diff --git a/MG5aMC/mg5amcnlo b/MG5aMC/mg5amcnlo
index cb4771efc5..3597d123f6 160000
--- a/MG5aMC/mg5amcnlo
+++ b/MG5aMC/mg5amcnlo
@@ -1 +1 @@
-Subproject commit cb4771efc507953c5ae03320ef5ee5496fabbff3
+Subproject commit 3597d123f6c954bdba4a49387b8e89a5f2b06dcc
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/MG5aMC_patches/PROD/patch.common b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/MG5aMC_patches/PROD/patch.common
index 2e272b419b..8f96dab5ef 100644
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/MG5aMC_patches/PROD/patch.common
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/MG5aMC_patches/PROD/patch.common
@@ -25,7 +25,7 @@ index 617f10b93..dbe08b846 100644
  	$(RM) *.o $(LIBRARIES) $(BINARIES)
  	cd PDF; make clean; cd ..
  	cd PDF/gammaUPC; make clean; cd ../../
-@@ -132,4 +132,9 @@ clean:
+@@ -132,4 +132,11 @@ clean:
  	cd BIAS/ptj_bias; make clean; cd ../..
  	if [ -d $(CUTTOOLSDIR) ]; then cd $(CUTTOOLSDIR); make clean; cd ..; fi
  	if [ -d $(IREGIDIR) ]; then cd $(IREGIDIR); make clean; cd ..; fi
@@ -33,8 +33,11 @@ index 617f10b93..dbe08b846 100644
 +clean: cleanSource
  	for i in `ls -d ../SubProcesses/P*`; do cd $$i; make clean; cd -; done;
 +
-+cleanall: cleanSource
++cleanavx:
 +	for i in `ls -d ../SubProcesses/P*`; do cd $$i; make cleanavxs; cd -; done;
++cleanall: cleanSource # THIS IS THE ONE
++	for i in `ls -d ../SubProcesses/P*`; do cd $$i; make cleanavxs; cd -; done;
++
 diff --git b/epochX/cudacpp/gg_tt.mad/SubProcesses/makefile a/epochX/cudacpp/gg_tt.mad/SubProcesses/makefile
 index 348c283be..74db44d84 100644
 --- b/epochX/cudacpp/gg_tt.mad/SubProcesses/makefile
@@ -78,12 +81,12 @@ index 348c283be..74db44d84 100644
 +CUDACPP_MAKEENV:=$(shell echo '$(.VARIABLES)' | tr " " "\n" | egrep "(USEBUILDDIR|AVX|FPTYPE|HELINL|HRDCOD)")
 +###$(info CUDACPP_MAKEENV=$(CUDACPP_MAKEENV))
 +###$(info $(foreach v,$(CUDACPP_MAKEENV),$(v)="$($(v))"))
-+CUDACPP_BUILDDIR:=$(shell $(MAKE) $(foreach v,$(CUDACPP_MAKEENV),$(v)="$($(v))") -f $(CUDACPP_MAKEFILE) -pn |& awk '/Building/{print $$3}' | sed s/BUILDDIR=//)
-+ifeq ($(CUDACPP_BUILDDIR),)
-+$(error CUDACPP_BUILDDIR='$(CUDACPP_BUILDDIR)' should not be empty!)
-+else
++CUDACPP_BUILDDIR:=$(shell $(MAKE) $(foreach v,$(CUDACPP_MAKEENV),$(v)="$($(v))") -f $(CUDACPP_MAKEFILE) -pn 2>/dev/null | awk '/Building/{print $$3}' | sed s/BUILDDIR=//)
++#ifeq ($(CUDACPP_BUILDDIR),)
++#$(error CUDACPP_BUILDDIR='$(CUDACPP_BUILDDIR)' should not be empty!)
++#else
 +$(info CUDACPP_BUILDDIR='$(CUDACPP_BUILDDIR)')
-+endif
++#endif
 +CUDACPP_COMMONLIB=mg5amc_common
 +CUDACPP_CXXLIB=mg5amc_$(processid_short)_cpp
 +CUDACPP_CULIB=mg5amc_$(processid_short)_cuda
@@ -110,7 +113,7 @@ index 348c283be..74db44d84 100644
  
 -$(PROG): $(PROCESS) auto_dsig.o $(LIBS) $(MATRIX)
 -	$(FC) -o $(PROG) $(PROCESS) $(MATRIX) $(LINKLIBS) $(LDFLAGS) $(BIASDEPENDENCIES) -fopenmp
-+LDFLAGS+=-Wl,--no-relax # avoid 'failed to convert GOTPCREL relocation' error #458
++#LDFLAGS+=-Wl,--no-relax # avoid 'failed to convert GOTPCREL relocation' error #458 (flag not universal -> skip?)
  
