CAELinux 2009
- John
- Offline
- New Member
-
Less
More
- Posts: 8
- Thank you received: 0
15 years 10 months ago #2973
by John
Replied by John on topic Re:CAELinux 2009
Hi Joël,
Great news about the new release, but I am a bit baffled about other packages that are not mentioned in your post, especially OpenFOAM...I hope you will still keep them in the project, right? If you do plan to remove some packages, would it be possible to make a list of removed material?
Many Thanks
Best Regards
JR
Great news about the new release, but I am a bit baffled about other packages that are not mentioned in your post, especially OpenFOAM...I hope you will still keep them in the project, right? If you do plan to remove some packages, would it be possible to make a list of removed material?
Many Thanks
Best Regards
JR
- Joël Cugnoni
-
- Offline
- Moderator
-
15 years 10 months ago #2974
by Joël Cugnoni
Joël Cugnoni - a.k.a admin
www.caelinux.com
Replied by Joël Cugnoni on topic Re:CAELinux 2009
@John
sorry, I missed OpenFOAM 1.5 in the list.. actually, it is not a definitive list as I am still trying to add a few codes... so stay tuned for the final announcement.
By the way, which other package would you suggest / need?
@Alexey
You are right, compiling all these codes with system-wise libraries is like trying to "cross the jungle of the dependencies": one package would need (for example) openmpi + a full lapack library + qt3 + med 2.1 and python < 2.4 while the other needs lam-mpi, but only needs atlas math libraries + qt 4 + a given version of glut + med 2.3 + python >= 2.4...
Actually, this is why I have not yet packaged these codes... and prefer to manually tune & compile a full distribution; trying to use system libraries whenever possible and recompiling specific libs when necessary (=very often...).
When compiling code, I try to keep all the files in a given folder in /opt, more or less like a stand-alone distribution, so that the software can be copied directly from one install to the other (you just need to install a few system dependencies and that's it). It is a very practical way of doing when you manage a small size cluster or laboratory for example:
1) compile and install the package in one machine, evaluate system dependencies, test it
2) copy it to all nodes with scp
3) on all nodes (script) install the missing system dependencies .
sorry, I missed OpenFOAM 1.5 in the list.. actually, it is not a definitive list as I am still trying to add a few codes... so stay tuned for the final announcement.
By the way, which other package would you suggest / need?
@Alexey
You are right, compiling all these codes with system-wise libraries is like trying to "cross the jungle of the dependencies": one package would need (for example) openmpi + a full lapack library + qt3 + med 2.1 and python < 2.4 while the other needs lam-mpi, but only needs atlas math libraries + qt 4 + a given version of glut + med 2.3 + python >= 2.4...
Actually, this is why I have not yet packaged these codes... and prefer to manually tune & compile a full distribution; trying to use system libraries whenever possible and recompiling specific libs when necessary (=very often...).
When compiling code, I try to keep all the files in a given folder in /opt, more or less like a stand-alone distribution, so that the software can be copied directly from one install to the other (you just need to install a few system dependencies and that's it). It is a very practical way of doing when you manage a small size cluster or laboratory for example:
1) compile and install the package in one machine, evaluate system dependencies, test it
2) copy it to all nodes with scp
3) on all nodes (script) install the missing system dependencies .
Joël Cugnoni - a.k.a admin
www.caelinux.com
- John
- Offline
- New Member
-
Less
More
- Posts: 8
- Thank you received: 0
15 years 10 months ago #2976
by John
Replied by John on topic Re:CAELinux 2009
Hi Joel
will do, definitely. OpenFOAM is an important tool; it would be also nice to include any patches for the current version. I would add OpenDX as a keeper. I am also extremely impressed by a package developed by NIST, called OOF ( www.ctcms.nist.gov/oof/oof2/index.html#download ). It is a brilliant way of modeling micro structures in order to establish mechanical properties of materials that cannot be tested directly in the lab. About what Alexey wrote, I agree. To resolve dependencies of the various packages with the main OS trunk (trying to trim some fat) is a hellish experience...
Best Regards
JR
will do, definitely. OpenFOAM is an important tool; it would be also nice to include any patches for the current version. I would add OpenDX as a keeper. I am also extremely impressed by a package developed by NIST, called OOF ( www.ctcms.nist.gov/oof/oof2/index.html#download ). It is a brilliant way of modeling micro structures in order to establish mechanical properties of materials that cannot be tested directly in the lab. About what Alexey wrote, I agree. To resolve dependencies of the various packages with the main OS trunk (trying to trim some fat) is a hellish experience...
