Figures 7-11

Figure 7. Speedup Curve for GAMESS(US) Running on the KSR-2. Page faults cause the larger basis calculations to deviate from ideal performance.

Figure 8. Speedup Curve for GAMESS(US) Running on the Intel Paragon. Page faults cause the larger basis calculations to deviate from ideal performance.

Figure 9. Log10-based Timing Comparison for GAMESS(US) Running on the KSR-2, Intel Paragon and CRAY C90. This is a graphical representation of the data contained in Tables 32 and 33.

Figure 10. Wall clock times for direct Hartree-Fock calculations on an SGI PowerChallenge with 75 MHz processors. The IBM 591 times were estimated by assuming a 15% improvement over the model 590. SUPERMOLECULE times are based on runs conducted entirely on the T3D (i.e. nonheterogeneous). The fastest SUPERMOLECULE times were obtained in heterogeneous mode.

Figure 11. Wall clock times for direct Hartree-Fock + gradient calculations on an SGI PowerChallenge with 75 MHz R8000 processors.


Table of Contents.
Document Number: PNNL-10481, Version 3.0
Contact: DF Feller
Webmaster: webmaster@emsl.pnl.gov
Revised: July 31, 1997
Disclaimer