Miquel Pericas

@pmiquel.bsky.social

Professor in parallel computing, HPC/AI runtimes and compilers. Metalhead, go player https://tinyurl.com/np5ntcdh

I just realized that Tony Hoare passed away on March 5th. Hoare, the 1980 Turing Award Winner, will be recalled for many fundamental contributions in CS, among others quicksort, communicating sequential processes and the dining philosophers problem.

Intrigued by FelixCLC's talk at FOSDEM 2026 about RISC-V design decisions. Agree about RVC (compressed) instructions. Seems to add just complication without much use case. Is there a chance that future specs deprecate it? #riscv

How many people can say their childhood interests followed them into adulthood? Now imagine rediscovering that interest later in life and using it to change your local community. For #BHM’2026, ACM shares the accomplishments of Marlene Mhangami and her impact on computing.

So, modern GPUs increasingly rely on FP emulation like Ozaki scheme to provide high FP64 flops using using low precision tensor cores. But that will benefit only GEMM codes? Rubin will only have 33 TFLOPS of native FP64, while 200TF of emulated FP64. #HPC

The OpenMP ARB announces that Python will be supported in 7.0! www.openmp.org/press-relea... I wonder if this means full support a la Fortran/C/C++ or if it will focus on a subset, or just bindings for loading/running OpenMP programs from Python. Excited to see this development!

New Python Subcommittee, and Anaconda joins OpenMP ARB - OpenMP

Python will be the fourth officially-supported language in the OpenMP API, and leading Python infrastructure company Anaconda joins the OpenMP ARB 

openmp.org

Excited after the first week of our project course on parallel computing. Over the next 7 weeks, 92 students, assembled into teams of 6-7, will tackle an LLM inference optimization challenge (CPU+GPU), and containerize+run key HPC benchmarks (HPL, HPCG, HPL-MP, Graph500) on our local cluster Vera.

Sad to learn about the passing of Mike Flynn. Mike made several core contributions to computer architecture, and will long be remembered for the Flynn taxonomy, which my students also study. This ACM sigarch blog summarizes his life and achievements www.sigarch.org/in-memoriam-... #hpc #comparch

In Memoriam: Remembering Mike Flynn

Michael J. Flynn is a widely respected contributor—indeed a giant—in the field of Computer Architecture.  He made highly significant and impactful contributions throughout his career, both in indus…

sigarch.org

Looking for a postdoc in HPC? We have one postdoctoral opening to work on compilation+runtimes for RISC-V based vector and matrix multicore processors. The research is part of the European @dare-eu.bsky.social project (dare-riscv.eu). Apply by Nov 21st 2025 here: www.chalmers.se/en/about-cha...

dare-project

A new era for supercomputing in Europe The DARE project aims to boost European digital transformation by developing novel computing and AI technologies Latest updates from DARE Nothing Fo...

dare-riscv.eu

I am recruiting one PhD to work on runtimes and code generation for next-gen RISC-V vector/matrix processors! This research will be part of the EuroHPC DARE project (dare-riscv.eu). Apply by Oct 31st, 2025! www.chalmers.se/en/about-cha... Looking for a postdoc? Please email! @dare-eu.bsky.social

dare-project

A new era for supercomputing in Europe The DARE project aims to boost European digital transformation by developing novel computing and AI technologies Latest updates from DARE Nothing Fo...

dare-riscv.eu

Excited to be part of the @dare-eu.bsky.social team! Chalmers will be working on optimizing parts of SYCL, OpenMP, and oneDNN for our next generation of European #RISCV processors. We will be posting job vacancies on these topics soon! Follow me for updates

DARE@dare-eu.bsky.social · last yr.

🚀 DARE SGA1 has launched! 🚀 A €240M and 38 partners initiative, to build #Europe’s first fully European #HPC & #AI stack—from #RISCV #processors to an optimized #software ecosystem. 🔗https://dare-riscv.eu/launch-of-dare-sga1-project/