Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
Thanks Ruben,
Could you please also send the output of make -k so that we can try to catch all other issues in one go?
Matthias
On Nov 22, 2010, at 10:12, Ruben Ghulghazaryan ghulr@yahoo.com wrote:
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
<alpsreleaseproblem.txt>
Hi Ruben,
Did you by chance set the variable LAPACK_64_BIT when configuring? If you did so, are you 100% sure that you have a 64 bit LAPACK and BLAS, which uses 8-byte instead of 4-byte integers?
Matthias
On 22 Nov 2010, at 10:12, Ruben Ghulghazaryan wrote:
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
<alpsreleaseproblem.txt>
Hi Matthias,
I send you attached the output generated by make -k command.
I used LAPACK_64_BIT ON when configuring. I used the default lapack and blas libraries from Kubuntu 10.10 distribution. I agree that the problem may be due to LAPACK and BLAS libraries.
Currently I unset LAPACK_64_BIT variable and running a new make.
I will let you know the result when it finish.
Thanks, Ruben
________________________________ From: Matthias Troyer troyer@phys.ethz.ch To: General discussion on the ALPS library components and their development comp-phys-alps-devel@lists.phys.ethz.ch Cc: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Mon, November 22, 2010 3:20:50 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Ruben,
Did you by chance set the variable LAPACK_64_BIT when configuring? If you did so, are you 100% sure that you have a 64 bit LAPACK and BLAS, which uses 8-byte instead of 4-byte integers?
Matthias
On 22 Nov 2010, at 10:12, Ruben Ghulghazaryan wrote:
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
<alpsreleaseproblem.txt>
Hi Matthias,
I seems the problem was the LAPACK and BLAS libraries due to my settting LAPACK_64_BIT ON. With the default setting LAPACK_64_BIT OFF the build went ok. All tests passed (make test).
I successfully installed ALPS 2.0 on Kubuntu 10.10 64bit version.
I thought that since I am using 64bit installation of Kubuntu 10.10 then LAPACK and BLAS libraries should also built with 64 bit support. It seems this is not the case.
Thanks for your help.
Regards, Ruben
________________________________ From: Ruben Ghulghazaryan ghulr@yahoo.com To: comp-phys-alps-users@lists.phys.ethz.ch Sent: Mon, November 22, 2010 3:52:59 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Matthias,
I send you attached the output generated by make -k command.
I used LAPACK_64_BIT ON when configuring. I used the default lapack and blas libraries from Kubuntu 10.10 distribution. I agree that the problem may be due to LAPACK and BLAS libraries.
Currently I unset LAPACK_64_BIT variable and running a new make.
I will let you know the result when it finish.
Thanks, Ruben
________________________________ From: Matthias Troyer troyer@phys.ethz.ch To: General discussion on the ALPS library components and their development comp-phys-alps-devel@lists.phys.ethz.ch Cc: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Mon, November 22, 2010 3:20:50 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Ruben,
Did you by chance set the variable LAPACK_64_BIT when configuring? If you did so, are you 100% sure that you have a 64 bit LAPACK and BLAS, which uses 8-byte instead of 4-byte integers?
Matthias
On 22 Nov 2010, at 10:12, Ruben Ghulghazaryan wrote:
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
<alpsreleaseproblem.txt>
Hi Ruben,
Thanks for uncovering a problem with 64 bit lapack. I'm glad that it all works for you now.
Matthias
On Nov 22, 2010, at 1:51 PM, Ruben Ghulghazaryan wrote:
Hi Matthias,
I seems the problem was the LAPACK and BLAS libraries due to my settting LAPACK_64_BIT ON. With the default setting LAPACK_64_BIT OFF the build went ok. All tests passed (make test).
I successfully installed ALPS 2.0 on Kubuntu 10.10 64bit version.
I thought that since I am using 64bit installation of Kubuntu 10.10 then LAPACK and BLAS libraries should also built with 64 bit support. It seems this is not the case.
Thanks for your help.
