From owner-chemistry@ccl.net Mon Sep 12 08:40:10 2005 From: "CCL" To: CCL Subject: CCL: Molpro and memory Message-Id: <-29122-050912045136-16222-DUcROUZqAAanpwYwt3FuqA!^!server.ccl.net> X-Original-From: Grzegorz Mazur Content-Disposition: inline Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="iso-8859-1" Date: Mon, 12 Sep 2005 09:49:09 +0200 MIME-Version: 1.0 Sent to CCL by: Grzegorz Mazur [mazur!^!chemia.uj.edu.pl] On Sunday 11 of September 2005 01:33, CCL wrote: > Sent to CCL by: Kirk Peterson [kipeters{=}wsu.edu] > Perhaps this is due to using gcc, but with the Portland Group compiler we > consistently allocate more than 1 GB of memory (up to 2) with molpro on our > 32-bit athlon cluster (dual node, 3GB RAM, 4GB of swap). Perhaps is the answer? For x86_64 in 64 bit mode the default integer size is 32 bits. Unfortunately this limitation makes g77 kind of useless for amd64. Hopefully gfortran is better, but I haven't yet had time to check. BTW, as far as I understand, the issue with compiling GAMESS with g77 in 64 bit mode has exactly the same origin, although it manifests itself in a completely different way. HTH, Grzesiek -- Grzegorz Mazur http://chemia.uj.edu.pl/~mazur From owner-chemistry@ccl.net Mon Sep 12 09:08:27 2005 From: "CCL" To: CCL Subject: CCL: Harris functional Message-Id: <-29123-050912090004-6290-DUcROUZqAAanpwYwt3FuqA[A]server.ccl.net> X-Original-From: "Telkuni Tsuru" Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="Windows-1252" Date: Mon, 12 Sep 2005 21:13:27 +0900 MIME-Version: 1.0 Sent to CCL by: "Telkuni Tsuru" [telkuni[A]venus.dti.ne.jp] Hello, CCLers. I like to study Harris functional of DFT and like to know the papers which describe it. I've gotten the papers: 1) Cullen J, J. Comput Chem., 25(5):637-48 (2004) 2) S. I. Simak et.al, Solid State Communications, 87 ,Pages 393-396 (1993) 3) K. Kobayashi et.al, Chem. Phys. Lett., 188 181 (1992). 4) N. Chetty et.al, Phys. Rev. B, 46 3798 (1992), If you know other good papers, please teach me. Sincerely yours, ---------------------------------------------------- Telkuni Tsuru telkuni[A]venus.dti.ne.jp From owner-chemistry@ccl.net Mon Sep 12 10:20:37 2005 From: "CCL" To: CCL Subject: CCL: Molpro and memory Message-Id: <-29124-050912101944-31460-DUcROUZqAAanpwYwt3FuqA(!)server.ccl.net> X-Original-From: Alexander Martins Silva Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Mon, 12 Sep 2005 11:24:18 -0300 MIME-Version: 1.0 Sent to CCL by: Alexander Martins Silva [alex.msilva(!)uol.com.br] Hi, "Is your kernel compiled with large memory support? " Yes, it is. You are using 1Gb per processor. It is the same problem. I will compile a paralel version of molpro to try to solve partialy my problem. Thank you, Alexander. CCL wrote: >Sent to CCL by: Kirk Peterson [kipeters{=}wsu.edu] >Perhaps this is due to using gcc, but with the Portland Group compiler we >consistently allocate more than 1 GB of memory (up to 2) with molpro on our >32-bit athlon cluster (dual node, 3GB RAM, 4GB of swap). >Is your kernel compiled with large memory support? > >-Kirk > > >On Fri, Sep 09, 2005 at 11:00:32PM -0400, CCL wrote: > > >>Sent to CCL by: "Igor Filippov [Contr]" [igorf|"|helix.nih.gov] >>I seem to remember reading about 1Gb limit per single process too, >>though I can't find the reference now. This is what Linus had said about >>the issue - according to him, it's more like 3Gb limit, but >>not all of it can be accessible - due to shared libs