Apparently, someone didn’t like this post and it got hidden. I thought the topic would be of specific interest to this community. I’ve removed the news link - the title should suffice.
11 Likes
is NOAA also on the chopping block? I believe I saw something
1 Like
All your questions and more answered on this years Weather and Climate Live Stream, now until 6 pm ET Wed. One great talk after another to inform and motivate action. (Full disclosure: I’m speaking at 5:40 am ET Tuesday morning.)
4 Likes
This thread has been brought to my attention. The original post was about a court decision on U.S. government policy regarding NCAR. NCAR traditionally uses a lot of Fortran, but threads focused on political or policy matters are generally off-topic here.
I’m leaving this thread open, but please keep discussion technical. If it shifts into politics we’ll have to close it.
2 Likes
I have recently read that interesting european (Germany and Swiss) paper using Fortran:
It is about how to parallelize the ICON Fortran model on two supercomputers: JUPITER in Germany and Alps in Switzerland. JUPITER, according to the November 2025 TOP500 list, has become —in its Booster configuration— the first exaflop-class (exascale) supercomputer in Europe and the fourth in the world (the top three being in the U.S.). At the time of this article’s publication, it featured 23,536 NVIDIA Grace Hopper GH200 superchips.
The authors take an approach that separates the Fortran code written by the researchers from the optimizations performed by the performance engineers to ensure it runs optimally on the supercomputer. In particular, they streamlined the most computationally intensive part of the atmospheric component (the dynamical core) to reduce it to 1,400 lines of code and used the DaCe framework (Data-Centric Parallel Programming) to parallelize it as effectively as possible, rather than using OpenACC. Another example: the authors take advantage of the fact that ocean responses are much slower than those of the atmosphere—so the ocean runs on the CPU, and the rest on the GPU.
The team reports a temporal compression of \tau = 145.7 simulated days per day of computation and concludes that in just under six months of computation, it will be possible to simulate two global warming scenarios, with variations, over the next thirty years.
(that short text is a DeepL translation of a post I made somewhere in French)
Such models of course need good data coming from all other the (common) world, and international collaborations.
5 Likes
Thank you @sblionel for posting it. It’s important for people to be aware of these developments. It’s also not widely known that NSF NCAR has historically been, and continues to be a significant contributor to the scientific ecosystem of Fortran libraries and applications, and to the advances in science in general. So, anyone that cares about Fortran will care about NCAR prevailing during these uncertain times.
The actions against NCAR are indeed of political nature. Discussing them here is thus considered off topic.
At the same time, any high level developments that pertain to Fortran, even if not strictly technical, are considered on topic here.
So, please continue posting such news that affect the Fortran world, while staying neutral and objective when it comes to interpretations.
This is my opinion as the original creator of Fortran Discourse. Other admins and moderators may do what they want with it.
12 Likes
Hi Milan, I agree with you since most of Fortran’s success is driven through its adoption at large research centres and universities the dismantling of such a place has a direct impact on the community.
3 Likes
Indeed, and worldwide. This is a translation of the conclusion of a French news:
Or take the young French researcher who won’t be doing a postdoc at NCAR due to uncertainty about funding. “The best want to work with the best,” emphasizes Josh Hacker, a former NCAR employee now working at a startup, who is concerned about the appeal of the Boulder hub. “When talent leaves, it’s harder to keep those who remain,” “it’s a deadly spiral.”
(moderators can delete my post if it is not neutral enough, I won’t take offense)
5 Likes
I first heard of NCAR in the 1980s because of the NCAR Graphics library. It was one of a handful of fortran based graphics libraries for various types of 2D graphics: line graphs, contour graphs, vector flow graphs, and so on. I used it a little because of some codes I inherited that used it. One of its competitors was GNUPLOT, which I also used and which I still use today. Some other similar libraries are PGplot, PLplot, and DISLIN, I thought one of the problems in the 1980s and 1990s was that there were too many options for creating graphs, and there was no trend toward a standard package that every programmer could use and that could be relied on to be available and supported. It seems that situation has pretty much remained even today, or perhaps even worse as the capabilities have extended to 3D graphics and animation.
My other connection to NCAR is that I know many scientists over the decades who have worked there as permanent and temporary staff, both in my field of chemistry and in other fields. The facility has produced some of the highest quality research over its time.
2 Likes
Discussion of the scientific contributions of NCAR is on-topic here but discussion of how much it should be funded is not, since it is political, so I edited one post.
3 Likes
written in C and Fortran has been superseded by a Python version, but LLMs make it easier to maintain projects.
2 Likes
For the full breadth of open-source software work, and many are in the project-specific orgs and not included here.
Among those, WRF remains the most widely used research-grade open-source atmospheric model in the world, and CESM likewise but for climate.
5 Likes
Science without conscience is nothing but ruin to the soul
François Rabelais
1 Like
And Science avec conscience (Science with conscience, 1982) is a book by Edgar Morin about complexity and “complex thought”. He died one week ago at 104 (he published his last book last year!).
Earth system is one of the most complex system we can study, and the most important for us little human beings, as we are just part of it (hence the complexity). I am happy that there are great scientists at that great NCAR laboratory with beautiful models and big supercomputers. We need them and I wish them the best.
2 Likes
I worked with NCL (NCAR Command Language) quite a bit and my colleagues, students, and I all benefit from the work done at NCAR. So does Fortran.
2 Likes
This is the link: https://www.ncl.ucar.edu/
It seems now discontinued: Pivot to Python - September 2019 Update
On February 6, 2019, NCAR announced the decision to adopt Python as the scripting language platform of choice for future development of analysis and visualization tools and to place NCL into maintenance mode. NCL Version 6.6.2 will be the last major release of NCL. Subsequent releases will be issued as resources allow for critical bug fixes.
1 Like
There’s a discussion about NCL in a previous post. Definitely, there’s some more Fortran code in there that needs rescuing.
4 Likes