Merge remote-tracking branch 'upstream/master' into patchMadsh-upstream

epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/process_function_definitions.inc

@@ -251,12 +251,12 @@ namespace mg5amcCpu
                        fptype* allDenominators,    // output: multichannel denominators[nevt], running_sum_over_helicities
 #endif
                        bool* isGoodHel )           // output: isGoodHel[ncomb] - device array (CUDA implementation)
-  { /* clang-format on */
-    fptype allMEsLast = 0;
+  {                                                         /* clang-format on */
     const int ievt = blockDim.x * blockIdx.x + threadIdx.x; // index of event (thread) in grid
-    allMEs[ievt] = 0;
     for( int ihel = 0; ihel < ncomb; ihel++ )
     {
+      // NEW IMPLEMENTATION OF GETGOODHEL (#630): RESET THE RUNNING SUM OVER HELICITIES TO 0 BEFORE ADDING A NEW HELICITY
+      allMEs[ievt] = 0;
       // NB: calculate_wavefunctions ADDS |M|^2 for a given ihel to the running sum of |M|^2 over helicities for the given event(s)
       constexpr fptype_sv* jamp2_sv = nullptr; // no need for color selection during helicity filtering
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
@@ -265,12 +265,11 @@ namespace mg5amcCpu
 #else
       calculate_wavefunctions( ihel, allmomenta, allcouplings, allMEs, jamp2_sv );
 #endif
-      if( allMEs[ievt] != allMEsLast )
+      if( allMEs[ievt] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
       {
         //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
         isGoodHel[ihel] = true;
       }
-      allMEsLast = allMEs[ievt]; // running sum up to helicity ihel for event ievt
     }
   }
 #else
@@ -289,19 +288,11 @@ namespace mg5amcCpu
     //assert( (size_t)(allMEs) %% mgOnGpu::cppAlign == 0 ); // SANITY CHECK: require SIMD-friendly alignment [COMMENT OUT TO TEST MISALIGNED ACCESS]
     // Allocate arrays at build time to contain at least 16 events (or at least neppV events if neppV>16, e.g. in future VPUs)
     constexpr int maxtry0 = std::max( 16, neppV ); // 16, but at least neppV (otherwise the npagV loop does not even start)
-    fptype allMEsLast[maxtry0] = { 0 };            // allocated at build time: maxtry0 must be a constexpr
     // Loop over only nevt events if nevt is < 16 (note that nevt is always >= neppV)
     assert( nevt >= neppV );
     const int maxtry = std::min( maxtry0, nevt ); // 16, but at most nevt (avoid invalid memory access if nevt<maxtry0)
 
-    // PART 0 - INITIALISATION (before calculate_wavefunctions)
-    // Reset the "matrix elements" - running sums of |M|^2 over helicities for the given event
-    for( int ievt = 0; ievt < maxtry; ++ievt )
-    {
-      allMEs[ievt] = 0; // all zeros
-    }
-
-    // PART 1 - HELICITY LOOP: CALCULATE WAVEFUNCTIONS
+    // HELICITY LOOP: CALCULATE WAVEFUNCTIONS
     const int npagV = maxtry / neppV;
 #if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
     // Mixed fptypes #537: float for color algebra and double elsewhere
@@ -320,6 +311,16 @@ namespace mg5amcCpu
 #endif
       for( int ihel = 0; ihel < ncomb; ihel++ )
       {
+        // NEW IMPLEMENTATION OF GETGOODHEL (#630): RESET THE RUNNING SUM OVER HELICITIES TO 0 BEFORE ADDING A NEW HELICITY
+        for( int ieppV = 0; ieppV < neppV; ++ieppV )
+        {
+          const int ievt = ievt00 + ieppV;
+          allMEs[ievt] = 0;
+#if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
+          const int ievt2 = ievt00 + ieppV + neppV;
+          allMEs[ievt2] = 0;
+#endif
+        }
         constexpr fptype_sv* jamp2_sv = nullptr; // no need for color selection during helicity filtering
         //std::cout << "sigmaKin_getGoodHel ihel=" << ihel << ( isGoodHel[ihel] ? " true" : " false" ) << std::endl;
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
@@ -331,22 +332,18 @@ namespace mg5amcCpu
         for( int ieppV = 0; ieppV < neppV; ++ieppV )
         {
           const int ievt = ievt00 + ieppV;
-          const bool differs = ( allMEs[ievt] != allMEsLast[ievt] );
-          if( differs )
+          if( allMEs[ievt] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
           {
             //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
             isGoodHel[ihel] = true;
           }
-          allMEsLast[ievt] = allMEs[ievt]; // running sum up to helicity ihel
 #if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
           const int ievt2 = ievt00 + ieppV + neppV;
-          const bool differs2 = ( allMEs[ievt2] != allMEsLast[ievt2] );
-          if( differs2 )
+          if( allMEs[ievt2] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
           {
             //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
             isGoodHel[ihel] = true;
           }
-          allMEsLast[ievt2] = allMEs[ievt2]; // running sum up to helicity ihel
 #endif
         }
       }