taking space etc. >>You need to move to 64-bit cpu. >> >>http://www.ussg.iu.edu/hypermail/linux/kernel/9801.2/0194.html >> >>Igor >> >> >>On Fri, 2005-09-09 at 19:04, CCL wrote: >> >> >>>Sent to CCL by: "Ben Swerts" [ben.swerts/./ua.ac.be] >>> >>> >>>Hi Alexander, >>> >>>I have absolutely no experience with Molpro but I can tell you this: >>>When using GCC as the compiler in 32 bit mode, I noticed a limit of 1 GB for >>>statically allocated memory when compiling fortran codes. I tried a lot of >>>things to circumvent this problem but never succeeded. Now we have 64 bit >>>machines and this 1 GB limit is finally a thing of the past. >>>I guess the problem is related to Linux. Applications just crash when >>>compiled with more than 1 GB. >>> >>>Greetz, >>> >>> >>> Ben >>> >>> >>> >>>>-----Original Message----- >>>>From: owner-chemistry(!)ccl.net [mailto:owner-chemistry(!)ccl.net] >>>>Sent: Friday, September 09, 2005 10:44 PM >>>>To: Swerts, Ben >>>>Subject: CCL: Molpro and memory >>>> >>>> >>>>Sent to CCL by: Alexander Martins Silva [alex.msilva---uol.com.br] >>>> >>>> Hi, >>>> >>>> This is my second message about the same issue: >>>>Molpro and memory. I've compiled the Molpro in a Xeon >>>>machine with 2 proc and 4 GB of memory. However, I couldn't >>>>use more than 1Gb of memory when executing jobs. I've >>>>already expanded the kernel.shmall and kernel.shmmax >>>>parameters up to 2Gb and nothing happens. The manual don't >>>>tell me nothing about this subject. Is it a problem related >>>>to the Intel Fotran Compiler? Anybody could help me? >>>> >>>> >>>> Thanks in advance, >>>> >>>> Alexander.> >>>> >>>> >> >> > > > From owner-chemistry@ccl.net Mon Sep 12 12:44:08 2005 From: "CCL" To: CCL Subject: CCL: Dual core processor Message-Id: <-29125-050912104645-10797-DUcROUZqAAanpwYwt3FuqA\a/server.ccl.net> X-Original-From: John McKelvey Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Mon, 12 Sep 2005 09:44:59 -0500 MIME-Version: 1.0 Sent to CCL by: John McKelvey [jmmckel\a/attglobal.net] All, Could I ask a question? The job referred to below is a frequence job; how much IO does this involve? If four heavy processes hit the two cpu's at the same time will the bus be able to handle it? This was a serious problem when the early quad processor boards became available. If there is a lot of IO presumably 10KRPM SATA drives would have a positive impact? Cheers, John McKelvey CCL wrote: >Sent to CCL by: "S.I.Gorelsky" [gorelsky : stanford.edu] > >For the information about the dual-core Opteron performance with G03 >please refer to > > http://www.sg-chem.net/cluster/ > > >Best regards, > >S. Gorelsky > > > >>Sent to CCL by: Marcel Torok [mtorok{}kosice.upjs.sk] >>Dear CCL'ers, >>has somebody experience with dual core processor solutions for >>computational chemistry, mainly G03 and GAMESS? Are there some >>computation benchmarks comparing them with standard SMP configurations? >> >>Thank you >> >>Marcel >>-- >>================================================= >>RNDr. Marcel Torok, PhD. >>Katedra organickej chemie >>Ustav chemickych vied, Prirodovedecka fakulta >>Universita P.J Safarika >>Moyzesova 11 >>SK-041 54 Kosice >> >>Tel.: +421 (0)55 62 283 32 >>FAX: +421 (0)55 62 221 24 >>email: mtorok\a/kosice.upjs.sk >>=================================================> >> >> >> >> > >---------------------------------------------------------------- > Dr S.I. Gorelsky, Department of Chemistry, Stanford University > Box 155, 