1
0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-19 00:07:05 +01:00

re-work double precision promotion for summit

This commit is contained in:
Christoph Lehner
2020-04-30 16:09:57 -04:00
parent f1fe444d4f
commit ddb192bac7
3 changed files with 90 additions and 6 deletions

View File

@ -234,7 +234,7 @@ innerProductNorm(ComplexD& ip, RealD &nrm, const Lattice<vobj> &left,const Latti
accelerator_for( ss, sites, nsimd,{
auto left_tmp = left_v(ss);
coalescedWrite(inner_tmp_v[ss],innerProduct(left_tmp,right_v(ss)));
coalescedWrite(norm_tmp_v[ss],innerProduct(left_tmp,left_tmp)));
coalescedWrite(norm_tmp_v[ss],innerProduct(left_tmp,left_tmp));
});
tmp[0] = TensorRemove(sumD_gpu(inner_tmp_v,sites));