Fwd: [Disksim-users] Facing problems in Disksim

Shahrukh Rohinton Tarapore shahrukh at cs.virginia.edu
Sat Mar 3 07:27:10 EST 2007


---------- Forwarded message ----------
From: Shahrukh Rohinton Tarapore <shahrukh at cs.virginia.edu>
Date: Mar 3, 2007 7:25 AM
Subject: Re: [Disksim-users] Facing problems in Disksim
To: shubhada nandarshi <shubhada.nandarshi at gmail.com>

Yes,  because you included "disk0 .. disk17".  See disksim by default
requires only 5 parameters.  If you add more then they must be in multiples
of three (3-tuples).  These 3-tuples are meant to easily change parameters
already defined in the parameter file.  You will probably not want to do
this.

Second the first 5 parameters you are using are not correct.  To run disksim
you need to enter this at the terminal:

diskim <parameter file> <output file> <trace type> <trace file> <synthethic
generation flag>

where, <parameter file> is a file which contains properly formatted
parameters for disksim
<output file> is a file (which may or may not exist) where disksim can dump
the statistics
<trace type> tells disksim what format the trace is in
<trace file> a file containing properly formatted trace of disk requests (in
<trace type> format)
<synthetic generation flag> is set to 1 if you will manually generate the
disk requests.  (this is usually set to zero since you have a trace file)

With all that the proper way to execute disksim would be:

*disksim cheetah9LP.parv cheetah9LP.out validate cheetah9LP.trace  0

This will execute disksim with the cheetah9LP parameter file, and the
statistics can be found in a file called cheetah9LP.out when the simulation
is complete, and the simulation executed on a trace found in
cheetah9LP.trace which is formatted in the validate type.  Finally no
synthetic generation was used.

Once the simulation is complete go and view cheetah9LP.out to see the
results of the simulation.


-- shahrukh
***

On 3/3/07, shubhada nandarshi <shubhada.nandarshi at gmail.com> wrote:
>
> yes we tried but we are facing a comment like"*parameter file overrides
> must be 3-tuples" *
>
> in command prompt we are using the follwing command :
>
> "*disksim cheetah9LP.parv cheetah9LP.runvalid valid/ascii.trace
> src/disksim_global.h  0 "disk0 .. disk17""
>
> *
> On 3/1/07, Shahrukh Rohinton Tarapore <shahrukh at cs.virginia.edu > wrote:
> >
> > Checkout this database of disksim parameter files some commonly modelled
> > disks.  Its unlikely you would want to write an entirely new parameter file,
> > so if there are just a few parameters you want to adjust I would simply
> > duplicate one of these parameter files and make the appropriate changes.  If
> > there is a specific disk you want to create a parameter file for you can try
> > using DIXtrac, I'm not familiar with it but it basically extracts the
> > parameters of a real disk which you can use to create a config file for
> > disksim.
> >
> > As for the output file, you do not write these.  The output file simply
> > contains the statistics of the execution of the simulator on a given
> > parameter file.  The output file gives you the results of your simulation,
> > so disksim will be the only one that writes to this file.
> >
> > Earlier you asked how to run disksim, there is an example in the
> > documentation in section 2.2.
> >
> > Hope this helps.
> >
> > -- shahrukh
> >
> > On 3/1/07, shubhada nandarshi < shubhada.nandarshi at gmail.com> wrote:
> > >
> > >  Hi,
> > > Is there any available code which we can refer for getting more
> > > idea about 'How to write a parameter file and output file' , so that we will
> > > be able to write a code for that.
> > >
> > > Thanks & Regards,
> > > Shubhada/Nilofer
> > >
> > >
> > > _______________________________________________
> > > Disksim-users mailing list
> > > Disksim-users at ece.cmu.edu
> > > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users
> > >
> > >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.andrew.cmu.edu/pipermail/disksim-users/attachments/20070303/5b7e9994/attachment.html>


More information about the Disksim-users mailing list