333 Campus Drive, Stanford, CA 94305-5080 USA > Phone: (650) 723-0041. Fax: (650) 723-0852. >----------------------------------------------------------------> > > > > > > From owner-chemistry@ccl.net Mon Sep 12 12:58:48 2005 From: "CCL" To: CCL Subject: CCL: Any open-source molecular descriptor software for short peptides? Message-Id: <-29127-050912112904-20062-o80D67vVBtiLxWnyey+vTg a server.ccl.net> X-Original-From: "tomoko niwa" Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="us-ascii" Date: Mon, 12 Sep 2005 23:48:26 +0900 MIME-Version: 1.0 Sent to CCL by: "tomoko niwa" [tomokoniwa a gamma.ocn.ne.jp] If you are interest in structure-activity relationships of bioactive peptides, the following article would be helpful. Fauchere, J. L.; Charton, M.; Kier, L. B.; Verloop, A.; Pliska, V. Amino Acid Side Chain Parameters for Correlation Studies in Biology and Pharmacology. Int. J. Pept. Protein Res. 1988, 32, 269-278. Application of these descriptors to structure-activity relationship studies, and definition of hydrophobic descriptors of amino acids are described in Sotomatsu-Niwa, T.; Ogino, A. Evaluation of the Hydrophobic Parameters of the Amino Acid Side Chains of Peptides and their Application in QSAR and Conformational Studies. THEOCHEM 1997, 392, 43-54. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Tomoko Niwa, Ph.D. Discovery Research Lab. Nippon Shinyaku Co., Ltd. 14, Nishinosho-Monguchi-cho, Kisshoin, Minami-ku Kyoto, 601-8550 Japan tel +81-75-321-9010, fax +81-75-321-9038 t.niwa a po.nippon-shinyaku.co.jp =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= On Tue, 2005-09-06 at 15:45 -0400, CCL wrote: > Sent to CCL by: "Lei Huang" [hxbus]* > Dear CCLers: > > Could anyone recommend a nice open-source molecular descriptor calculation software for short peptides with 8-15 amino acid long? I tried to search the archives but only got the internal error message. Thanks. > > Lei Huang > University of Illinois at Chicago (M/C563) > Chicago, IL 60612 > Email: hxbus at yahoo.com From owner-chemistry@ccl.net Mon Sep 12 12:58:48 2005 From: "CCL" To: CCL Subject: CCL: Harris functional Message-Id: <-29126-050912110815-12446-o80D67vVBtiLxWnyey+vTg|"|server.ccl.net> X-Original-From: Huub van Dam Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=windows-1252; format=flowed Date: Mon, 12 Sep 2005 15:31:11 +0100 MIME-Version: 1.0 Sent to CCL by: Huub van Dam [h.j.j.vandam|"|dl.ac.uk] Hi Telkuni, How about: J. Harris, Phys.Rev.B 31 (1985) 1770-1779. Best wishes, Huub CCL wrote: >Sent to CCL by: "Telkuni Tsuru" [telkuni[A]venus.dti.ne.jp] >Hello, CCLers. > >I like to study Harris functional of DFT and like to know the >papers which describe it. > >I've gotten the papers: >1) Cullen J, J. Comput Chem., 25(5):637-48 (2004) >2) S. I. Simak et.al, Solid State Communications, 87 ,Pages 393-396 (1993) >3) K. Kobayashi et.al, Chem. Phys. Lett., 188 181 (1992). >4) N. Chetty et.al, Phys. Rev. B, 46 3798 (1992), > >If you know other good papers, please teach me. > > > Sincerely yours, >---------------------------------------------------- > Telkuni Tsuru telkuni|"|venus.dti.ne.jp> > > > > -- ========================================================== Huub van Dam email: h.j.j.vandam|"|dl.ac.uk CCLRC Daresbury Laboratory phone: +44-1925-603933 Daresbury, Warrington fax: +44-1925-603634 Cheshire, UK WA4 4AD ========================================================== From owner-chemistry@ccl.net Mon Sep 12 13:41:07 