bench_ibm_16

IBM POWER3

Machine Description


					Model Number	POWER3 SMP High Node, 16-way
					Number of Processors	16
					Processor Type	POWER 630 @ 375 MHz
					Operating System	AIX 4.3

Problem and Results

Date Started	10/19/2000	07/11/2000	07/13/2000
Code Used	Example 11	Example 11	Example 11
Number of Equations	15,000	50,000	100,000
Number of Vectors	1	1	1
Matrix Sparsity	Full	Full	Full
Matrix Symmetry	Nonsymmetric with Pivoting	Nonsymmetric with Pivoting	Nonsymmetric with Pivoting
Data Type	16-byte Complex	16-byte Complex	16-byte Complex
Memory for FMS(Gbytes)	8	8	8
CPU Time (Day::Hr:Min:Sec)	2:02:05	3::03:25:03	24::20:02:01
Wall Time (Day::Hr:Min:Sec)	8:06	5:08:26	1::19:02:45
I/O Wait Time (Hr:Min:Sec)	Incore	18:08	3:35:51
Megaflops per Processor	1,228	1,227	1,242
Overall Megaflops	18,537	18,013	17,208
Parellel Speedup	15.09 out of 16	14.67 out of 16	13.85 out of 16

NOTES:

For the out-of-core benchmarks, AIX parameters were not optimized for FMS, resulting in a large number of page faults. There page faults between FMS CPU and I/O threads resulted in larger than normal I/O wait times.

CPU Time is the total amount of User and System time used by all processors.

Wall Time is the elapsed time measured on a dedicated machine. This includes the time spent processing and any time spent waiting for I/O to complete.

I/O Wait Time is the total time spent waiting for I/O to complete that is not overlapped by asynchronous I/O.

Megaflops per Processor is the theoretical number of floating point operations performed, divided by the CPU time.

Overall Megaflops is the total number of floating point operations performed, divided by the Wall Time.

Parallel Speedup is overall increase in performance due to parallel processing.

Copyright © 2000 Multipath Corporation