From owner-nwchem-users@emsl.pnl.gov Wed Jun 22 12:20:47 2005 Received: from odyssey.emsl.pnl.gov (localhost [127.0.0.1]) by odyssey.emsl.pnl.gov (8.12.10/8.12.10) with ESMTP id j5MJKlUW027935 for ; Wed, 22 Jun 2005 12:20:47 -0700 (PDT) Received: (from majordom@localhost) by odyssey.emsl.pnl.gov (8.12.10/8.12.10/Submit) id j5MJKlxB027934 for nwchem-users-outgoing; Wed, 22 Jun 2005 12:20:47 -0700 (PDT) Date: Wed, 22 Jun 2005 13:20:35 -0600 From: Matthew Asplund Subject: Re: x86_64, integer*8 blas libraries In-reply-to: To: Brian Barnes Cc: nwchem-users@emsl.pnl.gov Message-id: <032ACD5E-F8C4-4C3F-A298-0083894CD63E@byu.edu> MIME-version: 1.0 (Apple Message framework v730) X-Mailer: Apple Mail (2.730) Content-type: multipart/alternative; boundary=Apple-Mail-26--573085019 X-Scanned-By: MIMEDefang 2.42 References: Sender: owner-nwchem-users@emsl.pnl.gov Precedence: bulk --Apple-Mail-26--573085019 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed I just compiled it with IFC and atlas. I read the same thing you did, so I compiled the latest atlas with ifc, and then built nwchem. Dr. Matthew Asplund Department of Chemistry and Biochemistry Brigham Young University Provo, UT 84602 (801) 422-5275 On Jun 22, 2005, at 9:57 AM, Brian Barnes wrote: > hello, > > Our group has recently acquired an AMD Opteron cluster, and I have > been > setting up NWChem 4.7 and testing it. While I have run a > successful test > job using gnu compilers and ATLAS 3.6.0 (also compiled with gnu) > for the > BLAS routines, I am hoping to use the Pathscale f90 compiler to > optimize > speed. > > In the NWChem INSTALL file, it is recommended to _not_ use g77 on > x86_64 > because of NWChem's preference for 64-bit integers. It is also > mentioned > that the common math libraries (ESSL, ACML, and from looking at > some docs, > ATLAS) do not support 64-bit integers. Which leads to my > question: what > BLAS routines have people had success with, for x86_64 64-bit integer > compiles? > > thanks, > > Brian Barnes > Gelb Group > Washington University in St. Louis > --Apple-Mail-26--573085019 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1 I just compiled it with IFC and = atlas.=A0 I read the same thing you did, so I compiled the latest atlas = with ifc, and then built nwchem.=A0=A0



Dr. Matthew Asplund

Department of Chemistry and = Biochemistry

Brigham Young University

Provo, UT 84602

(801) 422-5275


On Jun = 22, 2005, at 9:57 AM, Brian Barnes wrote:

hello,

Our group has recently acquired = an AMD Opteron cluster, and I have been
setting = up NWChem 4.7 and testing it.=A0 = While I have run a successful test
job = using gnu compilers and ATLAS 3.6.0 (also compiled with gnu) for = the
BLAS routines, I am hoping to = use the Pathscale f90 compiler to optimize

In the NWChem INSTALL file, it is recommended to = _not_ use g77 on x86_64
because of = NWChem's preference for 64-bit integers.=A0 It is also = mentioned
that the common math libraries = (ESSL, ACML, and from looking at some docs,
ATLAS) = do not support 64-bit integers.=A0 = Which leads to my question:=A0= what
BLAS routines have people had = success with, for x86_64 64-bit integer

thanks,

Brian Barnes
Gelb Group
Washington = University in St. Louis


= --Apple-Mail-26--573085019--