Alex Granovsky
gran@classic.chem.msu.su
if the number of AOs is large and active space is not very large, the speedup will be really dramatic.
If the number of AOs is not large and active space is large, the speedup will be modest.
And sure, there exists lots of intermediate situations.
I'd suggest you to experimentally check how significant speedup is using your typical input files.
All the best,
Alex Granovsky
On Thu Apr 20 '17 3:49pm, GrEv wrote
------------------------------------
>Hello,
>I wonder what average speed-up of XMCQDPT2 calculations one can expect due to the use of RI? Are there some benchmark calculations.
>Best regards,
>Evgeniy