Regards, Ruben
From: Ruben Ghulghazaryan ghulr@yahoo.com To: comp-phys-alps-users@lists.phys.ethz.ch Sent: Mon, November 22, 2010 3:52:59 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Matthias,
I send you attached the output generated by make -k command.
I used LAPACK_64_BIT ON when configuring. I used the default lapack and blas libraries from Kubuntu 10.10 distribution. I agree that the problem may be due to LAPACK and BLAS libraries.
Currently I unset LAPACK_64_BIT variable and running a new make.
I will let you know the result when it finish.
Thanks, Ruben
From: Matthias Troyer troyer@phys.ethz.ch To: General discussion on the ALPS library components and their development comp-phys-alps-devel@lists.phys.ethz.ch Cc: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Mon, November 22, 2010 3:20:50 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Ruben,
Did you by chance set the variable LAPACK_64_BIT when configuring? If you did so, are you 100% sure that you have a 64 bit LAPACK and BLAS, which uses 8-byte instead of 4-byte integers?
Matthias
On 22 Nov 2010, at 10:12, Ruben Ghulghazaryan wrote:
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
<alpsreleaseproblem.txt>
There is no problem here. 64-bit Linux means sizeof(void*)=8 but says nothing about sizeof(int). The size of an integer is a compiler option and the default on both 32-bit and 64-bit Linux is that sizeof(int)=4. As one can see here - http://en.wikipedia.org/wiki/Limits.h - the size of long may vary with the compilation environment, but long is the not the type of the arguments passed to BLAS. However, the BLAS interface, regardless of implementation, lives within the world of Fortran, hence C standards do not apply. I can't find anything in either the 66 or 2003 Fortran standards that specifies how many bits an integer is on a given platform, but in practice it is 4-bytes unless one sets the 8-byte flag during compilation, but this is not how the 64-bit Linux distro packages of BLAS are generated.
Jeff
On Mon, Nov 22, 2010 at 6:53 AM, Matthias Troyer troyer@phys.ethz.ch wrote:
Hi Ruben, Thanks for uncovering a problem with 64 bit lapack. I'm glad that it all works for you now. Matthias On Nov 22, 2010, at 1:51 PM, Ruben Ghulghazaryan wrote:
Hi Matthias,
I seems the problem was the LAPACK and BLAS libraries due to my settting LAPACK_64_BIT ON. With the default setting LAPACK_64_BIT OFF the build went ok. All tests passed (make test).
I successfully installed ALPS 2.0 on Kubuntu 10.10 64bit version.
I thought that since I am using 64bit installation of Kubuntu 10.10 then LAPACK and BLAS libraries should also built with 64 bit support. It seems this is not the case.
Thanks for your help.
Regards, Ruben
From: Ruben Ghulghazaryan ghulr@yahoo.com To: comp-phys-alps-users@lists.phys.ethz.ch Sent: Mon, November 22, 2010 3:52:59 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Matthias,
I send you attached the output generated by make -k command.
I used LAPACK_64_BIT ON when configuring. I used the default lapack and blas libraries from Kubuntu 10.10 distribution. I agree that the problem may be due to LAPACK and BLAS libraries.
Currently I unset LAPACK_64_BIT variable and running a new make.
I will let you know the result when it finish.
Thanks, Ruben
From: Matthias Troyer troyer@phys.ethz.ch To: General discussion on the ALPS library components and their development comp-phys-alps-devel@lists.phys.ethz.ch Cc: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Mon, November 22, 2010 3:20:50 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Ruben, Did you by chance set the variable LAPACK_64_BIT when configuring? If you did so, are you 100% sure that you have a 64 bit LAPACK and BLAS, which uses 8-byte instead of 4-byte integers? Matthias On 22 Nov 2010, at 10:12, Ruben Ghulghazaryan wrote:
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
<alpsreleaseproblem.txt>
Indeed, but some (very few) people use a 64 bit version of blas. And they can set the integer size to 64 bit.this was however not done consistently and one application thus failed to compile. I have fixed that and will probably mark the 64 bit blas flag as advanced to prevent confusion.