 -$(PROG)_forhel: $(PROCESS) auto_dsig.o $(LIBS) $(MATRIX_HEL)
 -	$(FC) -o $(PROG)_forhel $(PROCESS) $(MATRIX_HEL) $(LINKLIBS) $(LDFLAGS) $(BIASDEPENDENCIES) -fopenmp
@@ -283,44 +286,6 @@ index 348c283be..74db44d84 100644
 +
 +distclean: cleanall # Clean all fortran and cudacpp builds as well as the googletest installation
 +	$(MAKE) -f $(CUDACPP_MAKEFILE) distclean
-diff --git b/epochX/cudacpp/gg_tt.mad/Source/make_opts a/epochX/cudacpp/gg_tt.mad/Source/make_opts
-index 57f5f7bb9..bd3c24228 100644
---- b/epochX/cudacpp/gg_tt.mad/Source/make_opts
-+++ a/epochX/cudacpp/gg_tt.mad/Source/make_opts
-@@ -1,12 +1,17 @@
-+pdlabel1=
-+pdlabel2=
-+lhapdf=
-+PYTHIA8_PATH=NotInstalled
-+MG5AMC_VERSION=3.5.0_lo_vect
- GLOBAL_FLAG=-O3 -ffast-math -fbounds-check
-+ALOHA_FLAG=
-+MATRIX_FLAG=
- DEFAULT_CPP_COMPILER=g++
- MACFLAG=
- STDLIB=-lstdc++
- STDLIB_FLAG=
- DEFAULT_F_COMPILER=gfortran
- DEFAULT_F2PY_COMPILER=f2py3
--MG5AMC_VERSION=SpecifiedByMG5aMCAtRunTime
--PYTHIA8_PATH=NotInstalled
- #end_of_make_opts_variables
- 
- BIASLIBDIR=../../../lib/
-diff --git b/epochX/cudacpp/gg_tt.mad/bin/internal/banner.py a/epochX/cudacpp/gg_tt.mad/bin/internal/banner.py
-index 8f8df219d..7624b9f55 100755
---- b/epochX/cudacpp/gg_tt.mad/bin/internal/banner.py
-+++ a/epochX/cudacpp/gg_tt.mad/bin/internal/banner.py
-@@ -4187,7 +4187,8 @@ class RunCardLO(RunCard):
-         self.add_param('mxxmin4pdg',[-1.], system=True)
-         self.add_param('mxxpart_antipart', [False], system=True)
-                      
--        
-+        # CUDACPP parameters
-+        self.add_param('cudacpp_backend', 'CPP', include=False, hidden=False)
-              
-     def check_validity(self):
-         """ """
 diff --git b/epochX/cudacpp/gg_tt.mad/bin/internal/gen_ximprove.py a/epochX/cudacpp/gg_tt.mad/bin/internal/gen_ximprove.py
 index 4dd71db86..3b8ec3121 100755
 --- b/epochX/cudacpp/gg_tt.mad/bin/internal/gen_ximprove.py
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/__init__.py b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/__init__.py
index f302f9c746..a0cd9dbfb3 100644
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/__init__.py
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/__init__.py
@@ -32,7 +32,12 @@
     #    allows the command "output myformat PATH" in madgraph.
     #    MYCLASS should inherit from class madgraph.iolibs.export_v4.VirtualExporter
     import PLUGIN.CUDACPP_OUTPUT.output as output
-    new_output = { 'standalone_cudacpp' : output.PLUGIN_ProcessExporter }
+    new_output = { 'madevent_simd' : output.SIMD_ProcessExporter,
+                   'madevent_gpu' : output.GPU_ProcessExporter,
+                   'standalone_cudacpp' : output.PLUGIN_ProcessExporter,
+                   'standalone_simd' :  output.SIMD_ProcessExporter,
+                   'standalone_cuda' :  output.GPU_ProcessExporter,
+                  }
 
     # 2. Define new way to handle the cluster.
     #    Example: new_cluster = {'mycluster': MYCLUSTERCLASS}
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/launch_plugin.py b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/launch_plugin.py
new file mode 100644
index 0000000000..23271b846e
--- /dev/null
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/launch_plugin.py
@@ -0,0 +1,100 @@
+
+import logging
+import os
+import subprocess
+pjoin = os.path.join
+logger = logging.getLogger('cmdprint') # for stdout
+
+try:
+    import madgraph
+except ImportError:
+    import internal.madevent_interface as madevent_interface
+    import internal.misc as misc
+    import internal.extended_cmd as extended_cmd
+    import internal.banner as banner_mod
+else:
+    import madgraph.interface.madevent_interface as madevent_interface
+    import madgraph.various.misc as misc
+    import madgraph.interface.extended_cmd as extended_cmd
+    import madgraph.various.banner as banner_mod
+
+class CPPMEInterface(madevent_interface.MadEventCmdShell):
+    
+    def compile(self, *args, **opts):
+        """ """
+        
+        import multiprocessing
+        if not self.options['nb_core'] or self.options['nb_core'] == 'None':
+            self.options['nb_core'] = multiprocessing.cpu_count()
+    
+        if args and args[0][0] == 'madevent' and hasattr(self, 'run_card'):
+            import pathlib
+            import os
+            pjoin = os.path.join
+            
+
+
+            
+            cudacpp_backend = self.run_card['cudacpp_backend'].upper() # the default value is defined in banner.py
+            logger.info("Building madevent in madevent_interface.py with '%s' matrix elements"%cudacpp_backend)
+            if cudacpp_backend == 'FORTRAN':
+                args[0][0] = 'madevent_fortran_link'
+            elif cudacpp_backend == 'CPP':
+                args[0][0] = 'madevent_cpp_link'
+            elif cudacpp_backend == 'CUDA':
+                args[0][0] = 'madevent_cuda_link'
+            else:
+                raise Exception("Invalid cudacpp_backend='%s': only 'FORTRAN', 'CPP', 'CUDA' are supported")
+            return misc.compile(nb_core=self.options['nb_core'], *args, **opts)
+        else:
+            return misc.compile(nb_core=self.options['nb_core'], *args, **opts)
+
+class CPPRunCard(banner_mod.RunCardLO):
+
+    def reset_simd(self, old_value, new_value, name):
+        if not hasattr(self, 'path'):
+            raise Exception
+
+        if name == "vector_size" and new_value <= int(old_value):
+            # code can handle the new size -> do not recompile
+            return
+        
+        Sourcedir = pjoin(os.path.dirname(os.path.dirname(self.path)), 'Source')
+        subprocess.call(['make', 'cleanavx'], cwd=Sourcedir, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+        
+    def plugin_input(self, finput):
+        return
+
+    def default_setup(self):
+        super().default_setup()
+        self.add_param('cudacpp_backend', 'CPP', include=False, hidden=False)
+        
+
+    def write_one_include_file(self, output_dir, incname, output_file=None):
+        """write one include file at the time"""
+
+        if incname == "vector.inc" and 'vector_size' not in self.user_set:
+            return
+        super().write_one_include_file(output_dir, incname, output_file)
+
+
+    def check_validity(self):
+        """ensure that PLUGIN information are consistent"""
+        
+        super().check_validity()
+
+        if self['SDE_strategy'] != 1:
+            logger.warning('SDE_strategy different of 1 is not supported with SMD/GPU mode')
+            self['sde_strategy'] = 1
+
+        if self['hel_recycling']:
+            self['hel_recycling'] = False
+            
+class GPURunCard(CPPRunCard):
+
+    def default_setup(self):
+        super(CPPRunCard, self).default_setup()
+        self.add_param('cudacpp_backend', 'CUDA', include=False, hidden=False)
+        
+MEINTERFACE = CPPMEInterface
+RunCard = CPPRunCard
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/MatrixElementKernels.cc b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/MatrixElementKernels.cc
index 30257195b6..cfed7d2af5 100644
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/MatrixElementKernels.cc
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/MatrixElementKernels.cc
@@ -112,10 +112,17 @@ namespace mg5amcCpu
     // See https://community.arm.com/arm-community-blogs/b/operating-systems-blog/posts/runtime-detection-of-cpu-features-on-an-armv8-a-cpu
     bool ok = true; // this is just an assumption!
     const std::string tag = "arm neon (128bit as in SSE4.2)";
-#else
+#elif  defined(__x86_64__) || defined(__i386__)
     bool known = true;
     bool ok = __builtin_cpu_supports( "sse4.2" );
     const std::string tag = "nehalem (SSE4.2)";
+#else
+     bool known = false; // __builtin_cpu_supports is not supported
+    // See https://gcc.gnu.org/onlinedocs/gcc/Basic-PowerPC-Built-in-Functions-Available-on-all-Configurations.html
+    // See https://stackoverflow.com/q/62783908
+    // See https://community.arm.com/arm-community-blogs/b/operating-systems-blog/posts/runtime-detection-of-cpu-features-on-an-armv8-a-cpu
+    bool ok = true; // this is just an assumption!
+    const std::string tag = "arm neon (128bit as in SSE4.2)";
 #endif
 #else
     bool known = true;
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/cudacpp.mk b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/cudacpp.mk
index 222d75f846..49a5856085 100644
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/cudacpp.mk
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/cudacpp.mk
@@ -27,6 +27,7 @@ UNAME_S := $(shell uname -s)
 UNAME_P := $(shell uname -p)
 ###$(info UNAME_P='$(UNAME_P)')
 
