We are in Rio next week IPDPS! Join Torsten Hoefler 🇨🇭 for his keynote talk at the AsHES workshop "Performance Portability with Data-Centric Parallel Programming", Monday May 20 9:00 and learn about Stateful DataFlow multiGraph (SDFG). ETH CS Department CSCS Lugano Oak Ridge Lab mcs.anl.gov/events/worksho…

We are enjoying Rio this week IPDPS! Tal Ben-Nun will present our recent paper "A Modular Benchmarking Infrastructure for High-Performance and Reproducible Deep Learning": bit.ly/2VwymLQ tomorrow, Tuesday May 21 at 10:00. ETH CS Department #Deep500 Code: github.com/deep500/deep500

Tal Tal Ben-Nun talked today on “High #Performance Programming in #MachineLearning” and demystifying #DeepLearning during the HIPS workshop: hosting.cs.vt.edu/hips2019/ IPDPS ETH CS Department Torsten Hoefler 🇨🇭

Tal Ben-Nun Tal Ben-Nun will give a talk tomorrow on “Neural Code Comprehension: A Learnable Representation of Code Semantics” in HG F3 at 16:45, and will present a novel processing technique to use #MachineLearning for code semantics PASC25 ETH CS Department

Full auditorium for Tal Ben-Nun SPCL@ETH “Demystifying Parallel and Distributed #DeepLearning” PASC25 #PASC19 ETH Zurich #Exascale and Beyond

Our survey on parallel and distributed #DeepLearning now on the frontpage at ACM's Computing Surveys (CSUR) journal csur.acm.org. SPCL@ETH dl.acm.org/citation.cfm?i… (longer version on arXiv) arxiv.org/abs/1802.09941

Tal Tal Ben-Nun gave a talk in Oxford this week on using #DeepLearning to improve weather uncertainty quantification in Ensemble Prediction Systems at the Workshop for #MachineLearning for Weather and Climate Modelling: users.ox.ac.uk/~phys0895/mlwc… ETH CS Department ETH Zurich

Did you know SC20 has a "Machine Learning and HPC" track? Submissions from both ML and HPC communities are welcome. Abstracts Submissions Close March 31st. sc20.supercomputing.org/submit/paper-s…

Take a look at our deep learning research on improving weather forecasts! We achieve an increase of over 26% in forecast skill using only half the original processing power for events such as the cold wave over South Asia. arxiv.org/abs/2005.08748 #MachineLearning #NeuralNetwork

Satoshi Matsuoka - a true HPC Guru ;-) - will open our new online seminar series SPCL_Bcast this Thursday! Join us live (via zoom) on 9/10 at 9am CET at spcl.inf.ethz.ch/Bcast/ for his talk: The First '#Exascale' #Supercomputer – Past, Present and Future SPCL@ETH RIKEN #HPC

The future of #DeepLearning is sparse! See our overview of the field and upcoming opportunities for how to gain 10-100x performance to fuel the next #AI revolution. #HPC techniques will be key as large-scale training is #supercomputing. arxiv.org/abs/2102.00554 #MachineLearning

Data Movement Is All You Need: A Case Study on Optimizing Transformers by Andrei Ivanov (ETH Zurich); Nikoli Dryden (ETH Zurich)*; Tal Ben-Nun (ETH Zurich); Shigang Li (ETH Zurich); Torsten Hoefler (ETH Zürich)

Can Graph Neural Networks be used for learning compiler tasks? Find out in our paper at #ICML2021, where we present ProGraML, a new graph-based program representation & DeepDataFlow, a compiler analysis #machinelearning dataset! spcl.inf.ethz.ch/Publications/.… @csateth Torsten Hoefler 🇨🇭 Tal Ben-Nun

Don’t miss Nikoli’s #ISC talk at 3:35 PM CEST in the ML systems and tools session! NoPFS helps solve the I/O bottleneck for #MachineLearning training on #HPC systems. More details: arxiv.org/abs/2101.08734

The EXCLAIM project is hiring software engineers: jobs.ethz.ch/job/view/JOPG_… - so check and apply if you want to work with us!

Are you interested in identifying and resolving performance bottlenecks in programs? We're conducting a survey on the dataflow program representation that helps to analyze bottlenecks. Help our research by participating in a short 10-minute survey! pa-survey.spcl.inf.ethz.ch #HPC

Don't miss the opportunity! Register now at wids.ch #widsworldwide #wids2022 #widszurich2022 #womenindatascience #zurich #datascience #wetechtogether

Ambitious partnership for reliable climate prediction yields fastest FV3 performance yet!A collaboration between Ai2,CSCS Lugano& ETH CS Department compiles the FV3 dynamical core, written entirely in Python,to run 3.9x faster on 2400 GPUs over production FORTRAN code.#SC22 👇paper:👇

#DaCe projects available in GSoC 2023!

Today at CGO'23, Tal Ben-Nun and Berke Ates presented our paper on bringing data-centric optimizations into the standard compiler pipeline with #DaCe and #MLIR. Check out the paper! 👇 spcl.inf.ethz.ch/Publications/i…