[log in to unmask] (Jeff Boehlert) wrote:
>Well speaking of performance measures a 960 is supposed to be a
>40% faster CPU. We recently switched from a 937 to a 960 to gain
>performance. I believe we had the same amount of memory and disk.
>Well as soon as the new system was in place everybody started
>complaining that it was 40 % SLOWER not faster. So obviously we
>shouldn't make decisions on numbers alone, else there is some
>tuning to be done. Sound familiar ?
>
 
People -- there are DOZENS of factors that contribute to performance.
If you keep 2 constant (memory & disk), change 1 (SPU processor), and
don't report on the other factors (configuration, free space, workload,
etc.etc.etc., then, "your mileage may vary".  For example, if my
memory serves me well, the 960 was a CIO (older) machine and the 937
was a NIO (newer) machine -- perhaps that is a contributing factor --
your IO structure differs in the two environments.
 
Without an examination of all the factors that are likely to impact
your performance, it is anyone's guess what's going on.
 
Jon Cohen
HP Commercial Systems Division