Chair News

Published on

We are happy to host two visiting Ph.D. students this summer. Shaokai Lin is a Ph.D. student at the University of California, Berkeley, advised by Prof. Edward A. Lee and Prof. Sanjit Seshia. Shaokai will work with us, particularly Christian Menard, in the next two months on the LinguaFranca project. This will include work on formal verification of LF programs and static scheduling strategies. Caio Vieira, in turn, recently started his Ph.D. at Universidade Federal do Rio Grande do Sul under the supervision of Prof. Antonio Beck and was granted by the Brazilian CAPES-PRINT program for this research stay. Caio will work on hyperdimensional computing and in-memory computing systems in the context of the ScaDS.AI center together with João Paulo Cardoso de Lima, Hamid Farzaneh, and Dr.-Ing. Asif Ali Khan.

Published on

Asif A. Khan was awarded the “3m5 Excellence Award – herausragende Dissertation” for his thesis “Design and Code Optimization for Systems with Next-generation Racetrack Memories”. Racetrack memories (RTMs) are an emerging type of memory with the potential to revolutionize computing systems by virtually removing the memory wall. Asif’s thesis contains a collection of outstanding innovative ideas, methods and tools to optimize systems with racetrack memories. His work on emerging memories, computer architectures and compilers has led to above 25 publications and several 3rd party-funded projects. Asif joined the CC Chair in 2017 and continues to work with us in his new role as Postdoc. The award was handed during this year's OUTPUT.DD science event. We are happy for Asif, congratulations! Photo credits: (c) 3m5.

Published on

This year's International Symposium on Highly Efficient Accelerators and Reconfigurable Technologies (HEART) took place near Kyoto on June 15 and 16. This 13th installment of the event marked its return to in-person attendance after a long stretch of remote operation. HEART connects a closely knit community of both research and industry actors focused on spreading specialized computing hardware to more domains and larger scales. In this context, programmability and innovative models of compute are considered the greatest current obstacles. In line with these challenges, Karl F. A. Friebel presented their work "Base2: An IR for Binary Numeral Types", which takes aim at reconfigurable targets using MLIR.

Published on

We are excited to share that the CC Chair took part once again in the annual Rewe Team Challenge with a total of 8 participants. This year, we formed two teams from our group: CCC-Ofast-run, a mixed team with Asif Ali Khan, Conny Okuma, Hamid Farzaneh, and Jerónimo Castrillón, and CCC-O3-run, an all-male team featuring Chris Okuma, Clément Fournier, João Paulo Cardoso de Lima, and Robert Khasanov. Thanks to the regular training sessions over the past few months, each participant achieved his/her personal records and milestones. Bring on the next challenge!

 

Published on

The CC chair hosted the first thematic workshop on computation-in-memory (CIM) and computation-near-memory (CNM) with great enthusiasm as part of the DFG special priority program on disruptive memory technologies (SPP2377). The workshop featured more than 25 experts and researchers from all across Germany, including a keynote speech by Prof. Thomas Mikolajick and participants from TU Dresden, FAU, KIT, TU Dortmund, TU Ilmenau, and the University of Osnabrück.

The CC Chair was represented by Jeronimo Castrillon, Asif Ali Khan, Hamid Farzaneh, and João Paulo Cardoso de Lima. Besides taking care of the organization, the CC team members presented their recent work on compilation for heterogeneous CIM and CNM systems in the context of their HetCIM project. In total, six presentations packed with lively discussions and knowledge-sharing sessions made the event a memorable experience for all participants.

Published on

It was a great week for compilers and computer architecture in Montreal during the cluster of conferences CC/CGO/HPCA/PPoPP, February 25 - March 1, 2023. Prof. Castrillon gave an overview of research topics for energy-efficient computing at the 1st Workshop on NetZero Carbon Computing (NetZero'23) on Saturday (25.02.). The workshop included insightful talks from experts on sustainable computing, an important line of research for the CC chair. Our highlight of the week was the presentation by Carlos Escuin in a collaborative work with Asif Khan with the title “Compression-Aware and Performance-Efficient Insertion Policies for Long-Lasting Hybrid LLCs”, on the first day of the the 29th IEEE International Symposium on High-Performance Computer Architecture (HPCA'23). Carlos talk was well visited and spawned great discussions. 

Published on

In 2023, Europe's annual industry and research forum HiPEAC entered its 7th phase. This year's HiPEAC Conference took place in Toulouse between January 16-18. Under the new focus of the HiPEAC project, High Performance Edge and Cloud Computing, the CC Chair represented some of its projects at different events.

Published on

The CC chair was present at the 15th IEEE International Symposium on Embedded Multicore/Many-core Systems-on-Chip (MCSoC-2022), held in Penang, Malaysia during December 19 to 22, 2022. The symposium targets emerging topics related to multicore computing architectures and systems. At MCSoC-2022, Julian presented his work on “Parameterizable Mobile Workloads for Adaptable Base Station Optimizations”. This paper introduces a flexible workload generator that allows to describe multiple spatio-temporal traffic scenarios for optimizations on baseband processing systems. Such models are key for the system-level optimizations we are working on in the context of the BMBF projects “E4C” (16ME0426K) and 6G-life (16KISK001K).

Published on

Nesrine Khouzami was selected along with other six women researchers in a competitive process to give a short lightening talk about her research entitled "Problem Solving Environment and Compiler Optimizations for High Performance Particle-Mesh Numerical Simulations" at the Women in HPC workshop, collocated with the International Conference for High-Performance Computing, Networking, Storage, and Analysis (SC22), November 13–18, 2022, at the Kay Bailey Hutchison Convention Center in Dallas, USA. SC is the most prominent and awaited conference in HPC field, this year it hosts together around 11000 attendees from all over the world including researchers, students and HPC-related hardware and software exhibitors under the theme "HPC accelerates".  

Published on

The International Conference on Field-Programmable Logic and Applications (FPL) was the first and remains the largest conference on field-programmable logic and reconfigurable computing, and the 32nd edition of the FPL Conference was hosted by Queen's University Belfast, UK, from the 29th of August to the 2nd of September. Two members of the CC Chair participated in this edition presenting the results of their work. On the 31st of August, João Paulo presented "Data and Computation reuse of CNN in Memristor TCAMs" as a regular paper at the main event. His paper explores data repetition and similarities in image and video recognition tasks, in which ternary content-addressable memories are used to retrieve previously computed activation values. On the 1st of September, Karl Friebel delivered his talk entitled "Creating reusable MLIR abstractions for heterogeneous systems" at the 3rd Workshop on DevOps Support for Cloud FPGA platforms (DevOps). In this talk, Karl shares his experiences with MLIR for heterogeneous systems and shows how his current dialect stack for high-performance reconfigurable systems creates composable, reusable tool flows.