Chair News

Published on

The CC Chair participated in the 59th IEEE/ACM International Symposium on Microarchitecture (MICRO 2026), a premier forum for computer architecture research. He are happy to have contributed to several of its specialized, co-located workshops. Prof. Jeronimo Castrillon delivered talks at two workshops. At the CHASM workshop (Composable Heterogeneous Architectures for Scalable Machine Learning), he presented "Towards Compilers for Emerging Heterogeneous Systems." He also gave a talk titled "Compilers for In-Memory Computing Systems" at the OASIS-AI workshop (Open Compute Architecture and Software Infrastructure for Heterogeneous AI Systems). Additionally, João Paulo C. de Lima gave an overview of our groups work on"Compiling High-Level Programs for Heterogeneous CIM/CNM Systems” at the RAISE-Arch (Refactoring Heterogeneous Systems Architectures for Scalable and Efficient AI Infrastructure), including joint work on building memory-centric and heterogeneous compilation stacks with Asif Khan and Hamid Farzaneh. João also presented our joint paper "Lift-and-Verify: Kernel-Transparent Equivalence-Based Verification in LLM Compilation" at CHASM, which is collaborative work Anderson Faustino da Silva, Guilherme de Oliveira Silva. It was a great opportunity to engage with the global computer architecture community, syncup with people in our network and make new connections with the community! 

Published on

The CC Chair was well represented at the second edition of the IEEE Cross-Disciplinary Conference on Memory-Centric Computing (CCMCC), held from October 28 to 30, 2026, in Lyon, France. The conference serves as a cross-disciplinary platform bringing together experts across the full computing stack to advance memory-centric computing. Our team contributed multiple presentations showcasing our latest research in this area. João Paulo C. de Lima presented "Noise-Driven Zeroth-Order Optimisation for In-Situ Training on Memristive Networks," which is joint work with Asif Khan. Alexander Kusnezoff, a long-time student at the chair, talked about his work building "A gem5-based Simulation Framework for Computing-in-DRAM," a collaborative effort with Asif and João. Furthermore, Luis Schoppik, who recently joined us as a new PhD student, discussed his ongoing work-in-progress titled "Neural Surrogate Compilation via Hypernetworks." The exchanges at this specialised event were great. We look forward to a continued engagement with the CCMCC community!

Published on

The Chair for Compiler Construction congratulates João Paulo Cardoso de Lima on his appointment as Assistant Professor (W1) of Embedded AI at FAU Erlangen-Nürnberg, effective October 2026.

João joined the Chair in 2022 through ScaDS.AI. His research focused on energy-efficient and high-performance computing through computing-in-memory and near-memory architectures, especially for machine learning and data analytics acceleration. His contributions span a wide range of architecture optimizations, compilation infrastructures, reliability of computing-in-memory systems and simulation of emerging memory technologies. He also collaborated with many members of the Chair and with external partners, supervised Bachelor’s and Master’s students, and helped establish new collaborations. Readers of our news may remember seeing him present the Chair’s papers and tutorials at conferences.

The Chair wishes him success in his new role in Erlangen. The collaboration continues through joint papers in preparation and planned joint project proposals. With Asif Ali Khan now at Paderborn University and Andrés Goéns at TU Darmstadt, we hope this marks the kick-off of a strong network of CCC alumni, working together on emerging computer architectures and the trustworthy compilers they require.

Published on

It is with mixed emotions that the CCC says goodbye to Asif Ali Khan, who is leaving us to join Paderborn University as a Professor of Computer Architecture!

While it is incredibly hard to see him go, we could not be prouder of his remarkable achievement. Asif has been an integral part of the CC Chair for nearly a decade, leaving a lasting mark on our research. Over the years, his work has spanned a wide spectrum of emerging memory technologies, from exotic paradigms like racetrack memory to novel optimizations across DRAM, and emerging non-volatile memories. His contributions in developing pioneering simulation and compilation frameworks for these technologies have been instrumental and are acknowledged and used by the community.

While we will miss his daily presence, insight, and energy at Dresden, we are thrilled for Asif as he steps into this exciting new chapter. We wish him all the best in his new role at Paderborn University, and we look forward to continuing our strong research collaborations across both short- and long-term horizons.

Thank you for everything, Asif, and congratulations!

Published on

On Tuesday, October 20, 2026, our wonderful administrative manager, Conny Okuma, was honored with a certificate celebrating her 25 years of dedicated service to TU Dresden. The certificate was officially handed over by the Chancellor of the university, Jan Gerken, during a special ceremony. Before joining our team, Conny worked with Prof. Wolfgang Nagel from mid 2001 to mid 2015. Since then, she has been an incredible support for our team at the Chair for Compiler Construction! We are deeply grateful for everything she does behind the scenes to keep the chair running smoothly and to foster such a positive environment for all our members. Thank you, Conny, for your outstanding dedication over the years, and congratulations on this remarkable milestone!

