forum@abinit.org
Subject: The ABINIT Users Mailing List ( CLOSED )
List archive
- From: Gilles Zerah <gilles.zerah@cea.fr>
- To: forum@abinit.org
- Subject: Re: (Solved) Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!
- Date: Wed, 16 May 2007 10:55:51 +0200
Dear Masayoshi (and all),
Thanks for the lesson you taught us....
One more, if you have some time: do you know exactly the routine(s) whose time increase (nonlop is made of many subroutines
like opernl etc..) when you change the optimization option?
Thanks
Gilles
Dear all,
Sorry for the series of the posts...
And very sorry to those who take his/her time
to dwell on this topic; In particular, Marc, I will be pleased
to owe you a bottle of special Japanese "sake" ... (next time !)
The solution : FCFLAGS_OPT="-O3 -tpp2",
whiche enables 10-15 times faster than "-O2"(default)
on Itanium2/ifort. (This is amazing !)
The previous results (the default configuration)
v.5.2.3:- Total wall clock time (s,m,h): 5824.1 97.07 1.618 (-O3 -tpp1)
v.5.2.4:- Total wall clock time (s,m,h): 12559.0 209.32 3.489 (-O2)
v.5.3.0:- Total wall clock time (s,m,h) 13056.1 217.60 3.627 (-O2)
v.5.3.2:- Total wall clock time (s,m,h) 74086.2 1234.77 20.580 (-O2)
v.5.3.3:- Total wall clock time (s,m,h): 74144.5 1235.74 20.596 (-O2)
v.5.3.4:- Total wall clock time (s,m,h): 74040.8 1234.01 20.567 (-O2)
Then, the same test with "-O3 -tpp2"
v.5.2.3:- Total wall clock time (s,m,h): 7652.5 127.54 2.126
v.5.2.4:- Total wall clock time (s,m,h): 7529.7 125.50 2.092
v.5.3.0:- Total wall clock time (s,m,h): 7634.1 127.23 2.121
v.5.3.2:- Total wall clock time (s,m,h): 7662.8 127.71 2.129
v.5.3.3:- Total wall clock time (s,m,h): 7687.2 128.12 2.135
v.5.3.4:- Total wall clock time (s,m,h): 7723.7 128.73 2.145
So, the gap between v.5.2.3 and v.5.2.4 (about twice times),
and the other gap between v.5.3.0 and v.5.3.2 (about 6-7 times)
can be overcome just with "-O3 -tpp2".
My preoccupation came from the experience with v.4.6.x,
where I used attached makefile_macros with "-O2" ...
Oh, yikes !! Pardon !
(Still, don't you think this optimization effect is awesome ?
So, the binary users might want to compile the source tar.gz
with such higher optimization ...)
Thus, I would like to suggest "-O3 -tpp2" as default in configuration
for Itanium2/ifort... Yann, how do you think ?
BTW, speaking of configuration,
knowing about "FCFLAGS_FREEFORM"/"FCFLAGS_FIXEDFORM"
might help for some old computers with old compilers that do not
understand free/fixed format automatically.
Bien a vous,
Masayoshi
begin:vcard fn:Gilles Zerah n:Zerah;Gilles email;internet:gilles.zerah@cea.fr tel;work:(33) 1 69 26 79 01 version:2.1 end:vcard
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, (continued)
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Marc Torrent, 05/09/2007
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Yann Pouillon, 05/09/2007
- Message not available
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Masayoshi Mikami, 05/11/2007
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Marc Torrent, 05/11/2007
- Message not available
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Masayoshi Mikami, 05/11/2007
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Yann Pouillon, 05/11/2007
- (Solved) Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Masayoshi Mikami, 05/14/2007
- Re: (Solved) Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Anglade Pierre-Matthieu, 05/15/2007
- configuration to mix -O2/-O3 for different routines, Masayoshi Mikami, 05/16/2007
- Re: [abinit-forum] configuration to mix -O2/-O3 for different routines, Masayoshi Mikami, 05/16/2007
- Re: (Solved) Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Gilles Zerah, 05/16/2007
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Masayoshi Mikami, 05/11/2007
- Message not available
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Marc Torrent, 05/11/2007
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Masayoshi Mikami, 05/11/2007
- Re: [abinit-forum] v.5.3.x is slower than v.5.2.x ?!, Marc Torrent, 05/09/2007
Archive powered by MHonArc 2.6.16.