2005 From: "CCL" To: CCL Subject: CCL: Molpro and memory Message-Id: <-29128-050912134014-9948-o80D67vVBtiLxWnyey+vTg|a|server.ccl.net> X-Original-From: Kirk Peterson Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Date: Mon, 12 Sep 2005 10:40:03 -0700 Mime-Version: 1.0 (Apple Message framework v622) Sent to CCL by: Kirk Peterson [kipeters|a|wsu.edu] Hi, I've re-built Molpro without parallel support and I can still allocate up to 2GB of memory on 1 processor. After looking through the Molpro users mailing list, I noted that there was a problem reported with very early versions of molpro2002 (this was back in Feb. of 2002). What version of Molpro are you running? The "fix" was posted back in July of 2002 where one should make sure the program is linked with the system gmalloc rather than the Molpro one. All subsequent versions of Molpro have this patched. (see also: http://www.molpro.net/molpro-user/archive/all/msg00485.html) Hope this helps, -Kirk On Sep 12, 2005, at 7:24 AM, CCL wrote: > > Sent to CCL by: Alexander Martins Silva [alex.msilva(!)uol.com.br] > Hi, > > "Is your kernel compiled with large memory support? " > > Yes, it is. You are using 1Gb per processor. It is the same problem. > I will compile a paralel version of molpro to try to solve partialy my > problem. > > > Thank you, > > > Alexander. > > > > CCL wrote: > >> Sent to CCL by: Kirk Peterson [kipeters{=}wsu.edu] >> Perhaps this is due to using gcc, but with the Portland Group >> compiler we consistently allocate more than 1 GB of memory (up to 2) >> with molpro on our >> 32-bit athlon cluster (dual node, 3GB RAM, 4GB of swap). Is your >> kernel compiled with large memory support? >> -Kirk >> >> >> On Fri, Sep 09, 2005 at 11:00:32PM -0400, CCL wrote: >> >>> Sent to CCL by: "Igor Filippov [Contr]" [igorf|"|helix.nih.gov] >>> I seem to remember reading about 1Gb limit per single process too, >>> though I can't find the reference now. This is what Linus had said >>> about >>> the issue - according to him, it's more like 3Gb limit, but >>> not all of it can be accessible - due to shared libs taking space >>> etc. >>> You need to move to 64-bit cpu. >>> >>> http://www.ussg.iu.edu/hypermail/linux/kernel/9801.2/0194.html >>> >>> Igor >>> >>> >>> On Fri, 2005-09-09 at 19:04, CCL wrote: >>> >>>> Sent to CCL by: "Ben Swerts" [ben.swerts/./ua.ac.be] >>>> >>>> >>>> Hi Alexander, >>>> >>>> I have absolutely no experience with Molpro but I can tell you this: >>>> When using GCC as the compiler in 32 bit mode, I noticed a limit of >>>> 1 GB for >>>> statically allocated memory when compiling fortran codes. I tried a >>>> lot of >>>> things to circumvent this problem but never succeeded. Now we have >>>> 64 bit >>>> machines and this 1 GB limit is finally a thing of the past. I >>>> guess the problem is related to Linux. Applications just crash when >>>> compiled with more than 1 GB. >>>> >>>> Greetz, >>>> >>>> >>>> Ben >>>> >>>> >>>>> -----Original Message----- >>>>> From: owner-chemistry|a|ccl.net [mailto:owner-chemistry|a|ccl.net] >>>>> Sent: Friday, September 09, 2005 10:44 PM >>>>> To: Swerts, Ben Subject: CCL: Molpro and memory >>>>> >>>>> >>>>> Sent to CCL by: Alexander Martins Silva [alex.msilva---uol.com.br] >>>>> >>>>> Hi, >>>>> >>>>> This is my second message about the same issue: Molpro >>>>> and memory. I've compiled the Molpro in a Xeon machine with 2 >>>>> proc and 4 GB of memory. However, I couldn't use more than 1Gb of >>>>> memory when executing jobs. I've already expanded the >>>>> kernel.shmall and kernel.shmmax parameters up to 2Gb and nothing >>>>> happens. The manual don't tell me nothing about this subject. Is >>>>> it a problem related to the Intel Fotran Compiler? Anybody could >>>>> help me? >>>>> >>>>> >>>>> Thanks in advance, >>>>> >>>>> Alexander.