[susy] regenerate susy_gg_t1t1.mad with Olivier's patch PR madgraph5#852

 for issue madgraph5#826

epochX/cudacpp/susy_gg_t1t1.mad/CODEGEN_mad_susy_gg_t1t1_log.txt

@@ -554,7 +554,7 @@ INFO: Please specify coupling orders to bypass this step.
 INFO: Trying coupling order WEIGHTED<=2: WEIGTHED IS QCD+2*QED 
 INFO: Trying process: g g > t1 t1~ WEIGHTED<=2 @1  
 INFO: Process has 6 diagrams 
-1 processes with 6 diagrams generated in 0.132 s
+1 processes with 6 diagrams generated in 0.131 s
 Total: 1 processes with 6 diagrams
 output madevent_simd ../TMPOUT/CODEGEN_mad_susy_gg_t1t1 --hel_recycling=False --vector_size=32
 Load PLUGIN.CUDACPP_OUTPUT
@@ -577,7 +577,7 @@ INFO: Generating Helas calls for process: g g > t1 t1~ WEIGHTED<=2 @1
 INFO: Processing color information for process: g g > t1 t1~ @1 
 INFO: Creating files in directory P1_gg_t1t1x 
 DEBUG:  kwargs[prefix] = 0 [model_handling.py at line 1152] 
-DEBUG:  process_exporter_cpp =  <PLUGIN.CUDACPP_OUTPUT.model_handling.PLUGIN_OneProcessExporter object at 0x7fded7eebf10> [export_v4.py at line 6261] 
+DEBUG:  process_exporter_cpp =  <PLUGIN.CUDACPP_OUTPUT.model_handling.PLUGIN_OneProcessExporter object at 0x7fabd6301e80> [export_v4.py at line 6261] 
 INFO: Creating files in directory . 
 FileWriter <class 'PLUGIN.CUDACPP_OUTPUT.model_handling.PLUGIN_CPPWriter'> for ././CPPProcess.h
 FileWriter <class 'PLUGIN.CUDACPP_OUTPUT.model_handling.PLUGIN_CPPWriter'> for ././CPPProcess.cc
@@ -592,13 +592,18 @@ INFO: Created files CPPProcess.h and CPPProcess.cc in directory ./.
 DEBUG:  vector, subproc_group,self.opt['vector_size'] =  32 True 32 [export_v4.py at line 1871] 
 INFO: Generating Feynman diagrams for Process: g g > t1 t1~ WEIGHTED<=2 @1 
 INFO: Finding symmetric diagrams for subprocess group gg_t1t1x 
-Generated helas calls for 1 subprocesses (6 diagrams) in 0.009 s
-Wrote files for 16 helas calls in 0.117 s
+DEBUG:  os.getcwd() =  /data/avalassi/GPU2023/madgraph4gpuX/MG5aMC/TMPOUT/CODEGEN_mad_susy_gg_t1t1/SubProcesses/P1_gg_t1t1x [export_v4.py at line 6438] 
+DEBUG:  len(subproc_diagrams_for_config) =  5 [model_handling.py at line 1520] 
+DEBUG:  replace_dict =  {'diag_to_channel': '     {2, 0},\n     {3, 1},\n     {4, 2},\n     {5, 3},\n     {6, 4},'} [model_handling.py at line 1543] 
+DEBUG:  multi_channel =  {1: [1], 2: [2], 3: [3], 4: [4], 5: [5]} [model_handling.py at line 1551] 
+DEBUG:  self.path, os.getcwd() =  . /data/avalassi/GPU2023/madgraph4gpuX/MG5aMC/TMPOUT/CODEGEN_mad_susy_gg_t1t1/SubProcesses/P1_gg_t1t1x [model_handling.py at line 1552] 
+Generated helas calls for 1 subprocesses (6 diagrams) in 0.008 s
+Wrote files for 16 helas calls in 0.130 s
 ALOHA: aloha starts to compute helicity amplitudes
 ALOHA: aloha creates VVV1 set of routines with options: P0
 ALOHA: aloha creates VSS1 routines
 ALOHA: aloha creates VVSS1 routines
-ALOHA: aloha creates 3 routines in  0.196 s
+ALOHA: aloha creates 3 routines in  0.197 s
 DEBUG:  Entering PLUGIN_ProcessExporter.convert_model (create the model) [output.py at line 205] 
 ALOHA: aloha starts to compute helicity amplitudes
 ALOHA: aloha creates VVV1 set of routines with options: P0
@@ -645,10 +650,10 @@ Type "launch" to generate events from this process, or see
 Run "open index.html" to see more information about this process.
 quit
 
-real	0m2.960s
-user	0m2.686s
-sys	0m0.271s
-Code generation completed in 3 seconds
+real	0m3.863s
+user	0m2.681s
+sys	0m0.265s
+Code generation completed in 4 seconds
 ************************************************************
 *                                                          *
 *                      W E L C O M E to                    *

epochX/cudacpp/susy_gg_t1t1.mad/SubProcesses/P1_gg_t1t1x/CPPProcess.cc

@@ -1009,7 +1009,7 @@ namespace mg5amcCpu
     // Event-by-event random choice of color #402
     if( channelId != 0 ) // no event-by-event choice of color if channelId == 0 (fix FPE #783)
     {
-      const unsigned int channelIdC = channelId - 1; // coloramps.h uses the C array indexing starting at 0
+      const unsigned int channelIdC = mgOnGpu::diag_to_channel[channelId]; // coloramps.h uses a channel ordering not the diagram id
       fptype targetamp[ncolor] = { 0 };
       for( int icolC = 0; icolC < ncolor; icolC++ )
       {
@@ -1052,7 +1052,7 @@ namespace mg5amcCpu
     // - firstprivate: give each thread its own copy, and initialise with value from outside
 #define _OMPLIST0 allcouplings, allMEs, allmomenta, allrndcol, allrndhel, allselcol, allselhel, cGoodHel, cNGoodHel, npagV2
 #ifdef MGONGPU_SUPPORTS_MULTICHANNEL
-#define _OMPLIST1 , allDenominators, allNumerators, channelId, mgOnGpu::icolamp
+#define _OMPLIST1 , allDenominators, allNumerators, channelId, mgOnGpu::icolamp, mgOnGpu::diag_to_channel
 #else
 #define _OMPLIST1
 #endif
@@ -1124,7 +1124,7 @@ namespace mg5amcCpu
       // Event-by-event random choice of color #402
       if( channelId != 0 ) // no event-by-event choice of color if channelId == 0 (fix FPE #783)
       {
-        const unsigned int channelIdC = channelId - 1; // coloramps.h uses the C array indexing starting at 0
+        const unsigned int channelIdC = mgOnGpu::diag_to_channel[channelId]; // coloramps.h uses a channel ordering not the diagram id
         fptype_sv targetamp[ncolor] = { 0 };
         for( int icolC = 0; icolC < ncolor; icolC++ )
         {

epochX/cudacpp/susy_gg_t1t1.mad/SubProcesses/P1_gg_t1t1x/coloramps.h

@@ -6,10 +6,19 @@
 #ifndef COLORAMPS_H
 #define COLORAMPS_H 1
 
+#include <map>
+
 namespace mgOnGpu
 {
 
-  __device__ constexpr bool icolamp[5][2] = {
+    __device__ std::map<int,int> diag_to_channel = {
+             {2, 0},
+     {3, 1},
+     {4, 2},
+     {5, 3},
+     {6, 4},
+  };          
+    __device__ constexpr bool icolamp[5][2] = {
     { true, true },
     { true, true },
     { true, false },