+include ../../Source/make_opts
 #-------------------------------------------------------------------------------
 
 #=== Configure common compiler flags for C++ and CUDA
@@ -220,7 +221,7 @@ else ifneq ($(shell $(CXX) --version | egrep '^(clang)'),)
 override OMPFLAGS = -fopenmp
 ###override OMPFLAGS = # disable OpenMP MT on clang (was not ok without or with nvcc before #578)
 else ifneq ($(shell $(CXX) --version | egrep '^(Apple clang)'),)
-override OMPFLAGS = # disable OpenMP MT on Apple clang (builds fail in the CI #578)
+override OMPFLAGS = -fopenmp # disable OpenMP MT on Apple clang (builds fail in the CI #578)
 else
 override OMPFLAGS = -fopenmp
 ###override OMPFLAGS = # disable OpenMP MT (default before #575)
@@ -554,7 +555,7 @@ endif
 $(LIBDIR)/lib$(MG5AMC_CXXLIB).so: $(BUILDDIR)/fbridge.o
 $(LIBDIR)/lib$(MG5AMC_CXXLIB).so: cxx_objects_lib += $(BUILDDIR)/fbridge.o
 $(LIBDIR)/lib$(MG5AMC_CXXLIB).so: $(LIBDIR)/lib$(MG5AMC_COMMONLIB).so $(cxx_objects_lib)
-	$(CXX) -shared -o $@ $(cxx_objects_lib) $(CXXLIBFLAGSRPATH2) -L$(LIBDIR) -l$(MG5AMC_COMMONLIB)
+	$(CXX) -shared -o $@ $(cxx_objects_lib) $(CXXLIBFLAGSRPATH2) -L$(LIBDIR) -l$(MG5AMC_COMMONLIB) $(LIBFLAGS) -fopenmp
 
 ifneq ($(NVCC),)
 $(LIBDIR)/lib$(MG5AMC_CULIB).so: $(BUILDDIR)/fbridge_cu.o
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/process_function_definitions.inc b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/process_function_definitions.inc
index c3c0812b94..a3ce5ad2dc 100644
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/process_function_definitions.inc
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/process_function_definitions.inc
@@ -405,8 +405,7 @@ namespace mg5amcCpu
     {
       // nprocesses>1 was last observed for "mirror processes" in uux_ttx in the 270 branch (see issue #343 and PRs #360 and #396)
       constexpr int nprocesses = %(nproc)i;
-      static_assert( nprocesses == 1, "Assume nprocesses == 1" );
-      // process_id corresponds to the index of DSIG1 Fortran functions (must be 1 because cudacpp is unable to handle DSIG2)
+      static_assert( nprocesses == 1 || nprocesses == 2, "Assume nprocesses == 1 or 2" );
       constexpr int process_id = %(proc_id)i; // code generation source: %(proc_id_source)s
       static_assert( process_id == 1, "Assume process_id == 1" );
     }
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/model_handling.py b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/model_handling.py
index a73c41d7c3..0e634c4fe6 100644
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/model_handling.py
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/model_handling.py
@@ -196,6 +196,7 @@ def get_header_txt(self, name=None, couplings=None,mode=''):
                 list_arg = '[]' # AV from cxtype_sv to fptype array (running alphas #373)
                 point = self.type2def['pointer_coup']
                 args.append('%s %s%s%s'% (type, point, argname, list_arg))
+                args.append('double Ccoeff%s'% argname[7:])
             else:
                 args.append('%s %s%s'% (type, argname, list_arg))
         if not self.offshell:
@@ -534,6 +535,15 @@ def write_MultContainer(self, obj, prefactor=True):
             text = '%(factors)s'
         return text % data
 
+    def change_var_format(self, obj):
+        """ """
+        if obj.startswith('COUP'):
+            out = super().change_var_format(obj)
+            postfix = out[4:]
+            return "Ccoeff%s*%s" % (postfix, out)
+        else:
+            return super().change_var_format(obj)
+
     # AV - new method (based on implementation of write_obj and write_MultVariable)
     def objIsSimpleVariable(self, obj) :
         ###print ( obj.vartype, obj.prefactor, len( obj ), obj ) # AV - FOR DEBUGGING
@@ -726,7 +736,6 @@ def write_hardcoded_parameters(self, params):
             for par in pars.split(','):
                 ###print(len(pardef_lines), par) # for debugging
                 pardef_lines[par] = ( 'constexpr ' + type + ' ' + par )
-        misc.sprint( 'pardef_lines size =', len(pardef_lines), ', keys size =', len(pardef_lines.keys()) )
         ###print( pardef_lines ) # for debugging
         ###for line in pardef_lines: misc.sprint(line) # for debugging
         parset_pars = []
@@ -1043,14 +1052,14 @@ class PLUGIN_OneProcessExporter(PLUGIN_export_cpp.OneProcessExporterGPU):
 