epochX/cudacpp/CODEGEN/generateAndCompare.sh

@@ -38,6 +38,12 @@ function codeGenAndDiff()
     gq_ttq)
       cmd="define q = u c d s u~ c~ d~ s~; generate g q > t t~ q"
       ;;
+    gu_ttu)
+      cmd="generate g u > t t~ u"
+      ;;
+    gg_uu)
+      cmd="generate g g > u u~"
+      ;;
     gq_ttllq)
       cmd="define q = u c d s u~ c~ d~ s~; generate g q > t t~ l- l+ q"
       ;;
@@ -66,11 +72,14 @@ function codeGenAndDiff()
       generate p p > t t~ z @0
       add process p p > t t~ z j @1"
       ;;
+    uu_dd)
+      cmd="generate u u~ > d d~"
+      ;;
     uu_tt)
       cmd="generate u u~ > t t~"
       ;;
-    uu_dd)
-      cmd="generate u u~ > d d~"
+    uu_ttg)
+      cmd="generate u u~ > t t~ g"
       ;;
     bb_tt)
       cmd="generate b b~ > t t~"

epochX/cudacpp/ee_mumu.mad/CODEGEN_mad_ee_mumu_log.txt

@@ -62,7 +62,7 @@ generate e+ e- > mu+ mu-
 No model currently active, so we import the Standard Model
 INFO: load particles 
 INFO: load vertices 
-DEBUG: model prefixing  takes 0.00445103645324707 
+DEBUG: model prefixing  takes 0.004861593246459961 
 INFO: Restrict model sm with file models/sm/restrict_default.dat . 
 DEBUG: Simplifying conditional expressions 
 DEBUG: remove interactions: u s w+ at order: QED=1 
@@ -176,7 +176,7 @@ INFO: Creating files in directory P1_epem_mupmum
 DEBUG:  Entering PLUGIN_OneProcessExporter.__init__ [model_handling.py at line 1039] 
 DEBUG:  kwargs[prefix] = 0 [model_handling.py at line 1040] 
 DEBUG:  proc_id =  1 [model_handling.py at line 1045] 
-DEBUG:  process_exporter_cpp =  <PLUGIN.CUDACPP_SA_OUTPUT.model_handling.PLUGIN_OneProcessExporter object at 0x7f7976656ca0> [export_v4.py at line 6179] 
+DEBUG:  process_exporter_cpp =  <PLUGIN.CUDACPP_SA_OUTPUT.model_handling.PLUGIN_OneProcessExporter object at 0x7f7e993895e0> [export_v4.py at line 6179] 
 INFO: Creating files in directory . 
 DEBUG:  Entering PLUGIN_OneProcessExporter.generate_process_files [model_handling.py at line 1297] 
 DEBUG:  self.include_multi_channel is already defined: this is madevent+second_exporter mode [model_handling.py at line 1299] 
@@ -210,7 +210,7 @@ INFO: Created files CPPProcess.h and CPPProcess.cc in directory ./.
 INFO: Generating Feynman diagrams for Process: e+ e- > mu+ mu- WEIGHTED<=4 @1 
 INFO: Finding symmetric diagrams for subprocess group epem_mupmum 
 Generated helas calls for 1 subprocesses (2 diagrams) in 0.004 s
-Wrote files for 8 helas calls in 0.094 s
+Wrote files for 8 helas calls in 0.092 s
 ALOHA: aloha starts to compute helicity amplitudes
 ALOHA: aloha creates FFV1 routines
 ALOHA: aloha creates FFV2 routines
@@ -222,7 +222,7 @@ ALOHA: aloha creates FFV1 routines
 ALOHA: aloha creates FFV2 routines
 ALOHA: aloha creates FFV4 routines
 ALOHA: aloha creates FFV2_4 routines
-ALOHA: aloha creates 7 routines in  0.215 s
+ALOHA: aloha creates 7 routines in  0.216 s
 <class 'aloha.create_aloha.AbstractRoutine'> FFV1
 <class 'aloha.create_aloha.AbstractRoutine'> FFV1
 <class 'aloha.create_aloha.AbstractRoutine'> FFV2
@@ -254,6 +254,6 @@ Type "launch" to generate events from this process, or see
 Run "open index.html" to see more information about this process.
 quit
 
