[klas3] vectorize the final calcualations in sigmakin (issue #178)

This squeezes one (last?) % of performance. Now 4.91E6, was 4.87E6.

On itscrd70.cern.ch:
-------------------------------------------------------------------------
Process                     = EPOCH1_EEMUMU_CPP [gcc (GCC) 9.2.0]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
Internal loops fptype_sv    = SCALAR ('none': ~vector[1], no SIMD)
OMP threads / `nproc --all` = 1 / 4
EvtsPerSec[MatrixElems] (3) = ( 1.307604e+06                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     7.176768 sec
real    0m7.186s
=Symbols in CPPProcess.o= (~sse4:  614) (avx2:    0) (512y:    0) (512z:    0)
-------------------------------------------------------------------------
Process                     = EPOCH1_EEMUMU_CUDA [nvcc 11.0.221]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
EvtsPerSec[MatrixElems] (3) = ( 7.341151e+08                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     0.731415 sec
real    0m1.031s
==PROF== Profiling "sigmaKin": launch__registers_per_thread 120
-------------------------------------------------------------------------
Process                     = EPOCH1_EEMUMU_CPP [gcc (GCC) 9.2.0]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
Internal loops fptype_sv    = VECTOR[2] ('sse4': SSE4.2, 128bit) [cxtype_ref=YES]
OMP threads / `nproc --all` = 1 / 4
EvtsPerSec[MatrixElems] (3) = ( 2.500771e+06                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     4.879965 sec
real    0m4.890s
=Symbols in CPPProcess.o= (~sse4: 3274) (avx2:    0) (512y:    0) (512z:    0)
-------------------------------------------------------------------------
Process                     = EPOCH1_EEMUMU_CPP [gcc (GCC) 9.2.0]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
Internal loops fptype_sv    = VECTOR[4] ('avx2': AVX2, 256bit) [cxtype_ref=YES]
OMP threads / `nproc --all` = 1 / 4
EvtsPerSec[MatrixElems] (3) = ( 4.600406e+06                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     3.653161 sec
real    0m3.663s
=Symbols in CPPProcess.o= (~sse4:    0) (avx2: 2746) (512y:    0) (512z:    0)
-------------------------------------------------------------------------
Process                     = EPOCH1_EEMUMU_CPP [gcc (GCC) 9.2.0]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
Internal loops fptype_sv    = VECTOR[4] ('512y': AVX512, 256bit) [cxtype_ref=YES]
OMP threads / `nproc --all` = 1 / 4
EvtsPerSec[MatrixElems] (3) = ( 4.909408e+06                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     3.581065 sec
real    0m3.591s
=Symbols in CPPProcess.o= (~sse4:    0) (avx2: 2572) (512y:   95) (512z:    0)
-------------------------------------------------------------------------
Process                     = EPOCH1_EEMUMU_CPP [gcc (GCC) 9.2.0]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
Internal loops fptype_sv    = VECTOR[8] ('512z': AVX512, 512bit) [cxtype_ref=YES]
OMP threads / `nproc --all` = 1 / 4
EvtsPerSec[MatrixElems] (3) = ( 3.736380e+06                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     3.995692 sec
real    0m4.006s
=Symbols in CPPProcess.o= (~sse4:    0) (avx2: 1127) (512y:  205) (512z: 2045)
-------------------------------------------------------------------------
Process                     = EPOCH2_EEMUMU_CPP [gcc (GCC) 9.2.0]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
OMP threads / `nproc --all` = 1 / 4
EvtsPerSec[MatrixElems] (3) = ( 1.145752e+06                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     7.854923 sec
real    0m7.865s
=Symbols in CPPProcess.o= (~sse4:  567) (avx2:    0) (512y:    0) (512z:    0)
-------------------------------------------------------------------------
Process                     = EPOCH2_EEMUMU_CUDA [nvcc 11.0.221]
FP precision                = DOUBLE (NaN/abnormal=0, zero=0)
EvtsPerSec[MatrixElems] (3) = ( 7.451168e+08                 )  sec^-1
MeanMatrixElemValue         = ( 1.371706e-02 +- 3.270315e-06 )  GeV^0
TOTAL       :     0.734622 sec
real    0m1.027s
==PROF== Profiling "sigmaKin": launch__registers_per_thread 164
-------------------------------------------------------------------------

epoch1/cuda/ee_mumu/SubProcesses/P1_Sigma_sm_epem_mupmum/CPPProcess.cc

@@ -424,18 +424,22 @@ namespace Proc
 #endif
 
     // PART 0 - INITIALISATION (before calculate_wavefunctions)
+    // Reset the "matrix elements" - running sums of |M|^2 over helicities for the given event
+    // FIXME: assume process.nprocesses == 1 for the moment (eventually: need a loop over processes here?)
+#ifdef MGONGPU_CPPSIMD
+    const int npagV = nevt/neppV;    
+    for ( int ipagV = 0; ipagV < npagV; ++ipagV )
+    {
+      allMEs[ipagV] = fptype_v{0}; // all zeros
+    }
+#else
 #ifndef __CUDACC__
     for ( int ievt = 0; ievt < nevt; ++ievt )
 #endif
     {
-      // Reset the "matrix elements" - running sums of |M|^2 over helicities for the given event
-      // FIXME: assume process.nprocesses == 1 for the moment (eventually: need a loop over processes here?)
-#ifndef MGONGPU_CPPSIMD
       allMEs[ievt] = 0; // all zeros
-#else
-      allMEs[ievt/neppV][ievt%neppV] = 0; // all zeros
-#endif
     }
+#endif
 
     // PART 1 - HELICITY LOOP: CALCULATE WAVEFUNCTIONS
     // (in both CUDA and C++, using precomputed good helicities)
@@ -450,20 +454,23 @@ namespace Proc
     }
 
     // PART 2 - FINALISATION (after calculate_wavefunctions)
+    // Get the final |M|^2 as an average over helicities/colors of running sum of |M|^2 over helicities for the given event
+    // [NB 'sum over final spins, average over initial spins', eg see
+    // https://www.uzh.ch/cmsssl/physik/dam/jcr:2e24b7b1-f4d7-4160-817e-47b13dbf1d7c/Handout_4_2016-UZH.pdf]
+    // FIXME: assume process.nprocesses == 1 for the moment (eventually: need a loop over processes here?)
+#ifdef MGONGPU_CPPSIMD
+    for ( int ipagV = 0; ipagV < npagV; ++ipagV )
+    {
+      allMEs[ipagV] /= denominators;
+    }
+#else
 #ifndef __CUDACC__
     for ( int ievt = 0; ievt < nevt; ++ievt )
 #endif
     {
-      // Get the final |M|^2 as an average over helicities/colors of running sum of |M|^2 over helicities for the given event
-      // [NB 'sum over final spins, average over initial spins', eg see
-      // https://www.uzh.ch/cmsssl/physik/dam/jcr:2e24b7b1-f4d7-4160-817e-47b13dbf1d7c/Handout_4_2016-UZH.pdf]
-      // FIXME: assume process.nprocesses == 1 for the moment (eventually: need a loop over processes here?)
-#ifndef MGONGPU_CPPSIMD
       allMEs[ievt] /= denominators;
-#else
-      allMEs[ievt/neppV][ievt%neppV] /= denominators;
-#endif
     }
+#endif
     mgDebugFinalise();
   }