XVII INFN International School on Efficient Scientific Computing (ESC26)

Oct 11, 2026, 5:30 AM → Oct 22, 2026, 11:19 PM Europe/Rome

Rivellino (Bertinoro)

EFFICIENT SCIENTIFIC COMPUTING SCHOOL  

12 - 22 October 2026

Bertinoro (FC) Italy

The 17th edition of the ESC International School organized by INFN about "Architectures, tools and methodologies for the development of large-scale scientific computing applications", will take place from 12 to 22 October 2026 at Ce.U.B. in Bertinoro (Forlì-Cesena) in Italy.

The school is primarily aimed at PhD students and young researchers who are actively involved in the development of scientific applications and software tools for research.

The programme includes introductory lectures on current trends in the evolution of processor architectures and parallel programming.  This will be followed by more in-depth lectures on advanced programming with C++, efficient use of memory, floating-point computation and programming in a heterogeneous environment, exploiting the potential of multi-threaded computing, GPUs and cluster computing with MPI.

Arrival date: Sunday 11th October 2026.

Hands-on material

Sun, October 11

4:30 PM → 5:00 PM Shuttle service at 17.35 and 19.00 from Forlì train station

5:00 PM → 7:00 PM Check-in and registrations

8:00 PM → 9:30 PM Welcome Dinner

Mon, October 12

9:00 AM → 9:30 AM Welcome

School Introduction

Speaker: Alberto Garfagnini (Istituto Nazionale di Fisica Nucleare)

AGarfagnini_ESC26_welcome.pdf

9:30 AM → 10:30 AM School Introduction

School Introduction

Speaker: Timothy Mattson (Human Learning Group)

CompArch_ESC_intro.pdf

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM Environment setup

Speakers: Daniele Cesini (Istituto Nazionale di Fisica Nucleare), Daniele Lattanzio (Istituto Nazionale di Fisica Nucleare)

12:30 PM → 2:00 PM Lunch Break

2:00 PM → 2:45 PM Parallel architectures

Speaker: Timothy Mattson (Human Learning Group)

Parallel_comp_emphasizing_simd.pdf

2:45 PM → 3:30 PM Floating-point and vectorization

Speaker: Timothy Mattson (Human Learning Group)

FlopTalk.pdf

3:30 PM → 4:00 PM Coffee break

4:00 PM → 5:30 PM Floating-point and vectorization

Speaker: Timothy Mattson (Human Learning Group)

5:30 PM → 6:00 PM Consolidation

6:00 PM → 7:00 PM Students lightning presentations

8:00 PM → 9:30 PM Dinner

Tue, October 13

9:00 AM → 10:30 AM Efficient C++ programming

Speaker: Francesco Giacomini (Istituto Nazionale di Fisica Nucleare)

cpp-esc26.pdf

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM Efficient C++ programming

Speaker: Francesco Giacomini (Istituto Nazionale di Fisica Nucleare)

12:30 PM → 2:00 PM Lunch Break

2:00 PM → 3:30 PM Efficient C++ programming

Speaker: Francesco Giacomini (Istituto Nazionale di Fisica Nucleare)

3:30 PM → 4:00 PM Coffee Break

4:00 PM → 6:00 PM Consolidation

6:00 PM → 7:00 PM Students lightning presentations

8:00 PM → 9:30 PM RESTAURANT - Dinner at "Nonna Rina"

-------> Link Restaurant Nonna Rina

Wed, October 14

9:00 AM → 10:30 AM Efficient C++ programming

Speaker: Francesco Giacomini (Istituto Nazionale di Fisica Nucleare)

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM Efficient C++ programming

Speaker: Francesco Giacomini (Istituto Nazionale di Fisica Nucleare)

12:30 PM → 2:00 PM Lunch Break

2:00 PM → 3:30 PM Efficient C++ programming

Speaker: Francesco Giacomini (Istituto Nazionale di Fisica Nucleare)

3:30 PM → 4:00 PM Coffee break

4:00 PM → 6:00 PM Consolidation

6:00 PM → 7:00 PM Seminar: "Use and Abuse of Random Numbers"

There are a wide range of problems where we find useful answers by “sampling” a solution space. These answers work if the samples are completely uncorrelated, that is, if they are randomly selected. Generating numbers that appear to be random is easy to do. The problem is it’s easy to generate sets of such numbers that seem to be randomly distributed but actually aren't. In this talk, we’ll discuss how to work with numbers that appear to be random and to do so in ways that aren’t stupid.

Speaker: Timothy Mattson (Human Learning Group)

8:00 PM → 10:00 PM Dinner

Thu, October 15

9:00 AM → 10:30 AM Introduction to Parallel Programming with OpenMP

Speaker: Timothy Mattson (Human Learning Group)

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM Introduction to Parallel Programming with OpenMP

Speaker: Timothy Mattson (Human Learning Group)

12:30 PM → 2:00 PM Lunch

2:00 PM → 3:30 PM Working with threads - The C++ memory model

Speaker: Timothy Mattson (Human Learning Group)

3:30 PM → 4:00 PM Coffee break

4:00 PM → 6:00 PM Consolidation

8:00 PM → 10:00 PM RESTAURANT - Pizza party at "La Grotta"