-real	0m1.729s
-user	0m1.513s
-sys	0m0.202s
+real	0m1.771s
+user	0m1.516s
+sys	0m0.203s

epochX/cudacpp/ee_mumu.mad/SubProcesses/P1_epem_mupmum/CPPProcess.cc

@@ -637,12 +637,12 @@ namespace mg5amcCpu
                        fptype* allDenominators,    // output: multichannel denominators[nevt], running_sum_over_helicities
 #endif
                        bool* isGoodHel )           // output: isGoodHel[ncomb] - device array (CUDA implementation)
-  { /* clang-format on */
-    fptype allMEsLast = 0;
+  {                                                         /* clang-format on */
     const int ievt = blockDim.x * blockIdx.x + threadIdx.x; // index of event (thread) in grid
-    allMEs[ievt] = 0;
     for( int ihel = 0; ihel < ncomb; ihel++ )
     {
+      // NEW IMPLEMENTATION OF GETGOODHEL (#630): RESET THE RUNNING SUM OVER HELICITIES TO 0 BEFORE ADDING A NEW HELICITY
+      allMEs[ievt] = 0;
       // NB: calculate_wavefunctions ADDS |M|^2 for a given ihel to the running sum of |M|^2 over helicities for the given event(s)
       constexpr fptype_sv* jamp2_sv = nullptr; // no need for color selection during helicity filtering
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
@@ -651,12 +651,11 @@ namespace mg5amcCpu
 #else
       calculate_wavefunctions( ihel, allmomenta, allcouplings, allMEs, jamp2_sv );
 #endif
-      if( allMEs[ievt] != allMEsLast )
+      if( allMEs[ievt] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
       {
         //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
         isGoodHel[ihel] = true;
       }
-      allMEsLast = allMEs[ievt]; // running sum up to helicity ihel for event ievt
     }
   }
 #else
@@ -675,19 +674,11 @@ namespace mg5amcCpu
     //assert( (size_t)(allMEs) % mgOnGpu::cppAlign == 0 ); // SANITY CHECK: require SIMD-friendly alignment [COMMENT OUT TO TEST MISALIGNED ACCESS]
     // Allocate arrays at build time to contain at least 16 events (or at least neppV events if neppV>16, e.g. in future VPUs)
     constexpr int maxtry0 = std::max( 16, neppV ); // 16, but at least neppV (otherwise the npagV loop does not even start)
-    fptype allMEsLast[maxtry0] = { 0 };            // allocated at build time: maxtry0 must be a constexpr
     // Loop over only nevt events if nevt is < 16 (note that nevt is always >= neppV)
     assert( nevt >= neppV );
     const int maxtry = std::min( maxtry0, nevt ); // 16, but at most nevt (avoid invalid memory access if nevt<maxtry0)
 