> To send e-mail to subscribers of CCL put the string CCL: on your > Subject: line> > Send your subscription/unsubscription requests to: > CHEMISTRY-REQUEST|a|ccl.net HOME Page: http://www.ccl.net | Jobs Page: > http://www.ccl.net/jobs > If your is mail bouncing from ccl.net domain due to spam filters, > please> -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- > +-+ > > > > From owner-chemistry@ccl.net Mon Sep 12 14:51:06 2005 From: "CCL" To: CCL Subject: CCL: Molpro and memory Message-Id: <-29129-050912144234-29559-o80D67vVBtiLxWnyey+vTg]_[server.ccl.net> X-Original-From: Alexander Martins Silva Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Mon, 12 Sep 2005 15:47:11 -0300 MIME-Version: 1.0 Sent to CCL by: Alexander Martins Silva [alex.msilva]_[uol.com.br] Hi, I'm running the 2002.3 version. Thanks. Alexander. CCL wrote: > > Sent to CCL by: Kirk Peterson [kipeters|a|wsu.edu] > Hi, > > I've re-built Molpro without parallel support and I can still > allocate up to 2GB of memory on 1 processor. After looking through > the Molpro users mailing list, I noted that there was a problem > reported with very early versions of molpro2002 (this was back in > Feb. of 2002). What version of Molpro are you running? The "fix" was > posted back in July of 2002 where one should make sure the program is > linked with the system gmalloc rather than the Molpro one. All > subsequent versions of Molpro have this patched. > (see also: http://www.molpro.net/molpro-user/archive/all/msg00485.html) > > Hope this helps, -Kirk > > On Sep 12, 2005, at 7:24 AM, CCL wrote: > >> >> Sent to CCL by: Alexander Martins Silva [alex.msilva(!)uol.com.br] >> Hi, >> >> "Is your kernel compiled with large memory support? " >> >> Yes, it is. You are using 1Gb per processor. It is the same >> problem. I will compile a paralel version of molpro to try to solve >> partialy my problem. >> >> >> Thank you, >> >> >> Alexander. >> >> >> >> CCL wrote: >> >>> Sent to CCL by: Kirk Peterson [kipeters{=}wsu.edu] >>> Perhaps this is due to using gcc, but with the Portland Group >>> compiler we consistently allocate more than 1 GB of memory (up to >>> 2) with molpro on our >>> 32-bit athlon cluster (dual node, 3GB RAM, 4GB of swap). Is your >>> kernel compiled with large memory support? >>> -Kirk >>> >>> >>> On Fri, Sep 09, 2005 at 11:00:32PM -0400, CCL wrote: >>> >>>> Sent to CCL by: "Igor Filippov [Contr]" [igorf|"|helix.nih.gov] >>>> I seem to remember reading about 1Gb limit per single process too, >>>> though I can't find the reference now. This is what Linus had said >>>> about >>>> the issue - according to him, it's more like 3Gb limit, but >>>> not all of it can be accessible - due to shared libs taking space >>>> etc. >>>> You need to move to 64-bit cpu. >>>> >>>> http://www.ussg.iu.edu/hypermail/linux/kernel/9801.2/0194.html >>>> >>>> Igor >>>> >>>> >>>> On Fri, 2005-09-09 at 19:04, CCL wrote: >>>> >>>>> Sent to CCL by: "Ben Swerts" [ben.swerts/./ua.ac.be] >>>>> >>>>> >>>>> Hi Alexander, >>>>> >>>>> I have absolutely no experience with Molpro but I can tell you this: >>>>> When using GCC as the compiler in 32 