCounter"] = std::make_unique<TH1I>("ErrorCounter", "ErrorCounter", maxErrorCode, -0.5, maxErrorCode - 0.5);
+  mMapHist["ErrorCounter"]->GetXaxis()->SetTitle("Error Codes");
+  mMapHist["ErrorCounter"]->GetYaxis()->SetTitle("Entries");
+  // for convienence, label each bin with the error name
+  for (size_t bin = 1; bin <= maxErrorCode; bin++) {
+    auto const& it = errorNames.find(bin);
+    if (it != errorNames.end()) {
+      mMapHist["ErrorCounter"]->GetXaxis()->SetBinLabel(bin, it->second);
+    } else {
+      mMapHist["ErrorCounter"]->GetXaxis()->SetBinLabel(bin, "NO_DEF");
+    }
+  }
+}
+//______________________________________________________________________________
+void GPUErrorQA::resetHistograms()
+{
+  for (const auto& pair : mMapHist) {
+    pair.second->Reset();
+  }
+}
+//______________________________________________________________________________
+void GPUErrorQA::processErrors(std::vector<std::array<uint32_t, 4>> errors)
+{
+  for (const auto& error : errors) {
+    uint32_t errorCode = error[0];
+    mMapHist["ErrorCounter"]->AddBinContent(errorCode);
+  }
+}
+
+//______________________________________________________________________________
+void GPUErrorQA::dumpToFile(const std::string filename)
+{
+  auto f = std::unique_ptr<TFile>(TFile::Open(filename.data(), "recreate"));
+  TObjArray arr;
+  arr.SetName("GPUErrorQA_Hists");
+  for ([[maybe_unused]] const auto& [name, hist] : mMapHist) {
+    arr.Add(hist.get());
+  }
+  arr.Write(arr.GetName(), TObject::kSingleKey);
+}

Detectors/TPC/qc/src/TPCQCLinkDef.h

@@ -24,6 +24,7 @@
 #pragma link C++ class o2::tpc::qc::SACs + ;
 #pragma link C++ class o2::tpc::qc::IDCsVsSACs + ;
 #pragma link C++ class o2::tpc::qc::TrackClusters + ;
+#pragma link C++ class o2::tpc::qc::GPUErrorQA + ;
 #pragma link C++ function o2::tpc::qc::helpers::makeLogBinning + ;
 #pragma link C++ function o2::tpc::qc::helpers::setStyleHistogram1D + ;
 #pragma link C++ function o2::tpc::qc::helpers::setStyleHistogram2D + ;

GPU/GPUTracking/Global/GPUErrors.cxx

@@ -48,15 +48,20 @@ void GPUErrors::clear()
   memset(mErrors, 0, GPUCA_MAX_ERRORS * sizeof(*mErrors));
 }
 
-static std::unordered_map<uint32_t, const char*> errorNames = {
+const std::unordered_map<uint32_t, const char*>& GPUErrors::getErrorNames()
+{
+  static std::unordered_map<uint32_t, const char*> errorNames = {
 #define GPUCA_ERROR_CODE(num, name, ...) {num, GPUCA_M_STR(name)},
 #include "GPUErrorCodes.h"
 #undef GPUCA_ERROR_CODE
-};
+  };
+  return errorNames;
+}
 
 bool GPUErrors::printErrors(bool silent, uint64_t mask)
 {
   bool retVal = 0;
+  const auto& errorNames = getErrorNames();
   for (uint32_t i = 0; i < std::min(*mErrors, GPUCA_MAX_ERRORS); i++) {
     uint32_t errorCode = mErrors[4 * i + 1];
     const auto& it = errorNames.find(errorCode);

GPU/GPUTracking/Global/GPUErrors.h

@@ -16,6 +16,9 @@
 #define GPUERRORS_H
 
 #include "GPUCommonDef.h"
+#ifndef GPUCA_GPUCODE
+#include <unordered_map>
+#endif
 
 namespace o2::gpu
 {
@@ -34,6 +37,9 @@ class GPUErrors
   void setMemory(GPUglobalref() uint32_t* m) { mErrors = m; }
   void clear();
   bool printErrors(bool silent = false, uint64_t mask = 0);
+#ifndef GPUCA_GPUCODE
+  static const std::unordered_map<uint32_t, const char*>& getErrorNames();
+#endif
   uint32_t getNErrors() const;
   const uint32_t* getErrorPtr() const;
   static uint32_t getMaxErrors();