     # AV - overload export_cpp.OneProcessExporterGPU constructor (rename gCPPProcess to CPPProcess, set include_multi_channel)
     def __init__(self, *args, **kwargs):
-        misc.sprint('Entering PLUGIN_OneProcessExporter.__init__')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.__init__')
         for kwarg in kwargs: misc.sprint( 'kwargs[%s] = %s' %( kwarg, kwargs[kwarg] ) )
         super().__init__(*args, **kwargs)
         self.process_class = 'CPPProcess'
         ###if self.in_madevent_mode: proc_id = kwargs['prefix']+1 # madevent+cudacpp (NB: HERE SELF.IN_MADEVENT_MODE DOES NOT WORK!)
         if 'prefix' in kwargs: proc_id = kwargs['prefix']+1 # madevent+cudacpp (ime+1 from ProcessExporterFortranMEGroup.generate_subprocess_directory)
         else: proc_id = 0 # standalone_cudacpp
-        misc.sprint(proc_id)
+#        misc.sprint(proc_id)
         self.proc_id = proc_id
 
     # AV - overload export_cpp.OneProcessExporterGPU method (indent comments in process_lines)
@@ -1148,12 +1157,18 @@ def get_process_function_definitions(self, write=True):
 
     # AV - modify export_cpp.OneProcessExporterGPU method (add debug printouts for multichannel #342)
     def get_sigmaKin_lines(self, color_amplitudes, write=True):
-        misc.sprint('Entering PLUGIN_OneProcessExporter.get_sigmaKin_lines')
-        misc.sprint(self.include_multi_channel)
-        misc.sprint(self.support_multichannel)
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.get_sigmaKin_lines')
+#        misc.sprint(self.include_multi_channel)
+#        misc.sprint(self.support_multichannel)
         replace_dict = super().get_sigmaKin_lines(color_amplitudes, write=False)
         replace_dict['proc_id'] = self.proc_id if self.proc_id>0 else 1
         replace_dict['proc_id_source'] = 'madevent + cudacpp exporter' if self.proc_id>0 else 'standalone_cudacpp' # FIXME? use self.in_madevent_mode instead?
+
+        # Extract denominator (avoid to extend size for mirroring)
+        den_factors = [str(me.get_denominator_factor()) for me in \
+                            self.matrix_elements]
+        replace_dict['den_factors'] = ",".join(den_factors)
+
         if write:
             file = self.read_template_file(self.process_sigmaKin_function_template) % replace_dict
             file = '\n'.join( file.split('\n')[8:] ) # skip first 8 lines in process_sigmaKin_function.inc (copyright)
@@ -1167,14 +1182,14 @@ def get_all_sigmaKin_lines(self, color_amplitudes, class_name):
         ret_lines = []
         if self.single_helicities:
             ###assert self.include_multi_channel # remove this assert: must handle both cases and produce two different code bases (#473)
-            misc.sprint(type(self.helas_call_writer))
-            misc.sprint(self.support_multichannel, self.include_multi_channel)
+#            misc.sprint(type(self.helas_call_writer))
+#            misc.sprint(self.support_multichannel, self.include_multi_channel)
             multi_channel = None
             if self.include_multi_channel:
                 if not self.support_multichannel:
                     raise Exception("link with madevent not supported")
                 multi_channel = self.get_multi_channel_dictionary(self.matrix_elements[0].get('diagrams'), self.include_multi_channel)
-                misc.sprint(multi_channel)
+#                misc.sprint(multi_channel)
             ###misc.sprint( 'before get_matrix_element_calls', self.matrix_elements[0].get_number_of_wavefunctions() ) # WRONG value of nwf, eg 7 for gg_tt
             helas_calls = self.helas_call_writer.get_matrix_element_calls(\
                                                     self.matrix_elements[0],
@@ -1302,12 +1317,9 @@ def get_process_info_lines(self, matrix_element):
     # AV - replace the export_cpp.OneProcessExporterGPU method (invert .cc/.cu, add debug printouts)
     def generate_process_files(self):
         """Generate mgOnGpuConfig.h, CPPProcess.cc, CPPProcess.h, check_sa.cc, gXXX.cu links"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.generate_process_files')
-        if self.include_multi_channel:
-            misc.sprint('self.include_multi_channel is already defined: this is madevent+second_exporter mode') # FIXME? use self.in_madevent_mode instead?
-        else:
-            misc.sprint('self.include_multi_channel is not yet defined: this is standalone_cudacpp mode') # see issue #473
-            # I move those line to standalone_cudacpp mode (but do we need those at all???)
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.generate_process_files')
+        if not self.include_multi_channel:
+            #this condition is likely wrong and need to be removed
             if self.matrix_elements[0].get('has_mirror_process'):
                 self.matrix_elements[0].set('has_mirror_process', False)
                 self.nprocesses/=2
@@ -1341,15 +1353,15 @@ def generate_process_files(self):
         ###template_ref = 'dump_CPUTest.'+self.process_name+'.txt'
         template_ref = self.template_path + '/../../../test/ref/' + 'dump_CPUTest.' + self.process_name + '.txt'
         if os.path.exists( template_ref ):
-            misc.sprint( 'Copying test reference file: ', template_ref )
+#            misc.sprint( 'Copying test reference file: ', template_ref )
             PLUGIN_export_cpp.cp( template_ref, self.path + '/../../test/ref' )
-        else:
-            misc.sprint( 'Test reference file does not exist and will not be copied: ', template_ref )
+#        else:
+#            misc.sprint( 'Test reference file does not exist and will not be copied: ', template_ref )
 
