[SATLUG] question of the measurement of processing power past and present

Brad Knowles brad at shub-internet.org
Tue Aug 12 21:41:52 CDT 2008


On 8/12/08, Todd W. Bucy wrote:

>                        I guess what I am looking for is something that
>  illustrates the dramatic increase of computing power over the last 50 or
>  so years.

CPU power really isn't relevant.  I/O capacity is always what has 
differentiated the mainframes from the PCs.  Same here.  You can't 
just compare integer or floating point processing.

Different systems for different types of jobs.  There's no sense 
trying to compare a Zweihander to surgical shears, even though 
they're both types of cutting instruments.

>  second question
>  is it generally true that the ability to code software (at a theoretical
>  level at least) has always outstripped the ability of the hardware to
>  process such software?

No.  Certainly not in the case of parallel systems.  It is 
unbelievably hard to program these systems, and the systems generally 
go mostly unused, because the software just can't make good enough 
use of the system.

And yes, this rule most definitely applies to all modern cluster-type 
supercomputers.  Some software is parallel-ized enough that it can 
make good use of the whole system, but most isn't.

-- 
Brad Knowles <brad at shub-internet.org>
LinkedIn Profile: <http://tinyurl.com/y8kpxu>


More information about the SATLUG mailing list