bit mode, I noticed a limit >>>>> of 1 GB for >>>>> statically allocated memory when compiling fortran codes. I tried >>>>> a lot of >>>>> things to circumvent this problem but never succeeded. Now we >>>>> have 64 bit >>>>> machines and this 1 GB limit is finally a thing of the past. I >>>>> guess the problem is related to Linux. Applications just crash when >>>>> compiled with more than 1 GB. >>>>> >>>>> Greetz, >>>>> >>>>> >>>>> Ben >>>>> >>>>> >>>>>> -----Original Message----- >>>>>> From: owner-chemistry]_[ccl.net [mailto:owner-chemistry]_[ccl.net] >>>>>> Sent: Friday, September 09, 2005 10:44 PM >>>>>> To: Swerts, Ben Subject: CCL: Molpro and memory >>>>>> >>>>>> >>>>>> Sent to CCL by: Alexander Martins Silva [alex.msilva---uol.com.br] >>>>>> >>>>>> Hi, >>>>>> >>>>>> This is my second message about the same issue: >>>>>> Molpro and memory. I've compiled the Molpro in a Xeon machine >>>>>> with 2 proc and 4 GB of memory. However, I couldn't use more >>>>>> than 1Gb of memory when executing jobs. I've already expanded >>>>>> the kernel.shmall and kernel.shmmax parameters up to 2Gb and >>>>>> nothing happens. The manual don't tell me nothing about this >>>>>> subject. Is it a problem related to the Intel Fotran Compiler? >>>>>> Anybody could help me? >>>>>> >>>>>> >>>>>> Thanks in advance, >>>>>> >>>>>> Alexander.> To send >>>>>> e-mail to subscribers of CCL put the string CCL: on your >>>>> >> Subject: line> >> Send your subscription/unsubscription requests to: >> CHEMISTRY-REQUEST]_[ccl.net HOME Page: http://www.ccl.net | Jobs >> Page: http://www.ccl.net/jobs >> If your is mail bouncing from ccl.net domain due to spam filters, >> please> >> -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- >> +-+> To send e-mail to subscribers of CCL put the string CCL: on your > Subject: line> > Send your subscription/unsubscription requests to: > CHEMISTRY-REQUEST]_[ccl.net HOME Page: http://www.ccl.net | Jobs Page: > http://www.ccl.net/jobs> > > > From owner-chemistry@ccl.net Mon Sep 12 17:04:55 2005 From: "CCL" To: CCL Subject: CCL: Dual core processor Message-Id: <-29130-050912161300-18668-o80D67vVBtiLxWnyey+vTg(a)server.ccl.net> X-Original-From: Matt Thompson Content-Transfer-Encoding: 7bit Content-Type: text/plain Date: Mon, 12 Sep 2005 13:08:09 -0600 Mime-Version: 1.0 Sent to CCL by: Matt Thompson [thompsma(a)jilau1.colorado.edu] On Mon, 2005-09-12 at 09:44 -0500, CCL wrote: > Sent to CCL by: John McKelvey [jmmckel\a/attglobal.net] > All, > > Could I ask a question? The job referred to below is a frequence job; > how much IO does this involve? If four heavy processes hit the two > cpu's at the same time will the bus be able to handle it? This was a > serious problem when the early quad processor boards became available. > If there is a lot of IO presumably 10KRPM SATA drives would have a > positive impact? Another thing to consider is the Intel dual-core construction versus AMD's. AMD, I believe, has an on-die memory controller for each core (as they did with their single core procs). Intel, meanwhile, has only one per 2 cores (one per processor) I think. That should have a pretty large impact on performance. It would be nice if someone out there with a dual-core Xeon could see how it scales with the same G03 test job. Matt -- The mayfly lives only one day, and sometimes it rains. - Geo. Carlin Matt Thompson -- http://ucsub.colorado.edu/~thompsma/ 440 UCB, Boulder, CO 80309-0440 JILA A510, 303-492-4662