MCCSys-4: 4th Workshop on Memory-Centric Computing Systems, held with HPCA 2026 - 1 February 2026
Keywords
Summary
143 words
Critical Evaluation
Value of the Information & Strength of the Argument
The video provides high value by offering a comprehensive overview of the current state of memory-centric computing, including both fundamental concepts and cutting-edge research. The argumentation is solid, grounded in quantitative data from published studies and real-world measurements. The speakers present compelling evidence for the need to rethink computing architectures, and they address potential challenges and trade-offs. The workshop format allows for diverse perspectives, enriching the discussion.
Scientific Rigor, Source Quality, Title Accuracy
The scientific rigor is high, as the content is based on peer-reviewed research and industry experience. The sources cited are reputable, including papers from top conferences and journals. The title accurately reflects the content, and the workshop is well-organized. The description provides links to recommended readings and related resources, which enhances the credibility of the content.
138 words
Title / Content Match
The title accurately reflects the content: a full-day workshop on memory-centric computing systems, held in conjunction with HPCA 2026.
Quality & Reliability
8/10
The workshop is organized by leading researchers in the field, and the content is based on peer-reviewed research and industry experience. The talks are presented by experts, and the recommended readings are from reputable venues. However, as a workshop recording, some technical details may be condensed, and the presentation style is informal.
Key Moments
Markers derived by PSI from the transcript: the creator did not define chapters.
- Workshop logistics and introduction by Ismail Yuksel.
- Opening remarks by Professor Onur Mutlu, highlighting the memory problem and the need for memory-centric computing.
- Discussion on technology scaling and its impact on memory robustness and energy.
- Presentation of data showing that processors spend most of their time waiting for memory, and energy disparity between computation and data movement.
- Introduction to processing-in-memory (PIM) as a solution, and the concept of computing where data resides.
- Discussion on the challenges of PIM, including programming models and system software support.
- Invited talk on Mastodon, a PIM architecture.
- Presentation on benchmarking real PIM hardware.
- Session on CXL-based computational memory and system software support.
- Discussion on security issues in PIM systems.
Cited Sources
- MCCSys Workshop Website — Official workshop page with agenda and materials.
- A Modern Primer on Processing in Memory — Recommended reading providing a comprehensive overview of PIM.
- Processing Data Where It Makes Sense: Enabling In-Memory Computation — Invited paper discussing the motivation and potential of in-memory computation.
- Processing-in-Memory: A Workload-Driven Perspective — Paper analyzing PIM from a workload perspective.
- Benchmarking a New Paradigm: An Experimental Analysis of a Real Processing-in-Memory Architecture — Tutorial and analysis of the UPMEM PIM architecture.
- Benchmarking Memory-Centric Computing Systems: Analysis of Real Processing-in-Memory Hardware — Paper on benchmarking real PIM hardware.
- SIMDRAM: An End-to-End Framework for Bit-Serial SIMD Computing in DRAM — Paper presenting a framework for bit-serial SIMD computing in DRAM.
- DAMOV: A New Methodology and Benchmark Suite for Evaluating Data Movement Bottlenecks — Paper introducing a benchmark suite for data movement bottlenecks.
- Intelligent Architectures for Intelligent Computing Systems — Invited paper on intelligent architectures.
- RowHammer: A Retrospective — Paper discussing the RowHammer vulnerability and its implications.
Concurring Sources
- Processing-in-Memory: A Workload-Driven Perspective — This paper aligns with the workshop's emphasis on workload-driven analysis of PIM.
- Benchmarking a New Paradigm: An Experimental Analysis of a Real Processing-in-Memory Architecture — The workshop includes talks on benchmarking real PIM hardware, consistent with this paper.
- SIMDRAM: An End-to-End Framework for Bit-Serial SIMD Computing in DRAM — The workshop discusses novel PIM ideas, including bit-serial computing in DRAM.
Dissenting Sources
- No discordant sources identified — The workshop presents a unified view on the benefits of PIM, with no opposing viewpoints presented.
External References
Contribution & Novelties
The workshop provides a comprehensive and up-to-date overview of memory-centric computing, highlighting both the challenges and the latest solutions. It brings together leading researchers and practitioners, offering a unique platform for discussing the paradigm shift from processor-centric to memory-centric architectures. The talks cover a wide range of topics, from fundamental concepts to specific hardware and software implementations, making it a valuable resource for both newcomers and experts in the field.
Pour aller plus loin :
- Processing-in-Memory (PIM) - Wikipedia — Provides a general overview of PIM concepts and history.
- UPMEM - Processing-in-Memory — A commercial PIM technology, often referenced in the workshop.
- Samsung HBM-PIM — An industry example of PIM in high-bandwidth memory.
- CXL - Compute Express Link — A standard for high-speed interconnects that enables memory-centric computing.
- RowHammer - Wikipedia — A DRAM reliability issue discussed in the workshop.
140 words
Radar Profile
The radar profile shows high scores in quantity of information, quality of information, technical level, and global reliability, indicating a comprehensive and technically deep workshop. The balance across these dimensions suggests a well-rounded and authoritative presentation.
💬 No comments were provided for analysis.