Matthias
On Nov 22, 2010, at 17:11, Jeff Hammond jeff.science@gmail.com wrote:
There is no problem here. 64-bit Linux means sizeof(void*)=8 but says nothing about sizeof(int). The size of an integer is a compiler option and the default on both 32-bit and 64-bit Linux is that sizeof(int)=4. As one can see here - http://en.wikipedia.org/wiki/Limits.h - the size of long may vary with the compilation environment, but long is the not the type of the arguments passed to BLAS. However, the BLAS interface, regardless of implementation, lives within the world of Fortran, hence C standards do not apply. I can't find anything in either the 66 or 2003 Fortran standards that specifies how many bits an integer is on a given platform, but in practice it is 4-bytes unless one sets the 8-byte flag during compilation, but this is not how the 64-bit Linux distro packages of BLAS are generated.
Jeff
On Mon, Nov 22, 2010 at 6:53 AM, Matthias Troyer troyer@phys.ethz.ch wrote:
Hi Ruben, Thanks for uncovering a problem with 64 bit lapack. I'm glad that it all works for you now. Matthias On Nov 22, 2010, at 1:51 PM, Ruben Ghulghazaryan wrote:
Hi Matthias,
I seems the problem was the LAPACK and BLAS libraries due to my settting LAPACK_64_BIT ON. With the default setting LAPACK_64_BIT OFF the build went ok. All tests passed (make test).
I successfully installed ALPS 2.0 on Kubuntu 10.10 64bit version.
I thought that since I am using 64bit installation of Kubuntu 10.10 then LAPACK and BLAS libraries should also built with 64 bit support. It seems this is not the case.
Thanks for your help.
Regards, Ruben
From: Ruben Ghulghazaryan ghulr@yahoo.com To: comp-phys-alps-users@lists.phys.ethz.ch Sent: Mon, November 22, 2010 3:52:59 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Matthias,
I send you attached the output generated by make -k command.
I used LAPACK_64_BIT ON when configuring. I used the default lapack and blas libraries from Kubuntu 10.10 distribution. I agree that the problem may be due to LAPACK and BLAS libraries.
Currently I unset LAPACK_64_BIT variable and running a new make.
I will let you know the result when it finish.
Thanks, Ruben
From: Matthias Troyer troyer@phys.ethz.ch To: General discussion on the ALPS library components and their development comp-phys-alps-devel@lists.phys.ethz.ch Cc: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Mon, November 22, 2010 3:20:50 PM Subject: Re: [ALPS-users] ALPS 2.0 release candidate
Hi Ruben, Did you by chance set the variable LAPACK_64_BIT when configuring? If you did so, are you 100% sure that you have a 64 bit LAPACK and BLAS, which uses 8-byte instead of 4-byte integers? Matthias On 22 Nov 2010, at 10:12, Ruben Ghulghazaryan wrote:
Hi,
I tried to build new ALPS 2.0 release build (alps-2.0.0rc1-r5175-src-with-boost.tar.gz) with Kubuntu 10.10 and got problem with compilation of qmc code.
Please see the make report of the problem in the attached file.
Regards, Ruben
----- Original Message ---- From: Matthias Troyer troyer@phys.ethz.ch To: comp-phys-alps-users@lists.phys.ethz.ch; comp-phys-alps-devel comp-phys-alps-devel@lists.comp-phys.org Sent: Sat, November 20, 2010 10:04:05 PM Subject: [ALPS-users] ALPS 2.0 release candidate
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
Matthias Troyer
<alpsreleaseproblem.txt>
-- Jeff Hammond Argonne Leadership Computing Facility jhammond@alcf.anl.gov / (630) 252-5381 http://www.linkedin.com/in/jeffhammond
On 20 Nov 2010, at 19:04, Matthias Troyer wrote:
Dear all,
we finally have an ALPS 2 release candidate available on the Web. The MacOS X 10.5 build will take a bit longer but will also come soon. Please try to use is, install it, check your codes with it and report any bug to the mailing list. We are nearing completion of ALPS 2.0
An updated second release candidate is available. Everyone who cares that ALPS 2.0 will work "out of the box" on their machines should please try to download and install it.
Matthias
comp-phys-alps-users@lists.phys.ethz.ch