-------> Link Restaurant La Grotta

Fri, October 16

9:00 AM → 10:30 AM Efficient Memory Usage

Speaker: Wahid Redjeb

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM Efficient Memory Usage

Speaker: Wahid Redjeb

12:30 PM → 2:00 PM Lunch Break

2:00 PM → 3:30 PM Debugging

Speaker: Francesco Giacomini (Istituto Nazionale di Fisica Nucleare)

3:30 PM → 4:00 PM Coffee break

4:00 PM → 5:00 PM Consolidation

5:30 PM → 7:30 PM Wine event - Discover the Albana Wine (upstairs in the Rocca Castle)

-------> Link to the event

8:00 PM → 10:00 PM Dinner

Sat, October 17

9:00 AM → 10:30 AM Efficient C++ programming

Speaker: Simone Balducci (Istituto Nazionale di Fisica Nucleare)

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM Python Bindings

Speaker: Simone Balducci (Istituto Nazionale di Fisica Nucleare)

12:30 PM → 2:00 PM Lunch

2:00 PM → 3:30 PM Consolidation / Study time

3:30 PM → 4:00 PM Coffee break

6:00 PM → 8:00 PM Poster session party (upstairs in the Rocca Castle)

Sun, October 18

7:30 AM → 8:00 AM FREE DAY (with breakfast at the canteen and lunch and dinner at restaurants, see details below)

8:30 AM → 10:00 AM Breakfast

12:00 PM → 1:20 PM RESTAURANT - Lunch at TBA

7:30 PM → 8:50 PM RESTAURANT - DINNER at Restaurant Cà De Bé (downstairs city center)

-------> Link Restaurant Cà de Bè

Mon, October 19

9:00 AM → 10:30 AM OpenMP and friends: task parallelism

Speaker: Timothy Mattson (Human Learning Group)

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM OpenMP beyond Symmetric Multiprocessors

Speaker: Timothy Mattson (Human Learning Group)

12:30 PM → 2:00 PM Lunch Break

2:00 PM → 3:30 PM GPU Programming

Speaker: Simone Balducci (Istituto Nazionale di Fisica Nucleare)

3:30 PM → 4:00 PM coffee Break

4:00 PM → 6:00 PM Consolidation

6:00 PM → 7:00 PM Seminar: Radiation-Induced Faults in Modern HPC Systems

Radiation-Induced Faults in Modern HPC Systems
Reliability is an increasingly important concern in high-performance and scientific computing, where large-scale systems integrate complex processors, memories, and accelerators and are expected to execute long-running, computation-intensive applications. Among the possible sources of failures, transient faults induced by energetic particles, such as atmospheric neutrons, can affect hardware components and propagate to software, potentially leading to incorrect results or application failures.
In this talk, we will introduce the main mechanisms behind radiation-induced transient faults and discuss their impact on modern HPC systems, with particular attention to emerging challenges in GPU-accelerated platforms. Experimental methodologies based on particle accelerators will be presented as a powerful tool for characterizing hardware sensitivity, identifying vulnerable components, and studying fault propagation up to scientific applications. Finally, the talk will discuss software-based techniques for detecting, tolerating, and mitigating transient faults.

Speaker: Filippo Reghenzani (PoliMI)

8:00 PM → 9:00 PM Dinner

Tue, October 20

9:00 AM → 10:30 AM GPU Programming

Speaker: Simone Balducci (Istituto Nazionale di Fisica Nucleare)

10:30 AM → 11:00 AM Coffee break

11:00 AM → 12:30 PM GPU Programming

Speaker: Simone Balducci (Istituto Nazionale di Fisica Nucleare)

12:30 PM → 2:00 PM Lunch

2:00 PM → 3:30 PM Cluster computing with MPI

Speaker: Daniele Cesini (Istituto Nazionale di Fisica Nucleare)

3:30 PM → 4:00 PM Coffee break

4:00 PM → 6:00 PM Cluster computing with MPI

6:00 PM → 7:00 PM Seminar: High-Performance GPU Computing in R for Bioinformatics

The explosion of single-cell measurement techniques has created a computational bottleneck in bioinformatics, where the need for rapid exploration often clashes with the performance limitations of high-level languages. While R is the standard for statistical analysis due to its expressivity and ecosystem, extending R capabilities into the realm of GPU-accelerated computing is still challenging. In this talk, I will examine recent trends in GPU acceleration for R, focusing on the transition from traditional CPU-based implementations to CUDA-accelerated kernels, presenting a case study where an API-equivalent rewrite of a core library resulted in speedups of several orders of magnitude.

Speaker: Gabriele Sales (Università di Padova)

8:00 PM → 9:30 PM Dinner

Wed, October 21

9:00 AM → 10:30 AM Consolidation

10:30 AM → 11:00 AM Coffee break

11:00 AM → 11:30 AM Consolidation

11:30 AM → 12:00 PM Students feedback

12:00 PM → 2:00 PM Lunch

2:00 PM → 3:30 PM Exam

3:30 PM → 4:00 PM Coffe break light

4:00 PM → 5:00 PM Discussion

8:00 PM → 10:00 PM RESTAURANT - Social dinner at "La Grotta"

-------> Link Restaurant La Grotta

Thu, October 22

10:00 AM → 10:30 AM DEPARTURES - At 10.00 the shuttle to the train station will wait for you close to the reception