-    // PART 0 - INITIALISATION (before calculate_wavefunctions)
-    // Reset the "matrix elements" - running sums of |M|^2 over helicities for the given event
-    for( int ievt = 0; ievt < maxtry; ++ievt )
-    {
-      allMEs[ievt] = 0; // all zeros
-    }
-
-    // PART 1 - HELICITY LOOP: CALCULATE WAVEFUNCTIONS
+    // HELICITY LOOP: CALCULATE WAVEFUNCTIONS
     const int npagV = maxtry / neppV;
 #if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
     // Mixed fptypes #537: float for color algebra and double elsewhere
@@ -706,6 +697,16 @@ namespace mg5amcCpu
 #endif
       for( int ihel = 0; ihel < ncomb; ihel++ )
       {
+        // NEW IMPLEMENTATION OF GETGOODHEL (#630): RESET THE RUNNING SUM OVER HELICITIES TO 0 BEFORE ADDING A NEW HELICITY
+        for( int ieppV = 0; ieppV < neppV; ++ieppV )
+        {
+          const int ievt = ievt00 + ieppV;
+          allMEs[ievt] = 0;
+#if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
+          const int ievt2 = ievt00 + ieppV + neppV;
+          allMEs[ievt2] = 0;
+#endif
+        }
         constexpr fptype_sv* jamp2_sv = nullptr; // no need for color selection during helicity filtering
         //std::cout << "sigmaKin_getGoodHel ihel=" << ihel << ( isGoodHel[ihel] ? " true" : " false" ) << std::endl;
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
@@ -717,22 +718,18 @@ namespace mg5amcCpu
         for( int ieppV = 0; ieppV < neppV; ++ieppV )
         {
           const int ievt = ievt00 + ieppV;
-          const bool differs = ( allMEs[ievt] != allMEsLast[ievt] );
-          if( differs )
+          if( allMEs[ievt] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
           {
             //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
             isGoodHel[ihel] = true;
           }
-          allMEsLast[ievt] = allMEs[ievt]; // running sum up to helicity ihel
 #if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
           const int ievt2 = ievt00 + ieppV + neppV;
-          const bool differs2 = ( allMEs[ievt2] != allMEsLast[ievt2] );
-          if( differs2 )
+          if( allMEs[ievt2] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
           {
             //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
             isGoodHel[ihel] = true;
           }
-          allMEsLast[ievt2] = allMEs[ievt2]; // running sum up to helicity ihel
 #endif
         }
       }

epochX/cudacpp/ee_mumu.sa/CODEGEN_cudacpp_ee_mumu_log.txt

@@ -62,7 +62,7 @@ generate e+ e- > mu+ mu-
 No model currently active, so we import the Standard Model
 INFO: load particles 
 INFO: load vertices 
-DEBUG: model prefixing  takes 0.004779815673828125 
+DEBUG: model prefixing  takes 0.004894733428955078 
 INFO: Restrict model sm with file models/sm/restrict_default.dat . 
 DEBUG: Simplifying conditional expressions 
 DEBUG: remove interactions: u s w+ at order: QED=1 
@@ -225,6 +225,6 @@ INFO: /data/avalassi/GPU2023/MG5aMC/ghav-mg5amcnlo/CODEGEN_cudacpp_ee_mumu/src/.
 DEBUG:  Entering PLUGIN_ProcessExporter.finalize [output.py at line 203] 
 quit
 
-real	0m0.644s
-user	0m0.577s
-sys	0m0.049s
+real	0m0.637s
+user	0m0.575s
+sys	0m0.055s