Published on

The CC Chair had a highly active and successful week at the 2026 ACM/IEEE Embedded Systems Week (ESWEEK) in Barcelona! Our group's participation kicked off with a tutorial on Sunday “on Hardware-Aware Compilation for Analog, Digital, and Heterogeneous Computing-in-Memory Systems," co-organized with Asif Ali Khan, João Paulo C. de Lima, and Clément Fournier. João and Clément led the session, providing participants with hands-on experience using our compilation framework for near-in-memory computing systems. All material from the tutorial is publicly available online. During the main CASES 2026 conference, Jiahong Bi presented our work, "LAKSA: An MLIR-based Higher-Level Synthesis Compiler for Edge Inference." His research demonstrates significantly better efficiency on resource-constrained FPGAs compared to state-of-the-art approaches. We are pleased to share that the work is available open source.

Later in the week, Prof. Jeronimo Castrillon delivered a talk on "Compilers for In-Memory Computing Systems" during a special CASES session entitled "A Cross-Stack Approach to Efficient and Scalable Generative AI," organized by Shubham Rai from Bosch and other colleagues. The presentation highlighted our ongoing research and synergies within the SPP 2377, the TRR 404 Active-3D project, ScaDS.AI Dresden/Leipzig, and the REC² Excellence Cluster. Finally, Tassilo Tanneberger closed out our group's participation with a presentation at the co-located workshop on Time-Centric Reactive Software. He presented our joint paper with Edward Lee and Erling Jellum, "Macros as Abstractions: Simplifying Code Generation for Lingua Franca,".We are incredibly proud of the team's contributions and thoroughly enjoyed the scientific exchange in Barcelona!

Published on

The CC Chair was represented at the CPS Summer School 2026, held in Alghero, Italy, from 21 to 25 September, by Prof. Jeronimo Castrillon, Jiahong Bi, Robert Khasanov, and Tauseef Ahmed. Seven years after his previous keynote at the school, Prof. Castrillon returned with an invited talk titled “Compilers for Emerging Heterogeneous Systems,” in which he discussed high-level programming and compiler abstractions for high-performance computing, cyber-physical systems, and machine learning. Building on his lecture, Jiahong Bi and Robert Khasanov led a hands-on tutorial introducing LAKSA, an MLIR-based compiler developed within the MYRTUS EU project. Participants followed how LAKSA progressively transforms an application, lowering its computations through MLIR representations, into a dataflow graph and then HLS-oriented code for FPGA implementation. The tutorial also showed how LAKSA’s exported dataflow graph and execution profiles can be used in Mocasin to model and simulate mappings onto heterogeneous platforms and explore system-level trade-offs. We thank the CPS Summer School organisers for the invitation and for hosting an engaging programme in Alghero.

Published on

The CC Chair travelled to São Paulo, Brazil, for this year's Brazilian Symposium on Programming Languages (SBLP), which was co-located with the CBSoft 2026 conference. It was a fantastic event bringing together researchers and practitioners from across the software engineering and programming languages communities. Prof. Jeronimo Castrillon had the honor of delivering the SBLP keynote address, titled "Towards Compilers for Emerging Heterogeneous Systems." Additionally, during the very first technical session of the symposium, Christian Schwarz presented our paper "Structure-Aware Compiler Auto-Tuning." This research is a collaborative effort with Anderson Faustino da Silva, who recently spent two years as a guest professor at our chair. Beyond the great technical program, it was awesome to interact with the welcoming Brazilian PL community! We had a fantastic time exchanging ideas, and catching up with colleagues. A big thank you to the organizers and the entire community for such a memorable event!

Published on

We are very happy to officially welcome Luis Schoppik to the CCC team as a PhD student! Luis is already a familiar face at the CC Chair, having worked closely with our group as a student research assistant since May 2025. He obtained his BSc in Computer Science at the University of Passau in 2023 before joining TU Dresden, where he recently completed his MSc in Computer Science. During his master's studies, Luis shifted focus toward compiler construction and emerging hardware which led to him completing his MSc thesis with us on "Compiler Stack and Optimizations for Neural Surrogate of Programs”. A thesis that was supervised by our team members Asif Khan and João de Lima. Prior to his thesis, he worked with us on Compute-in-Memory (CiM) architectures and gained hands-on experience building compilation pipelines with Clement Fournier. Most recently in our group, he has been implementing a novel CiM simulator in the gem5 framework. At the CC Chair, Luis will conduct his PhD research within the COMETH ERC Consolidator Grant project. We are thrilled that Luis has chosen to continue his academic career with us, and we look forward to his future contributions to compilers for extreme heterogeneity!

Published on

We are happy to welcome Zhiyang Wang to the CCC team! Zhiyang obtained his joint MSc degree in Computer Science from VU Amsterdam and the University of Amsterdam in March 2026. He conducted his master's thesis on formalizing weak memory models in Lean 4 via the LeanCat framework under the supervision of Andres Goens at the University of Amsterdam. We are particularly pleased about this connection, as Andres previously completed his own PhD in our group! Over the past years, Zhiyang has developed a strong background in compiler techniques, multicore hardware/software co-design, and formal verification. His previous work includes developing a Rust-based compiler frontend mapping to CIRCT/MLIR, as well as an internship at the Institute of Software, Chinese Academy of Sciences, where he worked on verifying RISC-V instructions using Sail and Lean 4. At the CC Chair, Zhiyang will contribute to the ERC Consolidator Grant project COMETH. He will apply his expertise to develop system-level software that bridges the gap between complex hardware and programmer productivity, exploring new programming and compilation methodologies for heterogeneous systems and emerging computing paradigms. We look forward to working with Zhiyang and are excited about having him in the team!