[parsec-users] Size of the output.dat.ddp file in dedup

Christian Bienia cbienia at CS.Princeton.EDU
Wed Dec 3 22:18:20 EST 2008


Hi M. Angeles,

The output of dedup is non-deterministic, just ignore that. To check an
output file for correctness you can use dedup to uncompress the output
archive. It should give you the original input file.

- Chris


-----Original Message-----
From: parsec-users-bounces at lists.cs.princeton.edu
[mailto:parsec-users-bounces at lists.cs.princeton.edu] On Behalf Of Mª Angeles
Gonzalez Navarro
Sent: Wednesday, December 03, 2008 6:20 PM
To: parsec-users at lists.cs.princeton.edu
Subject: [parsec-users] Size of the output.dat.ddp file in dedup

Hi,

I have been testing the parallel implementation of the dedup kernel. I have
noticed that the size of the output file generated by the code
(output.dat.ddp), changes from run to run when the no. of threads is greater
or equal to 2.  I mean, if I fix  the no. of  threads and try several runs,
the size of the output file is always different. In fact, the size tends to
increase when I increase the no. of threads. Is that behavior  acceptable?
By the way, when I execute the serial  version or the parallel version with
n=1, I always get the same size.

Thank you very much in advance,

M.A.

--
--
M. Angeles Gonzalez Navarro
IBM T.J. Watson Research Center, Office 29250c
1101 Kitchawan Road,
Yorktown Heights, NY, 10598
Phone: +1 914 945 1681
http://www.ac.uma.es/~angeles/




More information about the parsec-users mailing list