Brought together our two #TitanComputers that used to support two different projects to set up a small #MPI / #OpenMP cluster (96 physical cores, 1TB DRAM, 40TB storage). We will probably drop a couple of GPUs inside them later on, but for now, they will support our EMR data analytics on the #CPU
Explore the latest OpenMP 6.0 webinar by POP!
Learn about the strategic roadmap, enhanced base language support (C23, C++23, FORTRAN), innovative tasking (free-agent threads, taskgraphs, iteration concepts), advanced loop transformations (user-defined inductions), threadsets, transparent tasks, and new device offloading features.
Watch here: https://youtu.be/YTbPPhOjO58
Slides: https://bit.ly/3OpnMP6
Call for Papers – IWOMP 2025
The premier OpenMP workshop invites unpublished research on all aspects of parallel programming.
Submission deadline: May 16, 2025 (AoE)
Oct 1–3 | 
UNC Charlotte, USA | Co-located with #EuroMPI2025
2025 Theme: Balancing Productivity and Performance Portability
Accepted papers will be published in Springer LNCS.
My brain is absolutely fried.
Today is the last day of coursework submissions for this semester. What a hectic month.
DNN with PyTorch, Brain model parallelisation with MPI, SYCL and OpenMP offloading of percolation models,hand optimizing serial codes for performance.
Two submissions due today. Submitted one and finalising my report for the second one.
Definitely having a pint after this
Ready to test the limits of performance?
Join the @EPCC Hackathon on AMD GPUs and explore the cutting-edge #MI300A and AMD’s Next Generation #Fortran Compiler with #OpenMP offload!
Bring your code, ideas, and curiosity.
Optimize, accelerate, and innovate with us.
Let’s see what you can build!
https://www.archer2.ac.uk/training/courses/250527-amd-hackathon/
Interested in #MPI and #OpenMP parallel programming to speed up your scientific applications written in #C, #Cpp, #Fortran or #Python (with #numpy)?
Attend our course in #Mainz at the Johannes Gutenberg University (#JGU) for a 4-day course from 1. April to 4. April 2025! We still have plenty of space available!
See our announcement page for further details and to register: https://indico.zdv.uni-mainz.de/event/34/
Note, it is an on-site course.
Interested in #MPI and #OpenMP parallel programming to speed up your scientific applications written in #C, #Cpp, #Fortran or #Python (with #numpy)?
Attend our course in #Mainz at the Johannes Gutenberg University (#JGU) for a 4-day course from 1. April to 4. April 2025!
See our announcement page for further details and to register: https://indico.zdv.uni-mainz.de/event/34/
Note, it is an on-site course.
Any #OpenMP lawyers?
Is this not legal?
nvfortran complains that nloop is not shared or private. Do loop trip-counts really need to be?
And for reduction variables (loopsum) too?
program test_omp_private
implicit none
integer :: iloop, nloop
integer :: loopsum
nloop = 10000
loopsum = 0
!$omp parallel do default(none) firstprivate(loopsum) reduction(+:loopsum)
do iloop = 1,nloop
loopsum = loopsum + iloop
end do
!$omp end parallel do
end program
@berkeleylab and NERSC are proud to have hosted the OpenMP ARB Language Committee for a productive face-to-face meeting, working toward the future of the portable OpenMP API beyond the recent release of OpenMP 6.0. The committee put their heads together to discuss a range of topics in the areas of accelerators, C/C++, Fortran, tasking, examples, and more. It's been a great chance to connect, innovate, and plan!
Unleashing Matrix Multiplication: A Deep Dive into Optimizing C Code for Performance
In a groundbreaking update, a new implementation of matrix multiplication in C has emerged, outpacing established libraries like OpenBLAS. This article explores the intricate steps taken to enhance pe...
We're used to leaning on children's books in Computer Science - with Gulliver's big-endian vs little-endian. Back at Supercomputing hashtag#SC24, I spoke at the hashtag#Intel booth all about open standards, performance portability, and the journey up the Yellow Brick Road to see the Wizard of Oz. Check out the video of the talk on YouTube:
https://youtu.be/xO8FGAOScpo?si=_BnVilvTBa0Ns6dX
#performanceportability #OpenMP #SYCL
@EricCarroll this is btop - i prefer it in the higher end servers (I use atop in the smaller #arm64 and htop in the mid range Arms). I think the scroll bar of the process subwindow was not fully scrolled to the bottom (should have shown a few #Rstats processes with very high PIDs, each of which was slaving away with #openmp tasks).
This is a simulation for a software adaptation of methods for large-scale survival analysis in electronic health records in shared memory threaded environments
I run a tight ship , and no one gets to slack off with my #openmp tasks
I run a tight ship , and no one gets to slack off with my #openmp tasks
Fajne rzeczy dzieją się z #LLVM w #Gentoo:
1. Już o tym wspominałem, ale wszystkie paczki teraz siedzą w llvm-*/, więc łatwiej je znaleźć.
2. Wrzuciłem llvm-runtimes/offload, przywracając obsługę offloadowania #OpenMP (dla 19.1.6 wzwyż).
3. Wrzuciłem llvm-core/mlir, korzystając ze świeżutkiej biblioteki dynamicznej (dostępne w najnowszym snapshocie 20.x, do 19.x raczej nie wejdzie już).
4. Przy okazji zacząłem pracować nad Flangiem, ale nic nie obiecuję. Chciałbym chociaż na poziomie projektu wprowadzić obsługę biblioteki dynamicznej MLIR-a.
5. Wysłałem łatki do wsparcia samodzielnego budowania #ClangIR, więc jak wszystko dobrze pójdzie, od pierwszego dnia będzie nam działało.
6. Na przegląd poszła nowa eclass llvm-r2, ze znacznie lepszym wsparciem kompilacji skrośnej. https://public-inbox.gentoo.org/gentoo-dev/20241221160053.780079-1-mgorny@gentoo.org/T/#t
7. Powoli też pracuję nad poprawą wsparcia kompilacji skrośnej w samych paczkach LLVM, ale nie wiem, kiedy skończę. Chciałbym chociaż wyeliminować konieczność budowania drugiej pełnej kopii LLVM i Clanga na CBUILD, tylko po to, by skompilować Clanga.
Some recent #LLVM goodies in #Gentoo:
1. Well, I've said that already but all packages are now in llvm-*/, so they're easier to find.
2. I've added llvm-runtimes/offload, which restores the offloading functions of #OpenMP (available in 19.1.6+).
3. I've added llvm-core/mlir, now that it provides a proper dylib to link to (available in the most recent 20.x snapshot, no plans to backport to 19.x).
4. I've started working on #Flang too, but no promises. At the very least, I'd like to make it support #MLIR dylib upstream.
5. I've sent patches for standalone build support in #ClangIR, so hopefully it'll be ready for integration when it lands in mainstream.
6. I've sent a new llvm-r2 eclass for review, now with much cleaner support for cross-compilation scenarios. https://public-inbox.gentoo.org/gentoo-dev/20241221160053.780079-1-mgorny@gentoo.org/T/#t
7. I'm also slowly working on improving cross-compilation support for LLVM packages themselves, but no ETA on this. At the very least, I'd like to remove the necessity of building another complete copy of LLVM+Clang for CBUILD to compile #Clang.
Do you want to learn about the new features in OpenMP 6.0?
Today, Michael Klemm (AMD) and myself (RWTH) will present the 31st POP webinar titles "OpenMP 6.0 Part 1: New Host-side Features and Enhancements".
Thesis: Leveraging the potential of task-based programming with OpenMP task graphs
