Thanks for you reply Tibor.

>> Try executing both to grid and compare. Perhaps try with a smaller row number and work upwards and

I had done that. I mentioned 20000 in the original post because it
wasn't a huge amount of data and it demonstrated the order of magnatude
speed difference between the tools.

This was a bizarre problem. This had been occurring for a number of days
and then this morning I came into work and the problem was gone. To be
honest I don't know if we'll ever know what caused the (temporary ?)
problem.

Hopefully this one does not return.

Regards
Clint.