2017
DOI: 10.1002/jcc.24756
|View full text |Cite
|
Sign up to set email alerts
|

Saving time and energy with oversubscription and semi‐direct M øller– P lesset second order perturbation methods

Abstract: In this work, the effect of oversubscription is evaluated, via calling 2n, 3n, or 4n processes for n physical cores, on semi-direct MP2 energy and gradient calculations and RI-MP2 energy calculations with the cc-pVTZ basis using NWChem. Results indicate that on both Intel and AMD platforms, oversubscription reduces total time to solution on average for semi-direct MP2 energy calculations by 25-45% and reduces total energy consumed by the CPU and DRAM on average by 10-15% on the Intel platform. Semi-direct grad… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

0
4
0

Year Published

2019
2019
2021
2021

Publication Types

Select...
2

Relationship

2
0

Authors

Journals

citations
Cited by 2 publications
(4 citation statements)
references
References 18 publications
0
4
0
Order By: Relevance
“…Similar to execution on a single node, oversubscription results in significant improvements in total MP2 gradient wall times. Figure provides normalized total execution times for all six systems examined using both two and three nodes.…”
Section: Resultsmentioning
confidence: 99%
See 1 more Smart Citation
“…Similar to execution on a single node, oversubscription results in significant improvements in total MP2 gradient wall times. Figure provides normalized total execution times for all six systems examined using both two and three nodes.…”
Section: Resultsmentioning
confidence: 99%
“…In 2017, we reported on the effects of oversubscription on MP2 on a single node within the NWChem software package. MP2 is a popular and relatively inexpensive correlation method and has multiple implementations within NWChem.…”
Section: Introductionmentioning
confidence: 99%
“…An explicit tie to external storage can make it hard to compose/nest calculations (e.g., perform a parameter sweep in parallel). Petascale and exascale platforms will have intermediate storage facilities using nonvolatile random-access memory (NVRAM) or solid-state disks (SSD). Resilience, energy, and power: These exascale concerns are rarely represented in the toolchains of today’s computational chemistry codes. …”
Section: Design Challenges and Principlesmentioning
confidence: 99%
“…Resilience, energy, and power: These exascale concerns are rarely represented in the toolchains of today’s computational chemistry codes.…”
Section: Design Challenges and Principlesmentioning
confidence: 99%