Hi Nilay,
I attach 3 sets of config.ini and stats.txt:
1) before_{config.ini,stats.txt}: without changeset 10313
2) after_{config.ini,stats.txt}: with changeset 10313
3) 10365_{config.ini,stats.txt}: gem5 at 10365 changeset, currently most
updated version
Because the files are too big to be sent, here is my dropbox link:
https://www.dropbox.com/sh/6zb4sneadcf57cz/AADpv7s9d8dMIRtTzEQJpZeXa?dl=0
All files are inside.
Thanks.
--
Best Regards
Yan Zi
On 9 Sep 2014, at 10:34, Nilay Vaish via gem5-dev wrote:
Is it possible for you to post the stats.txt and config.ini files from
these three runs you talk about?
--
Nilay
On Tue, 9 Sep 2014, Zi Yan via gem5-dev wrote:
Hi All,
I am running a segment of kmeans benchmark on hadoop in x86.
I find out changeset 10313 (http://reviews.gem5.org/r/2349/) causes
that
the number of committed instructions in x86 O3 CPU increases from
2.1 billion to 3.5 billion.
Here is what I did:
1) I run in gem5 O3 CPU at changeset 10299 plus patches mentioned in
http://www.mail-archive.com/gem5-dev@gem5.org/msg11941.html, and the
benchmark
committed 2.1 billion instructions.
2) afterwards, I apply http://reviews.gem5.org/r/2349/, and the
benchmark
committed 3.5 billion instructions.
Could anyone give some information about this patch? I want to look
into
this problem.
BTW, there is a huge gap between atomic CPU and O3 CPU while
executing
this segment of kmeans benchmark. atomic CPU takes only 340 million
instruction, O3 CPU takes 2.1 billion instructions at changeset
10299.
I also need help with this.
Thanks.
--
Best Regards
Yan Zi
_______________________________________________
gem5-dev mailing list
gem5-dev@gem5.org
http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
gem5-dev@gem5.org
http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
gem5-dev@gem5.org
http://m5sim.org/mailman/listinfo/gem5-dev