I can successfully install Molflow 2.7.7 (Fedora/Redhat linux). But for all the versions newer than 2.7.7, I only get a “black screen” upon starting molflow. I guess there must be some well-known procedure that I did not perform, otherwise it would have been mentioned here already by other people.
My system runs RedHat RHEL 7.
Could you tell me what I need to do in order to get the newest Molflow, for instance, 2.7.11, installed successfully on Fedora/Redhat linux?
Dear Chi-Yang,
For our Fedora builds we are testing on a CentOS 7 Version based on RHEL 7.7 combined with SDL 2.10.
Could you tell us which exact RHEL and SDL2 version you are using?
Also, could you provide us with the terminal output you get from launching Molflow with the startup script ?
Thank you very much for your message. Actually, I appreciate very much the fact you told me that the Fedora builds are tested on a CentOS 7 based on RHEL7.7 combined with SDL 2.10. So there must be something wrong with my shell environment, not the Molflow distribution. And I found that is exactly the case.
In my .bashrc file, the LD_LIBRARY_PATH points to a graphic library pertaining to ANSYS R19.3. That particular path is responsible for the “black screen” upon starting Molflow.
So the problem is resolved. Thank you very much for your response that helped me find my own problem.
Best regards,
Chi-Yang
PS: For the sake of reference, here is the offending line in my .bashrc file that causes the display issues:
Great to hear that you pinpointed the problem, and many thanks for the detailed feedback.
We don’t have many Linux users so it’s really helpful for future Redhat enthusiasts,
Cheers, Marton
Thank you for your message. The speedup on a Linux cluster running parallel Molflow is really outstanding. I will be happy to share with you and everyone on the forum the “layman’s benchmark results” some time next week. In short, running parallel Molflow on Linux is a no-brainer if the model is complex.
It would be really appreciated.
Pascal is experimenting on speeding up Molflow on a GPU, which is not straightforward (very high speed but less ray tracing precisions). We planned clusters as next step, but didn’t get to it yet. (For your info, Pascal is working on scriptability to launch Molflow automatically without the graphical interface)
Indeed, if you could spare the time for feedback on cluster results, we’d be more than interested.
Regards, Marton