Skip to content

Recent Topics
  • 49 Topics
    233 Posts

    Hi Mo,

    I believe the correct way to remove an atom while keeping the basis set is to use the GHOSTS keyword (see page 82 of the User manual) instead of ATOMSUBS, without modifying the original basis set.

    Could you try that and check if the energy makes sense afterward?

  • 9 Topics
    29 Posts

    Thanks a lot! I will take your advice.

  • Seek assistance, discuss troubleshooting tips for any technical problem you encounter and report bugs

    6 Topics
    35 Posts

    job314 said in No space left on device:

    those are huge HPC nodes… they can’t be possibly out of disk…

    On our cluster, although the total disk space is huge, there are limited disk quotas for each user. Maybe it is the same there for you?

    job314 said in No space left on device:

    will it affect my convergence or calculation speed?

    Possibly, but not much I think.

  • Discuss tools and techniques for visualizing simulated data

    4 Topics
    16 Posts

    Dear Jonas,

    A good option just for visualization is using the CRYSPLOT webpage, which includes a tool for this based on JSMol and only requires your output file. Another option would be to use JMol, which requires a local installation of Java. Using JSMol locally is not an easy task, as its developed for its usage on web servers. If you really need to do so, you can find some guidance in the the following sites (it will depend on which browser you want to use, and probably also on your OS):

    Recent thread on usage of JSMol locally JSMol Wiki post

    Hope this helps.

    Marcos

  • Communications for the community and updates on upcoming events

    5 Topics
    6 Posts

    Dear CRYSTAL community,

    We’re excited to share our recent work on accelerating linear algebra operations in the CRYSTAL code using GPUs. Our implementation boosts the performance of self-consistent field (SCF) calculations by offloading key matrix operations like multiplication, diagonalization, inversion, and Cholesky decomposition to GPUs.

    In the manuscript, we first analyze the performance and limitations of the standard parallel version of the code (Pcrystal) and then we evaluate the scalability of the new GPU-accelerated approach with 1 to 8 GPUs, observing remarkable scaling. To highlight these improvements, we present benchmark results on different systems, such as the example below.

    post_forum_1.png

    We expected significant speedups for large systems due to the limited number of k points, each requiring substantial computational effort. To ensure a fair comparison, we ran calculations using the massively parallel version of CRYSTAL (MPPcrystal) on a large MOF structure with over 30000 basis functions. Surprisingly, a single GPU on one node performed comparably to 512–1024 CPU cores running across 4–8 nodes.

    To find out more, read the full paper here.

    We aim to make this GPU-accelerated version of CRYSTAL available in the upcoming release, allowing all users to benefit from its enhanced performance for large-scale simulations. We look forward to reading your thoughts and discussing potential applications or further improvements.

    A big thanks to Lorenzo Donà, Chiara Ribaldone, and Filippo Spiga for their contributions to the development of this code!

Suggested Topics

  • Forum Rules

    Welcome to the official forum for CRYSTAL software users! This is a space to share knowledge, find support, and connect with others interested in solid-state simulations. To maintain a productive and respectful environment, we ask all members to adhere to the following rules...

  • CRYSTALClear

    CRYSTALClear is an open source project that provides an easy Python interface with CRYSTAL. The package allows you to quickly extract information from the CRYSTAL output files and to easily generate customizable plots...


Top Users

4

Online

160

Users

73

Topics

319

Posts