Abstract
The continuous increment in the power of modern high-performance computing (HPC) platforms has further stimulated the interest of the electronic structure calculations community for more computationally challenging studies.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
For marketing reasons these execution units are sometimes called “cores”, although they are not comparable to a CPU core.
- 2.
Fortunately for programmers, newer GPUs include a small cache. This makes GPU code optimisation much simpler.
- 3.
This has the additional advantage that integer multiplications by the block size can be done using the much faster bit shift operations.
- 4.
Note that it is the block that is transposed with respect to the standard ordering, not the whole set of orbitals.
- 5.
Unrealistic in the sense that the speed-up is larger than the theoretical GPU/CPU performance ratio.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2012 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Andrade, X., Genovese, L. (2012). Harnessing the Power of Graphic Processing Units. In: Marques, M., Maitra, N., Nogueira, F., Gross, E., Rubio, A. (eds) Fundamentals of Time-Dependent Density Functional Theory. Lecture Notes in Physics, vol 837. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-23518-4_21
Download citation
DOI: https://doi.org/10.1007/978-3-642-23518-4_21
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-23517-7
Online ISBN: 978-3-642-23518-4
eBook Packages: Physics and AstronomyPhysics and Astronomy (R0)