Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Numerical performance and throughput benchmark for electronic structure calculations in PC-Linux systems with new

Jen-Shiang K Yu1, Jenn-Kang Hwang, Chuan Yi Tang

  • 1Department of Biological Science and Technology, National Chiao Tung University, No. 75, Po-Ai Street, Hsinchu 300, Taiwan. jsyu@mail.nctu.edu.tw

Journal of Chemical Information and Computer Sciences
|March 23, 2004
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Origin of Substituent-Modulated Regioselectivity in Phosphine-Catalyzed [3 + 2] Cyclization of Allenoates and Enones: A Kinetic Shift toward Curtin-Hammett Control.

The Journal of organic chemistry·2025
Same author

sp-Hybridized Seesaw Ge<sup>0</sup> Complexes via Germylone-to-Seesaw Isomerization in a Four-Electron Cyclic N<sub>2</sub>Ge<sub>2</sub> Ligand.

Inorganic chemistry·2025
Same author

A Computational Study of Reactions with Boric Acid Aimed to Promote the Utilization of Lignin.

Chemistry (Weinheim an der Bergstrasse, Germany)·2024
Same author

Catalytic role of the enol ether intermediate in the intramolecular Stetter reaction: a computational perspective.

Physical chemistry chemical physics : PCCP·2024
Same author

Formation of Breslow Intermediates under Aprotic Conditions: A Computational Study.

The Journal of organic chemistry·2022
Same author

Fragmentation and rearrangement of Breslow intermediates: branches to both radical and ionic pathways.

Physical chemistry chemical physics : PCCP·2021

New numerical libraries and updated Fortran compilers significantly boost Gaussian 98 performance. Hardware tuning further enhances memory bandwidth and floating-point throughput on modern architectures like AMD64.

Area of Science:

  • Computational Chemistry
  • High-Performance Computing
  • Software Optimization

Background:

  • Recent numerical libraries like ATLAS, MKL, GOTO, and ACML are now compatible with Gaussian 98.
  • Updated Fortran compilers (Intel 7.1, PGI 5.0) offer performance improvements over older versions.

Purpose of the Study:

  • To evaluate the performance impact of new numerical libraries and Fortran compilers on Gaussian 98.
  • To assess the effectiveness of compiler tuning options and hardware-level optimizations.

Main Methods:

  • Linking Gaussian 98 with various numerical libraries (ATLAS, MKL, GOTO, ACML).
  • Compiling Gaussian 98 using updated Fortran compilers (Intel, PGI) with and without extensive tuning.
  • Measuring floating-point (FP) throughput using simultaneous execution of test jobs.

Related Experiment Videos

  • Evaluating hardware-level tuning of memory bandwidth (DRAM timing).

    • Main Results:

    • Intel Fortran compiler 7.1 shows a ~3% improvement on 32-bit systems over version 6.0.
    • PGI Fortran compiler 5.0 offers ~3% improvement with default settings, and up to 25% with extensive tuning.
    • Optimized numerical libraries show similar performance.
    • SSE2 instruction sets are functional on AMD Opteron in 32-bit mode, with Intel Fortran showing better optimization.
    • Hardware tuning improved memory bandwidth by 2.6% (CL2 vs. CL2.5).
    • IA64 and AMD64 architectures demonstrate significantly higher FP throughput than IA32.

    Conclusions:

    • Updated compilers and numerical libraries provide substantial performance gains for electronic structure calculations.
    • Extensive compiler tuning can yield significant speedups.
    • Modern architectures (IA64, AMD64) offer superior computational throughput compared to older ones (IA32).