I have an application to do random reads from a bunch of files and do
simple calculations on AIX 5.2. The very first time it runs, the time
command reports

>time myfoo.tsk
Real 512.35
User 7.46
System 6.31

If it runs again, time command reports

>time myfoo.tsk
Real 12.50
User 5.06
System 3.11

The user and system times look reasonable. But I have no clue what
causes the huge discrepency between the real times. This happens
consistently, not caused by other heavy activities when the
application runs first time. I guess it is a process scheduling
problem. I'd appreciate if you can share your input.

Thanks. Shane

Shane wrote:
> I have an application to do random reads from a bunch of files and do
> simple calculations on AIX 5.2. The very first time it runs, the time
> command reports
>
[SNIP]
Have you considered the effect of disk cacheing on subsequent run times?

Cheers,
Gary B-)

--

Speaking strictly for myself.

In article <40127376.0411100644.49b2c27b@posting.google.com >, xhu1
@bloomberg.net says...
> I have an application to do random reads from a bunch of files and do
> simple calculations on AIX 5.2. The very first time it runs, the time
> command reports
>
> >time myfoo.tsk
> Real 512.35
> User 7.46
> System 6.31
>
> If it runs again, time command reports
>
> >time myfoo.tsk
> Real 12.50
> User 5.06
> System 3.11
>
> The user and system times look reasonable. But I have no clue what
> causes the huge discrepency between the real times. This happens
> consistently, not caused by other heavy activities when the
> application runs first time. I guess it is a process scheduling
> problem. I'd appreciate if you can share your input.
>
> Thanks. Shane
>

How about the third time? And the fourth? Are you sure your i/o is not
getting cached in main memory? And what's a ".tsk" file?

It also could be that the SCSI has been spun-down due ti inactivity
and the first time needs to be brought back up to speed.

On one of my machines, the lag can be large. to the point I thought the
machine had crashed and was about to hit the reset button, just as it sprung
to life.
thanks Nigel

"Nigel Griffiths" <nag@uk.ibm.com> wrote in
news:cn2mvk$blr$1@sparta.btinternet.com:

> It also could be that the SCSI has been spun-down due ti inactivity
> and the first time needs to be brought back up to speed.
>
> On one of my machines, the lag can be large. to the point I thought
> the machine had crashed and was about to hit the reset button, just as
> it sprung to life.
> thanks Nigel

Just disable the power management. It has no place in a server.

--
Doing AIX support was the most monty-pythonesque
activity available at the time.
Eagerly awaiting my thin chocolat mint.