Got unpreconditioned conjugate gradient to run and converge on a random (uniform random,

not even SU(3) for now) gauge field. Convergence history is correctly indepdendent of decomposition
on 1,2,4,8,16 mpi tasks.
Found a couple of simd bugs which required fixed and enhanced the Grid_simd.cc test suite.
Implemented the Mdag, M, MdagM, Meooe Mooee schur type stuff in the wilson dop.

benchmarks/Grid_wilson_cg_unprec.cc

@@ -0,0 +1,56 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+  std::vector<int> seeds({1,2,3,4});
+  GridParallelRNG          pRNG(&Grid);  pRNG.SeedFixedIntegers(seeds);
+
+  LatticeFermion src(&Grid); random(pRNG,src);
+  std::cout << "src norm" << norm2(src)<<std::endl;
+  LatticeFermion result(&Grid); result=zero;
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,&Grid);
+
+  double volume=1;
+  for(int mu=0;mu<Nd;mu++){
+    volume=volume*latt_size[mu];
+  }  
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.5;
+  WilsonMatrix Dw(Umu,mass);
+  HermitianOperator<WilsonMatrix,LatticeFermion> HermOp(Dw);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOp,src,result);
+
+  
+  Grid_finalize();
+}