     # SR - generate CMakeLists.txt file inside the P* directory
     def edit_CMakeLists(self):
         """Generate CMakeLists.txt"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_CMakeLists')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_CMakeLists')
         template = open(pjoin(self.template_path,'CMake/SubProcesses/CMakeLists_P.txt'),'r').read()
         ff = open(pjoin(self.path, 'CMakeLists.txt'),'w')
         ff.write(template)
@@ -1358,7 +1370,7 @@ def edit_CMakeLists(self):
     # AV - replace the export_cpp.OneProcessExporterGPU method (invert .cc/.cu, add debug printouts)
     def edit_check_sa(self):
         """Generate check_sa.cc and fcheck_sa.f"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_check_sa')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_check_sa')
         ff = open(pjoin(self.path, 'check_sa.cc'),'w')
         template = open(pjoin(self.template_path,'gpu','check_sa.cc'),'r').read()
         ff.write(template) # nothing to replace in check_sa.cc
@@ -1375,7 +1387,7 @@ def edit_check_sa(self):
     # AV - replace the export_cpp.OneProcessExporterGPU method (add debug printouts and multichannel handling #473) 
     def edit_mgonGPU(self):
         """Generate mgOnGpuConfig.h"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_mgonGPU')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_mgonGPU')
         template = open(pjoin(self.template_path,'gpu','mgOnGpuConfig.h'),'r').read()
         replace_dict = {}
         nexternal, nincoming = self.matrix_elements[0].get_nexternal_ninitial()
@@ -1395,7 +1407,7 @@ def edit_mgonGPU(self):
     # AV - new method
     def edit_processidfile(self):
         """Generate epoch_process_id.h"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_processidfile')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_processidfile')
         template = open(pjoin(self.template_path,'gpu','epoch_process_id.h'),'r').read()
         replace_dict = {}
         replace_dict['processid'] = self.get_process_name()
@@ -1407,7 +1419,7 @@ def edit_processidfile(self):
     # AV - new method
     def edit_coloramps(self):
         """Generate coloramps.h"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_coloramps')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_coloramps')
         template = open(pjoin(self.template_path,'gpu','coloramps.h'),'r').read()
         ff = open(pjoin(self.path, 'coloramps.h'),'w')
         # The following five lines from OneProcessExporterCPP.get_sigmaKin_lines (using OneProcessExporterCPP.get_icolamp_lines)
@@ -1425,7 +1437,7 @@ def edit_coloramps(self):
     # AV - new method
     def edit_testxxx(self):
         """Generate testxxx.cc"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_testxxx')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_testxxx')
         template = open(pjoin(self.template_path,'gpu','testxxx.cc'),'r').read()
         replace_dict = {}
         replace_dict['model_name'] = self.model_name
@@ -1436,7 +1448,7 @@ def edit_testxxx(self):
     # AV - new method
     def edit_memorybuffers(self):
         """Generate MemoryBuffers.h"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_memorybuffers')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_memorybuffers')
         template = open(pjoin(self.template_path,'gpu','MemoryBuffers.h'),'r').read()
         replace_dict = {}
         replace_dict['model_name'] = self.model_name
@@ -1447,7 +1459,7 @@ def edit_memorybuffers(self):
     # AV - new method
     def edit_memoryaccesscouplings(self):
         """Generate MemoryAccessCouplings.h"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_memoryaccesscouplings')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.edit_memoryaccesscouplings')
         template = open(pjoin(self.template_path,'gpu','MemoryAccessCouplings.h'),'r').read()
         replace_dict = {}
         replace_dict['model_name'] = self.model_name
@@ -1459,7 +1471,7 @@ def edit_memoryaccesscouplings(self):
     # [*NB export_cpp.UFOModelConverterGPU.write_process_h_file is not called!*]
     def write_process_h_file(self, writer):
         """Generate final gCPPProcess.h"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.write_process_h_file')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.write_process_h_file')
         out = super().write_process_h_file(writer)
         writer.seek(-1, os.SEEK_CUR)
         writer.truncate()
@@ -1481,7 +1493,7 @@ def super_write_process_cc_file(self, writer):
     # AV - overload the export_cpp.OneProcessExporterGPU method (add debug printout and truncate last \n)
     def write_process_cc_file(self, writer):
         """Generate CPPProcess.cc"""
-        misc.sprint('Entering PLUGIN_OneProcessExporter.write_process_cc_file')
+#        misc.sprint('Entering PLUGIN_OneProcessExporter.write_process_cc_file')
         ###out = super().write_process_cc_file(writer)
         out = self.super_write_process_cc_file(writer)
         writer.seek(-1, os.SEEK_CUR)
@@ -1625,13 +1637,14 @@ def format_coupling(self, call):
                 alias[coup] = len(alias)
             if name == 'cIPD':
                 call = call.replace('m_pars->%s%s' % (sign, coup),
-                                    '%s%s[%s]' % (sign, name, alias[coup]))
+                                    '%s%s[%s]' % (sign, name, alias[coup]))                        
             else:
                 ###call = call.replace('m_pars->%s%s' % (sign, coup),
                 ###                    '%scxmake( cIPC[%s], cIPC[%s] )' %
                 ###                    (sign, 2*alias[coup],2*alias[coup]+1))
+                # AV from cIPCs to COUP array (running alphas #373)
                 call = call.replace('m_pars->%s%s' % (sign, coup),
-                                    '%sCOUPs[%s]' % (sign, alias[coup])) # AV from cIPCs to COUP array (running alphas #373)
+                                    'COUPs[%s], %s' % (alias[coup], '1.0' if not sign else '-1.0')) 
         return call
 