Best Regards
JR
- Jakob Andersen
- Offline
- Junior Member
-
Less
More
- Posts: 24
- Thank you received: 0
15 years 10 months ago #2984
by Jakob Andersen
Replied by Jakob Andersen on topic Re:CAELinux 2009
Hi Joel,
I have been using Gmsh as a meshing tool for flow simulations in Code Saturne. Often however, I've come across the same problem - describing the boandary layers in detail. In the commercial code ccm+, which I used up intil recently, this was easily solved with a prism layer - a feature that is not (please correct me if I am wrong) included in gmsh.
In the caelinux forum, I did however hear another package mentioned - the enGrid meshing utility, that is able to introduce a prism layer to existing gmsh meshes. Also, as I understand it, the utility is able to export to openFOAM, which might also be handy.
Has this package been under consideration or is it in fact made obsolete by existing meshing tools?
Thank you,
Jakob Andersen
PS: As said so many times before - your work with developing CAE Linux is greatly appreciated!
I have been using Gmsh as a meshing tool for flow simulations in Code Saturne. Often however, I've come across the same problem - describing the boandary layers in detail. In the commercial code ccm+, which I used up intil recently, this was easily solved with a prism layer - a feature that is not (please correct me if I am wrong) included in gmsh.
In the caelinux forum, I did however hear another package mentioned - the enGrid meshing utility, that is able to introduce a prism layer to existing gmsh meshes. Also, as I understand it, the utility is able to export to openFOAM, which might also be handy.
Has this package been under consideration or is it in fact made obsolete by existing meshing tools?
Thank you,
Jakob Andersen
PS: As said so many times before - your work with developing CAE Linux is greatly appreciated!
- Jonathan Arata
- Offline
- New Member
-
Less
More
- Posts: 5
- Thank you received: 0
15 years 10 months ago #2985
by Jonathan Arata
Replied by Jonathan Arata on topic Re:CAELinux 2009
Hello,
First of all, as a new user please accept my thanks for what has to be an enormous amount of work required in putting-together CAELinux. Very well done.
Just following up - are you still thinking that perhaps (fingers crossed!) the latest/greatest CAELinux will be available this week (your original estimate was ~10 June)? Just trying to time my own work (I have to make a home for the latest version and that will take a couple of days - but I want to jump right on it as soon as it's available).
Many thanks again, and best wishes.
Jon
First of all, as a new user please accept my thanks for what has to be an enormous amount of work required in putting-together CAELinux. Very well done.
Just following up - are you still thinking that perhaps (fingers crossed!) the latest/greatest CAELinux will be available this week (your original estimate was ~10 June)? Just trying to time my own work (I have to make a home for the latest version and that will take a couple of days - but I want to jump right on it as soon as it's available).
Many thanks again, and best wishes.
Jon
- Joël Cugnoni
-
- Offline
- Moderator
-
15 years 10 months ago #2986
by Joël Cugnoni
Joël Cugnoni - a.k.a admin
www.caelinux.com
Replied by Joël Cugnoni on topic Re:CAELinux 2009
@Jakob
Yes, enGrid is included and seems to be a great tool for CFD meshing!
I tried it on simple cases, and it can give nice prismatic boundary layer elements. The only drawback that I see at the moment is that it takes a lot of time to mesh. But it is fully automatic so you can have a coffee in the mean time.
@Jon
I am progressing well, the distro itself is done, but I have found some issues with Code-Aster configuration (SSH not working in liveDVD, problems when changing hostname) so I need to update my automatic configuration scripts. Concerning release date, I will probably need to spend part of the week-end to finish the "fine tuning" and release it early next week.
Yes, enGrid is included and seems to be a great tool for CFD meshing!
I tried it on simple cases, and it can give nice prismatic boundary layer elements. The only drawback that I see at the moment is that it takes a lot of time to mesh. But it is fully automatic so you can have a coffee in the mean time.
@Jon
I am progressing well, the distro itself is done, but I have found some issues with Code-Aster configuration (SSH not working in liveDVD, problems when changing hostname) so I need to update my automatic configuration scripts. Concerning release date, I will probably need to spend part of the week-end to finish the "fine tuning" and release it early next week.
Joël Cugnoni - a.k.a admin
www.caelinux.com
Moderators: catux
Time to create page: 0.130 seconds