epochX/cudacpp/ee_mumu.sa/SubProcesses/P1_Sigma_sm_epem_mupmum/CPPProcess.cc

@@ -635,12 +635,12 @@ namespace mg5amcCpu
                        fptype* allDenominators,    // output: multichannel denominators[nevt], running_sum_over_helicities
 #endif
                        bool* isGoodHel )           // output: isGoodHel[ncomb] - device array (CUDA implementation)
-  { /* clang-format on */
-    fptype allMEsLast = 0;
+  {                                                         /* clang-format on */
     const int ievt = blockDim.x * blockIdx.x + threadIdx.x; // index of event (thread) in grid
-    allMEs[ievt] = 0;
     for( int ihel = 0; ihel < ncomb; ihel++ )
     {
+      // NEW IMPLEMENTATION OF GETGOODHEL (#630): RESET THE RUNNING SUM OVER HELICITIES TO 0 BEFORE ADDING A NEW HELICITY
+      allMEs[ievt] = 0;
       // NB: calculate_wavefunctions ADDS |M|^2 for a given ihel to the running sum of |M|^2 over helicities for the given event(s)
       constexpr fptype_sv* jamp2_sv = nullptr; // no need for color selection during helicity filtering
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
@@ -649,12 +649,11 @@ namespace mg5amcCpu
 #else
       calculate_wavefunctions( ihel, allmomenta, allcouplings, allMEs, jamp2_sv );
 #endif
-      if( allMEs[ievt] != allMEsLast )
+      if( allMEs[ievt] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
       {
         //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
         isGoodHel[ihel] = true;
       }
-      allMEsLast = allMEs[ievt]; // running sum up to helicity ihel for event ievt
     }
   }
 #else
@@ -673,19 +672,11 @@ namespace mg5amcCpu
     //assert( (size_t)(allMEs) % mgOnGpu::cppAlign == 0 ); // SANITY CHECK: require SIMD-friendly alignment [COMMENT OUT TO TEST MISALIGNED ACCESS]
     // Allocate arrays at build time to contain at least 16 events (or at least neppV events if neppV>16, e.g. in future VPUs)
     constexpr int maxtry0 = std::max( 16, neppV ); // 16, but at least neppV (otherwise the npagV loop does not even start)
-    fptype allMEsLast[maxtry0] = { 0 };            // allocated at build time: maxtry0 must be a constexpr
     // Loop over only nevt events if nevt is < 16 (note that nevt is always >= neppV)
     assert( nevt >= neppV );
     const int maxtry = std::min( maxtry0, nevt ); // 16, but at most nevt (avoid invalid memory access if nevt<maxtry0)
 
-    // PART 0 - INITIALISATION (before calculate_wavefunctions)
-    // Reset the "matrix elements" - running sums of |M|^2 over helicities for the given event
-    for( int ievt = 0; ievt < maxtry; ++ievt )
-    {
-      allMEs[ievt] = 0; // all zeros
-    }
-
-    // PART 1 - HELICITY LOOP: CALCULATE WAVEFUNCTIONS
+    // HELICITY LOOP: CALCULATE WAVEFUNCTIONS
     const int npagV = maxtry / neppV;
 #if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
     // Mixed fptypes #537: float for color algebra and double elsewhere
@@ -704,6 +695,16 @@ namespace mg5amcCpu
 #endif
       for( int ihel = 0; ihel < ncomb; ihel++ )
       {
+        // NEW IMPLEMENTATION OF GETGOODHEL (#630): RESET THE RUNNING SUM OVER HELICITIES TO 0 BEFORE ADDING A NEW HELICITY
+        for( int ieppV = 0; ieppV < neppV; ++ieppV )
+        {
+          const int ievt = ievt00 + ieppV;
+          allMEs[ievt] = 0;
+#if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
+          const int ievt2 = ievt00 + ieppV + neppV;
+          allMEs[ievt2] = 0;
+#endif
+        }
         constexpr fptype_sv* jamp2_sv = nullptr; // no need for color selection during helicity filtering
         //std::cout << "sigmaKin_getGoodHel ihel=" << ihel << ( isGoodHel[ihel] ? " true" : " false" ) << std::endl;
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
@@ -715,22 +716,18 @@ namespace mg5amcCpu
         for( int ieppV = 0; ieppV < neppV; ++ieppV )
         {
           const int ievt = ievt00 + ieppV;
-          const bool differs = ( allMEs[ievt] != allMEsLast[ievt] );
-          if( differs )
+          if( allMEs[ievt] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
           {
             //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
             isGoodHel[ihel] = true;
           }
-          allMEsLast[ievt] = allMEs[ievt]; // running sum up to helicity ihel
 #if defined MGONGPU_CPPSIMD and defined MGONGPU_FPTYPE_DOUBLE and defined MGONGPU_FPTYPE2_FLOAT
           const int ievt2 = ievt00 + ieppV + neppV;
-          const bool differs2 = ( allMEs[ievt2] != allMEsLast[ievt2] );
-          if( differs2 )
+          if( allMEs[ievt2] != 0 ) // NEW IMPLEMENTATION OF GETGOODHEL (#630): COMPARE EACH HELICITY CONTRIBUTION TO 0
           {
             //if ( !isGoodHel[ihel] ) std::cout << "sigmaKin_getGoodHel ihel=" << ihel << " TRUE" << std::endl;
             isGoodHel[ihel] = true;
           }
-          allMEsLast[ievt2] = allMEs[ievt2]; // running sum up to helicity ihel
 #endif
         }
       }