     # AV - new method for formatting wavefunction/amplitude calls
@@ -1660,7 +1673,7 @@ def super_get_matrix_element_calls(self, matrix_element, color_amplitudes, multi
                 color[namp][njamp] = coeff
         me = matrix_element.get('diagrams')
         matrix_element.reuse_outdated_wavefunctions(me)
-        misc.sprint(multi_channel_map)
+#        misc.sprint(multi_channel_map)
         res = []
         ###res.append('for(int i=0;i<%s;i++){jamp[i] = cxtype(0.,0.);}' % len(color_amplitudes))
         res.append("""constexpr size_t nxcoup = ndcoup + nicoup; // both dependent and independent couplings
@@ -1715,7 +1728,7 @@ def super_get_matrix_element_calls(self, matrix_element, color_amplitudes, multi
                                   sum([diagrams[idiag].get('amplitudes') for \
                                        idiag in multi_channel_map[config]], [])]
                 diag_to_config[amp[0]] = config
-        misc.sprint(diag_to_config)
+#        misc.sprint(diag_to_config)
         id_amp = 0
         for diagram in matrix_element.get('diagrams'):
             ###print('DIAGRAM %3d: #wavefunctions=%3d, #diagrams=%3d' %
@@ -1803,71 +1816,51 @@ def get_external(self, wf, argument):
     def get_external_line(self, wf, argument):
         call = ''
         call = call + helas_call_writers.HelasCallWriter.mother_dict[\
-                argument.get_spin_state_number()].lower()
-        if wf.get('mass').lower() != 'zero' or argument.get('spin') != 2:
-            # Fill out with X up to 6 positions
-            call = call + 'x' * (6 - len(call))
-            # Specify namespace for Helas calls
-            call = call + '( momenta,'
-            if argument.get('spin') != 1:
-                # For non-scalars, need mass and helicity
-                call = call + 'm_pars->%s, cHel[ihel][%d],'
-            else:
-                # AV This seems to be for scalars (spin==1???), pass neither mass nor helicity (#351)
-                ###call = call + 'm_pars->%s,'
-                call = call
-            call = call + '%+d, w_sv[%d], %d );'
-            if argument.get('spin') == 1:
-                # AV This seems to be for scalars (spin==1???), pass neither mass nor helicity (#351)
-                return call % \
-                                (
-                                 ###wf.get('mass'),
-                                 # For boson, need initial/final here
-                                 (-1) ** (wf.get('state') == 'initial'),
-                                 wf.get('me_id')-1,
-                                 wf.get('number_external')-1)
-            elif argument.is_boson():
-                misc.sprint(call)
-                misc.sprint( (wf.get('mass'),
-                                 wf.get('number_external')-1,
-                                 # For boson, need initial/final here
-                                 (-1) ** (wf.get('state') == 'initial'),
-                                 wf.get('me_id')-1,
-                                 wf.get('number_external')-1))
-                return  self.format_coupling(call % \
-                                (wf.get('mass'),
-                                 wf.get('number_external')-1,
-                                 # For boson, need initial/final here
-                                 (-1) ** (wf.get('state') == 'initial'),
-                                 wf.get('me_id')-1,
-                                 wf.get('number_external')-1))
-            else:
-                return self.format_coupling(call % \
-                                (wf.get('mass'),
-                                 wf.get('number_external')-1,
-                                 # For fermions, need particle/antiparticle
-                                 - (-1) ** wf.get_with_flow('is_part'),
-                                 wf.get('me_id')-1,
-                                 wf.get('number_external')-1))
+            argument.get_spin_state_number()].lower()
+        # Fill out with X up to 6 positions
+        call = call + 'x' * (6 - len(call))
+        # Specify namespace for Helas calls
+        call = call + '( momenta,'
+        if argument.get('spin') != 1:
+            # For non-scalars, need mass and helicity
+            call = call + 'm_pars->%s, cHel[ihel][%d],'
+        else:
+            # AV This seems to be for scalars (spin==1???), pass neither mass nor helicity (#351)
+            ###call = call + 'm_pars->%s,'
+            call = call
+        call = call + '%+d, w_sv[%d], %d );'
+        if argument.get('spin') == 1:
+            # AV This seems to be for scalars (spin==1???), pass neither mass nor helicity (#351)
+            return call % \
+                            (
+                                ###wf.get('mass'),
+                                # For boson, need initial/final here
+                                (-1) ** (wf.get('state') == 'initial'),
+                                wf.get('me_id')-1,
+                                wf.get('number_external')-1)
+        elif argument.is_boson():
+#                misc.sprint(call)
+#                misc.sprint( (wf.get('mass'),
+#                                 wf.get('number_external')-1,
+#                                 # For boson, need initial/final here
+#                                 (-1) ** (wf.get('state') == 'initial'),
+#                                 wf.get('me_id')-1,
+#                                 wf.get('number_external')-1))
+            return  self.format_coupling(call % \
+                            (wf.get('mass'),
+                                wf.get('number_external')-1,
+                                # For boson, need initial/final here
+                                (-1) ** (wf.get('state') == 'initial'),
+                                wf.get('me_id')-1,
+                                wf.get('number_external')-1))
         else:
-            if wf.get('number_external') == 1:
-                call += 'pz'
-            elif wf.get('number_external') == 2:
-                call += 'mz'
-            else:
-                call += 'xz'
-                comment = '' # AV
-            call = call + 'x' * (6 - len(call))
-            if wf.get('number_external') == 1 or wf.get('number_external') == 2: # AV
-                comment = ' // NB: ' + call + ' only uses pz' # AV skip '(not E,px,py)' to avoid interference with comma parsing in get_external
-            # Specify namespace for Helas calls
-            call = call + '( momenta, cHel[ihel][%d], %+d, w_sv[%d], %d );' + comment # AV vectorize and add comment
             return self.format_coupling(call % \
-                                (wf.get('number_external')-1,
-                                 # For fermions, need particle/antiparticle
-                                 - (-1) ** wf.get_with_flow('is_part'),
-                                 wf.get('me_id')-1,
-                                 wf.get('number_external')-1))
+                            (wf.get('mass'),
+                                wf.get('number_external')-1,
+                                # For fermions, need particle/antiparticle
+                                - (-1) ** wf.get_with_flow('is_part'),
+                                wf.get('me_id')-1,
+                                wf.get('number_external')-1))
 
     # AV - replace helas_call_writers.GPUFOHelasCallWriter method (vectorize w_sv and amp_sv)
     def generate_helas_call(self, argument):
@@ -1931,7 +1924,8 @@ def generate_helas_call(self, argument):
             # AV FOR PR #434: determine if this call needs aS-dependent or aS-independent parameters
             usesdepcoupl = None
             for coup in argument.get('coupling'):
-                if coup.startswith('-'): coup = coup[1:]
+                if coup.startswith('-'): 
+                    coup = coup[1:]
                 # Use the same implementation as in UFOModelConverterCPP.prepare_couplings (assume self.model is the same)
                 for key, coup_list in self.get('model')['couplings'].items():
                     if coup in coup_list:
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/output.py b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/output.py
index 14522ebe14..53cbf7f621 100644
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/output.py
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/output.py
@@ -4,6 +4,7 @@
 # Further modified by: O. Mattelaer, S. Roiser, A. Valassi, Z. Wettersten (2021-2023) for the MG5aMC CUDACPP plugin.
 
 import os
+import subprocess
 
 # AV - load an independent 2nd copy of the export_cpp module (as PLUGIN_export_cpp) and use that within the plugin (workaround for #341)
 # See https://stackoverflow.com/a/11285504
@@ -34,6 +35,7 @@
 
 from os.path import join as pjoin
 import madgraph.various.misc as misc
+import madgraph.iolibs.files as files
 
