[parsec-users] Default value for options

Mahmood Naderan nt_mahmood at yahoo.com
Mon Jul 26 15:32:41 EDT 2010


Thanks for your reply. It was useful.
 
// Naderan *Mahmood;




________________________________
From: Christian Bienia <cbienia at CS.Princeton.EDU>
To: PARSEC Users <parsec-users at lists.cs.princeton.edu>
Sent: Mon, July 26, 2010 9:10:34 PM
Subject: Re: [parsec-users] Default value for options


You should always selected the largest input set that is realistically possible 
to execute for your methodology. For most simulators that is simlarge. The frame 
option in fluidanimate determines how many frames the program is to compute, it 
doesn’t affect the framerate at all.
 
To emphasize it one more time, there are usually no default values for any of 
the input values. The range of what is possible and makes sense is typically 
very large. In some cases there are standards which can be considered as a 
default, but even then you often have multiple standards to choose from which 
co-exist in practice.
 
- Chris
 
 
From:parsec-users-bounces at lists.cs.princeton.edu 
[mailto:parsec-users-bounces at lists.cs.princeton.edu] On Behalf Of Mahmood 
Naderan
Sent: Monday, July 26, 2010 11:47 AM
To: PARSEC Users
Subject: Re: [parsec-users] Default value for options
 
I know that. In many papers I saw that simlarge is selected. I can also select 
simlarge. What I mean are the options. For example in fluidanimate, with 
simlarge we have many frame rates (25, 30, 50, ...). Is there a default value 
for frame rate? if not can I choose 30 frames because it is more realistic?
 
Regards, 
// Naderan *Mahmood;
 
 

________________________________

From:Christian Bienia <cbienia at CS.Princeton.EDU>
To: PARSEC Users <parsec-users at lists.cs.princeton.edu>
Sent: Mon, July 26, 2010 7:58:43 PM
Subject: Re: [parsec-users] Default value for options
No, you must assume that all types of input affect your simulation. The 
simulation inputs are something like a compromise between short simulation time 
and realistic behavior.
 
The choice of an input for a benchmark can have as much impact on the simulation 
results as the choice of the benchmark itself.
 
- Chris
 
 
From:parsec-users-bounces at lists.cs.princeton.edu 
[mailto:parsec-users-bounces at lists.cs.princeton.edu] On Behalf Of Mahmood 
Naderan
Sent: Monday, July 26, 2010 10:54 AM
To: PARSEC Users
Subject: Re: [parsec-users] Default value for options
 
>The input sets are the closest thing to something like a default
 
You mean simlarge, simmedium and simsmall? If you mean that, then I will assume 
that options does not affect the simulation. Am I right?
 
// Naderan *Mahmood;
 
 

________________________________

From:Christian Bienia <cbienia at CS.Princeton.EDU>
To: PARSEC Users <parsec-users at lists.cs.princeton.edu>
Sent: Mon, July 26, 2010 7:20:37 PM
Subject: Re: [parsec-users] Default value for options
Hi,
 
There are no default options. The input sets are the closest thing to something 
like a default. They were picked to approximate real-world settings as much as 
possible.
 
- Chris
 
 
From:parsec-users-bounces at lists.cs.princeton.edu 
[mailto:parsec-users-bounces at lists.cs.princeton.edu] On Behalf Of Mahmood 
Naderan
Sent: Monday, July 26, 2010 4:08 AM
To: PARSEC
Subject: [parsec-users] Default value for options
 
Hi,
Is there any default values for selecting command line options rather than data 
set (simlarge, simmedium, simsmall). For example in Freqmine, I don't know what 
is the recommended value for <MINSUP>. As another example, in Canneal, we have 
<TEMP> and <NUMBER OF SWAPS>. Fluidanimate also has <FRAMENUMBER>. 

 
Thanks,
// Naderan *Mahmood;


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/parsec-users/attachments/20100726/f9172163/attachment.html>


More information about the parsec-users mailing list