National Institute of Standards and Technology

Computational Chemistry Comparison and Benchmark DataBase
Release
NIST Standard Reference Database 101

I	Introduction
II	Experimental data
III	Calculated data
IV	Data comparisons
V	Cost comparisons
VI	Input and output files
VII	Tutorials and Units
VIII	Links to other sites
IX	Feedback
X	Older CCCBDB versions
XII	Geometries
XIII	Vibrations
XIV	Reaction data
XV	Entropy data
XVI	Bibliographic data
XVII	Ion data
XVIII	Bad calculations
XIX	Index of properties
XX	H-bond dimers
XXI	Oddities

---

NIST policy on privacy, security, and accessibility.

---

© 2013 copyright by the U.S. Secretary of Commerce on behalf of the United States of America. All rights reserved.

The National Institute of Standards and Technology (NIST) is an agency of the U.S. Department of Commerce.

Please send questions, comments, corrections, additions and suggestions to cccbdb@nist.gov.

V.A.

Cost Comparison

The following is from:
Irikura, K. K.; Frurip, D. J. In "Computational Thermochemistry: Prediction and Estimation of Molecular Thermodynamics" (ACS Symposium Series 677); Irikura, K. K., Frurip, D. J., Eds.; American Chemical Society: Washington, DC, 1998; Chapter 1.

Some Common Methods of Computational Chemistry^a

Method	Applicability		Cost			Reliabilityb
	Type	Size	Software	Computer	Personnel	Accuracyc	Precisiond
Empirical (e.g. Benson Groups)	common organic	any	$	¢	$	A-	B+
Molecular Mechanics (e.g. MM3)	common organic	100000 atoms	$	¢¢¢	$$	A-	B
semi-empirical MO theory (e.g. MNDO/d)	organic, some inorganic	500 atoms	$	$	$$	C+	C-
density functional theory (e.g. B3LYP)	alle	50 atomsf	$$$	$$	$$$	B-	C+
CBS-4 (MO theory)	alle	20 atomsf	$$$	$$$	$$$	B	B-
BAC-MP4 (corr. MO theory)	organic, some inorganic	20 atomsf	$$$	$$$	$$$$	B+	B
PCI-80 (corr. MO theory)	alle	20 atomsf	$$$	$$$	$$$$	B+	B-
G2 (MO theory)	alle	6 heaviesf	$$$	$$$$	$$$	A-	B
CCSD(T) with basis set extrapolation (MO theory)	alle	3 heaviesf	$$$	$$$$$$	$$$$	A	A-

^aThe opinions in this Table are those of the authors and do not necessarily represent those of the authors of the other chapters in this book.
^bDFT methods are more robust than MO methods for molecules that contain transition metals, although both are less accurate for transition metals than for organics.
^cTypical performance for a large set of moelcules.(i.e. mean error)
^dWorst-case performance among a large set of moelcules (i.e. largest error)
^eSubject to availability of basis sets.
^fLight atoms add less than heavy atoms to the computational expense; "heavy" means lithium or heavier.