 # AV - define the plugin's process exporter
 # (NB: this is the plugin's main class, enabled in the new_output dictionary in __init__.py)
@@ -201,13 +203,24 @@ def finalize(self, matrix_element, cmdhistory, MG5options, outputflag):
 	    cmdhistory is the list of command used so far.
 	    MG5options are all the options of the main interface
 	    outputflags is a list of options provided when doing the output command"""
-        misc.sprint('Entering PLUGIN_ProcessExporter.finalize', self.in_madevent_mode)
+        misc.sprint('Entering PLUGIN_ProcessExporter.finalize', self.in_madevent_mode, type(self))
         if self.in_madevent_mode:
             self.add_input_for_banner()
             if 'CUDACPP_CODEGEN_PATCHLEVEL' in os.environ: patchlevel = os.environ['CUDACPP_CODEGEN_PATCHLEVEL']
             else: patchlevel = ''
-            if os.system(PLUGINDIR + os.sep + 'patchMad.sh ' + self.dir_path + ' PROD ' + patchlevel) != 0:
+            plugin_path = os.path.dirname(os.path.realpath( __file__ ))
+#            path = os.path.realpath(os.curdir + os.sep + 'PLUGIN' + os.sep + 'CUDACPP_OUTPUT')
+#            misc.sprint(path)
+            p = subprocess.Popen([pjoin(plugin_path, 'patchMad.sh'), self.dir_path , 'PROD', str(patchlevel)],
+                                 stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+            stdout, stderr = p.communicate()
+            if p.returncode != 0:
+                logger.debug("####### \n stdout is \n %s", stdout)
+                logger.info("####### \n stderr is \n %s", stderr)
+                logger.info("return code is %s\n", p.returncode)
                 raise Exception('ERROR! the O/S call to patchMad.sh failed')
+            
+            self.add_madevent_plugin_fct()
         return super().finalize(matrix_element, cmdhistory, MG5options, outputflag)
 
     # AV (default from OM's tutorial) - overload settings and add a debug printout
@@ -228,4 +241,55 @@ def add_input_for_banner(self):
         for entry in new_parameters:
             finput.write(entry)
 
+    # OM adding a new way to "patch" python file such that the launch command of MG5aMC is working
+    # this consist in a file plugin_interface.py
+    # which contains a series of functions and one dictionary variable TO_OVERWRITE
+    # that will be used to have temporary overwrite of all the key variable passed as string by their value.
+    # all variable that are file related should be called as madgraph.dir.file.variable
+    def add_madevent_plugin_fct(self):
+        """this consist in a file plugin_interface.py
+        which contains a series of functions and one dictionary variable TO_OVERWRITE
+        that will be used to have temporary overwrite of all the key variable passed as string by their value.
+        all variable that are file related should be called as madgraph.dir.file.variable
+        """
+        
+        plugin_path = os.path.dirname(os.path.realpath( __file__ ))
+        files.cp(pjoin(plugin_path, 'plugin_interface.py'), pjoin(self.dir_path, 'bin', 'internal'))
+        files.cp(pjoin(plugin_path, 'launch_plugin.py'), pjoin(self.dir_path, 'bin', 'internal'))
+        files.ln( pjoin(self.dir_path, 'lib'),  pjoin(self.dir_path, 'SubProcesses'))
+
+
 #------------------------------------------------------------------------------------
+class SIMD_ProcessExporter(PLUGIN_ProcessExporter):
+
+    def change_output_args(args, cmd):
+        """ """
+        cmd._export_format = "madevent"
+        args.append('--hel_recycling=False')
+        args.append('--me_exporter=standalone_simd')
+        if 'vector_size' not in ''.join(args):
+            args.append('--vector_size=16')
+        return args
+
+        
+    
+class GPU_ProcessExporter(PLUGIN_ProcessExporter):
+    
+    def change_output_args(args, cmd):
+        """ """
+        cmd._export_format = "madevent"
+        args.append('--hel_recycling=False')
+        args.append('--me_exporter=standalone_cuda')
+        if 'vector_size' not in ''.join(args):
+            args.append('--vector_size=16384')
+        return args
+        
+    def finalize(self, matrix_element, cmdhistory, MG5options, outputflag):
+
+        misc.sprint("enter dedicated function")
+        out = super().finalize(matrix_element, cmdhistory, MG5options, outputflag)
+        #change RunCard class to have default for GPU
+        text = open(pjoin(self.dir_path, 'bin', 'internal', 'launch_plugin.py'), 'r').read()
+        text = text.replace('RunCard = CPPRunCard', 'RunCard = GPURunCard')
+        open(pjoin(self.dir_path, 'bin', 'internal', 'launch_plugin.py'), 'w').write(text)
+        return out
diff --git a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/patchMad.sh b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/patchMad.sh
index 9a9bfd55cc..548a7fdcc8 100755
--- a/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/patchMad.sh
+++ b/epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/patchMad.sh
@@ -41,53 +41,43 @@ dir_patches=$2
 if [ ! -e ${dir} ]; then echo "ERROR! Directory $dir does not exist"; exit 1; fi
 
 # These two steps are part of "cd Source; make" but they actually are code-generating steps
-${dir}/bin/madevent treatcards run
-${dir}/bin/madevent treatcards param
+#${dir}/bin/madevent treatcards run
+#${dir}/bin/madevent treatcards param
 
 # Cleanup
-\rm -f ${dir}/crossx.html
-\rm -f ${dir}/index.html
-\rm -f ${dir}/madevent.tar.gz
-\rm -f ${dir}/Cards/delphes_trigger.dat
-\rm -f ${dir}/Cards/plot_card.dat
-\rm -f ${dir}/bin/internal/run_plot*
-\rm -f ${dir}/HTML/*
-\rm -rf ${dir}/bin/internal/__pycache__
-\rm -rf ${dir}/bin/internal/ufomodel/__pycache__
-touch ${dir}/HTML/.keep # new file
+#\rm -f ${dir}/crossx.html
+#\rm -f ${dir}/index.html
+#\rm -f ${dir}/madevent.tar.gz
+#\rm -f ${dir}/Cards/delphes_trigger.dat
+#\rm -f ${dir}/Cards/plot_card.dat
+#\rm -f ${dir}/bin/internal/run_plot*
+#\rm -f ${dir}/HTML/*
+#\rm -rf ${dir}/bin/internal/__pycache__
+#\rm -rf ${dir}/bin/internal/ufomodel/__pycache__
+#touch ${dir}/HTML/.keep # new file
 
