[parsec-users] Cross Compiling Dedup for Alpha Arch using M5 Simulator

Christian A Bienia cbienia at CS.Princeton.EDU
Wed Aug 5 20:56:41 EDT 2009


Hi Ef,

In addition to all the (significant) data structures of fixed size, dedup needs a multiple of the amount of input data to do it's work (uncompressed input data, compressed version of input data, SHA1 sums for chunks, meta information, etc). We hope to streamline the program in future versions of PARSEC to make it a little more forgiving, but keep in mind that dedup is a commercial workloads which always need huge amounts of memory and disk space so it'll never be happy with limited resources.

- Chris


----- Original Message -----
From: "ef" <snorlaxgb at gmail.com>
To: "PARSEC Users" <parsec-users at lists.cs.princeton.edu>
Sent: Wednesday, August 5, 2009 7:17:03 PM GMT -05:00 US/Canada Eastern
Subject: Re: [parsec-users] Cross Compiling Dedup for Alpha Arch using M5 Simulator


I was able to allocate 4GB of Memory to run Dedup_small. Everything went smoothly, now I am testing sim_medium then sim_large... 
Question: 
Can anyone explain the reasoning of why Dedup 1.0 was ok running with 512MB and now Dedup 2.0 requires around ~4GB? Is it scalability, efficiency, etc...? 

Thanks 
EF 

_______________________________________________
parsec-users mailing list
parsec-users at lists.cs.princeton.edu
https://lists.cs.princeton.edu/mailman/listinfo/parsec-users


More information about the parsec-users mailing list