188   PetscInt             gid, row, start_slice, cid;  in matmult_seqsell_tiled_kernel8()  local
193 …  gid = gridDim.x * 32 * BLOCKY * iter + blockIdx.x * BLOCKY * 32 + threadIdx.y * 32 + threadIdx.x;  in matmult_seqsell_tiled_kernel8()
194     if (gid < nrows) y[gid] = 0.0;  in matmult_seqsell_tiled_kernel8()
200       gid         = cid * BLOCKY * 32 + threadIdx.y * 32 + threadIdx.x;  in matmult_seqsell_tiled_kernel8()
206         while (gid < totalentries && gid >= sliidx[slice_id + 1]) slice_id++;  in matmult_seqsell_tiled_kernel8()
209         if (row < nrows && gid < totalentries) t = aval[gid] * x[acolidx[gid]];  in matmult_seqsell_tiled_kernel8()
212         if (row < nrows && gid < totalentries && write) atomAdd(y[row], t);  in matmult_seqsell_tiled_kernel8()
216         if (row < nrows) t += aval[gid] * x[acolidx[gid]];  in matmult_seqsell_tiled_kernel8()
243   PetscInt             gid, row, start_slice, cid;  in matmultadd_seqsell_tiled_kernel8()  local
248 …  gid = gridDim.x * 32 * BLOCKY * iter + blockIdx.x * BLOCKY * 32 + threadIdx.y * 32 + threadIdx.x;  in matmultadd_seqsell_tiled_kernel8()
249     if (gid < nrows) z[gid] = y[gid];  in matmultadd_seqsell_tiled_kernel8()
255       gid         = cid * BLOCKY * 32 + threadIdx.y * 32 + threadIdx.x;  in matmultadd_seqsell_tiled_kernel8()
261         while (gid < totalentries && gid >= sliidx[slice_id + 1]) slice_id++;  in matmultadd_seqsell_tiled_kernel8()
264         if (row < nrows && gid < totalentries) t = aval[gid] * x[acolidx[gid]];  in matmultadd_seqsell_tiled_kernel8()
267         if (row < nrows && gid < totalentries && write) atomAdd(z[row], t);  in matmultadd_seqsell_tiled_kernel8()
271         if (row < nrows) t += aval[gid] * x[acolidx[gid]];  in matmultadd_seqsell_tiled_kernel8()