 # Exit here for patchlevel 0 (--upstream)
 if [ "${patchlevel}" == "0" ]; then exit $status; fi
 
 # Add global flag '-O3 -ffast-math -fbounds-check' as in previous gridpacks
-echo "GLOBAL_FLAG=-O3 -ffast-math -fbounds-check" > ${dir}/Source/make_opts.new
-cat ${dir}/Source/make_opts >> ${dir}/Source/make_opts.new
-\mv ${dir}/Source/make_opts.new ${dir}/Source/make_opts
+#echo "GLOBAL_FLAG=-O3 -ffast-math -fbounds-check" > ${dir}/Source/make_opts.new
+#cat ${dir}/Source/make_opts >> ${dir}/Source/make_opts.new
+#\mv ${dir}/Source/make_opts.new ${dir}/Source/make_opts
 
 # Patch the default Fortran code to provide the integration with the cudacpp plugin
 # (1) Process-independent patches
 touch ${dir}/Events/.keep # this file should already be present (mg5amcnlo copies it from Template/LO/Events/.keep) 
-\cp -dpr ${scrdir}/MG5aMC_patches/${dir_patches}/fbridge_common.inc ${dir}/SubProcesses # new file
-\cp -dpr ${scrdir}/MG5aMC_patches/${dir_patches}/{counters.cc,ompnumthreads.cc} ${dir}/SubProcesses/
-sed -i 's/2  = sde_strategy/1  = sde_strategy/' ${dir}/Cards/run_card.dat # use strategy SDE=1 in multichannel mode (see #419)
-sed -i 's/SDE_STRAT = 2/SDE_STRAT = 1/' ${dir}/Source/run_card.inc # use strategy SDE=1 in multichannel mode (see #419)
+\cp -pr ${scrdir}/MG5aMC_patches/${dir_patches}/fbridge_common.inc ${dir}/SubProcesses # new file
+\cp -pr ${scrdir}/MG5aMC_patches/${dir_patches}/{counters.cc,ompnumthreads.cc} ${dir}/SubProcesses/
+#sed -i 's/2  = sde_strategy/1  = sde_strategy/' ${dir}/Cards/run_card.dat # use strategy SDE=1 in multichannel mode (see #419)
+#sed -i 's/SDE_STRAT = 2/SDE_STRAT = 1/' ${dir}/Source/run_card.inc # use strategy SDE=1 in multichannel mode (see #419)
 if [ "${patchlevel}" == "2" ]; then
   cd ${dir}
-  sed -i 's/DEFAULT_F2PY_COMPILER=f2py3.*/DEFAULT_F2PY_COMPILER=f2py3/' Source/make_opts
+  #sed -i 's/DEFAULT_F2PY_COMPILER=f2py3.*/DEFAULT_F2PY_COMPILER=f2py3/' Source/make_opts
   echo "DEBUG: cd ${PWD}; patch -p4 -i ${scrdir}/MG5aMC_patches/${dir_patches}/patch.common"
   if ! patch -p4 -i ${scrdir}/MG5aMC_patches/${dir_patches}/patch.common; then status=1; fi  
   \rm -f Source/*.orig
   \rm -f bin/internal/*.orig
-  echo "
-#*********************************************************************
-# Options for the cudacpp plugin
-#*********************************************************************
-
-# Set cudacpp-specific values of non-cudacpp-specific options
--O3 -ffast-math -fbounds-check = global_flag ! build flags for Fortran code (for a fair comparison to cudacpp)
-
-# New cudacpp-specific options (default values are defined in banner.py)
-CPP = cudacpp_backend ! valid backends are FORTRAN, CPP, CUDA" >> Cards/run_card.dat
   cd - > /dev/null
 fi
 for p1dir in ${dir}/SubProcesses/P*; do
diff --git a/epochX/kokkos/ee_mumu.sa/SubProcesses/P1_Sigma_sm_epem_mupmum/Makefile b/epochX/kokkos/ee_mumu.sa/SubProcesses/P1_Sigma_sm_epem_mupmum/Makefile
deleted file mode 120000
index d0b0e8e008..0000000000
--- a/epochX/kokkos/ee_mumu.sa/SubProcesses/P1_Sigma_sm_epem_mupmum/Makefile
+++ /dev/null
@@ -1 +0,0 @@
-../Makefile
\ No newline at end of file
diff --git a/epochX/kokkos/gg_tt.sa/SubProcesses/P1_Sigma_sm_gg_ttx/Makefile b/epochX/kokkos/gg_tt.sa/SubProcesses/P1_Sigma_sm_gg_ttx/Makefile
deleted file mode 120000
index d0b0e8e008..0000000000
--- a/epochX/kokkos/gg_tt.sa/SubProcesses/P1_Sigma_sm_gg_ttx/Makefile
+++ /dev/null
@@ -1 +0,0 @@
-../Makefile
\ No newline at end of file
diff --git a/epochX/kokkos/gg_ttg.sa/SubProcesses/P1_Sigma_sm_gg_ttxg/Makefile b/epochX/kokkos/gg_ttg.sa/SubProcesses/P1_Sigma_sm_gg_ttxg/Makefile
deleted file mode 120000
index d0b0e8e008..0000000000
--- a/epochX/kokkos/gg_ttg.sa/SubProcesses/P1_Sigma_sm_gg_ttxg/Makefile
+++ /dev/null
@@ -1 +0,0 @@
-../Makefile
\ No newline at end of file
diff --git a/epochX/kokkos/gg_ttgg.sa/SubProcesses/P1_Sigma_sm_gg_ttxgg/Makefile b/epochX/kokkos/gg_ttgg.sa/SubProcesses/P1_Sigma_sm_gg_ttxgg/Makefile
deleted file mode 120000
index d0b0e8e008..0000000000
--- a/epochX/kokkos/gg_ttgg.sa/SubProcesses/P1_Sigma_sm_gg_ttxgg/Makefile
+++ /dev/null
@@ -1 +0,0 @@
-../Makefile
\ No newline at end of file
diff --git a/epochX/kokkos/gg_ttggg.sa/SubProcesses/P1_Sigma_sm_gg_ttxggg/Makefile b/epochX/kokkos/gg_ttggg.sa/SubProcesses/P1_Sigma_sm_gg_ttxggg/Makefile
deleted file mode 120000
index d0b0e8e008..0000000000
--- a/epochX/kokkos/gg_ttggg.sa/SubProcesses/P1_Sigma_sm_gg_ttxggg/Makefile
+++ /dev/null
@@ -1 +0,0 @@
-../Makefile
\ No newline at end of file