‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌   ‌
 
Click here to see this online
 
 
 
 

ACE Newsletter June 2023

 
 
 
 

FROM THE DIRECTOR'S DESK

Greetings from the ACE Center for Evolvable Computing, a Semiconductor Research Corporation (SRC) Joint University Microelectronics Program (JUMP 2.0) Center on Systems and Architectures for Distributed Computing. We are 21 PIs and over 80 students from 13 academic institutions across the nation, led from the University of Illinois. The goal of ACE is to devise novel hardware and software technologies for scalable distributed computing that will put the computing ecosystem on a dramatically-steeper curve of performance and energy-efficiency improvement. We focus on several themes: Heterogeneous computing platforms, Distributed evolvable memory and storage, Fine-grained communication and coordination, and Security, privacy, and correctness. We are prototyping our ideas through two cross-cutting demonstrator platforms. Through our work and close collaboration with the participating SRC companies, we hope to make an impact on the microelectronics industry. As we ramp up our efforts, we encourage our partners to keep in touch and attend our technical seminars, which are held on Fridays at Noon CST (even months) and on Tuesdays at 6pm CST (odd months).

Josep Torrellas
Director, ACE
Saburo Muroga Professor of Computer Science
University of Illinois Urbana-Champaign 
 
 
     
 

SPOTLIGHT ON ACE RESEARCH

 
 
 
Figure: PUMICE speedups for the Phoenix benchmark suite using 32k and 131k vector length configurations, normalized to the performance of a decoupled Content Associative Processing Engine (CAPE) of 32k vector length [Caminal et al., HPCA’21]. For reference, a CAPE-131k configuration is also shown.  
 

ACE PI and students propose a tightly integrated CPU-PUM architecture

Existing SIMD extensions in scalar CPUs (e.g., SSE, AVX, etc.) can leverage instruction-level parallelism (ILP) because of their tight integration with the CPU pipeline. However, the vectors they employ are quite short, and this limits their ability to exploit data-level parallelism (DLP). Conversely, processing-using-memory (PUM) accelerators are capable of exploiting massive amounts of DLP, as they typically perform computation on very long vectors (tens of thousands of elements) within the memory itself. Yet PUM architectures are largely decoupled from the CPU itself, thereby limiting their ability to tap the CPU's ILP the way SIMD extensions do. To address this problem, PI Jose Martinez and his students Socrates Wong and Cecilio Tamarit from Cornell propose PUMICE, a tightly integrated CPU-PUM architecture that simultaneously exploits DLP and ILP for very long vector operations. As a result of this tight integration, PUMICE delivers significant performance gains. Pumice will be presented in DAC in July 2023.

 

 
 
 
Figure: Architecture of 𝜇Manycore.  
 

Collaboration between two ACE PIs leads to a novel manycore design for microservice workloads 

The teams of PI Muhammad Shahbaz at Purdue and PI Josep Torrellas at UIUC are collaborating to create μManycore, a chiplet-based manycore architecture specifically tailored for cloud-native microservice workloads. Unlike current multicores, which focus on improving average speedups, the goal of μManycore is to minimize tail latency. μManycore is an architecture composed of small, cache-coherent Villages that eliminate the need for package-wide hardware cache coherence. Clusters of villages are interconnected with an on-package leaf-spine network, which has many redundant, low-hop-count paths between clusters — hence minimizing tail latency. The result is high performance and low tail latency. μManycore will be presented in ISCA in June 2023.

 
     
 

SPOTLIGHT ON ACE STUDENTS

 
 

Ahan Gupta is a first year PhD student working with Charith Mendis at the University of Illinois at Urbana-Champaign. He is interested in building high performance abstractions for machine learning. He did his undergrad in sunny Singapore at the National University of Singapore. Together with ACE students Yixiao Du and Annus Zulfiqar (spotlighted next), Ahan leads the ACE Student Organization, which channels students' initiatives and feedback. His personal website is www.ahangupta.com.  

Yixiao Du is a third-year PhD student at Cornell University, working with Prof. Zhiru Zhang. He earned his bachelor's degree in Microelectronics from the University of Electronic Science and Technology of China. His current interests include hardware acceleration for sparse data processing and high-level synthesis. His personal website is https://yxd97.github.io/.

Annus Zulfiqar is a PhD student and Ross Fellow in the Computer science department at Purdue University, where he is advised by Muhammad Shahbaz. His research focuses on designing the next-generation hardware/software abstractions and architectures for emerging workloads (e.g. in network machine learning). Before joining Purdue, he was a Design Engineer at the Center for Advanced Research in Engineering (CARE), Pakistan, where he developed WIFI/Ethernet/LTE-Capable IoT sensor networks for industrial machine telemetry. He received his bachelor of Electrical Engineering from National University of Sciences and Technology (NUST), Pakistan. His personal website is https://annuszulfiqar2021.github.io/.

 
     
 

AWARDS

 
 
 
Mohammad Alian  
 

ACE PI Received an NSF CAREER Award

ACE PI Mohammad Alian from the University of Kansas School of Engineering has been awarded a National Science Foundation Early Career Development (CAREER) award for work on Near-Memory Datacenter Networks.   CAREER awards are considered among the NSF’s most prestigious, given to about 500 early-career faculty each year with the potential to serve as academic role models in both research and education. https://engr.ku.edu/ku-engineering-professor-wins-nsf-career-award-datacenter-networking-research

 
 
 
Zhiru Zhang  
 

ACE PI Paper Inducted into Hall of Fame

ACE PI Zhiru Zhang, from Cornell University, will have a paper inducted in the ACM SIGDA Technical Committee of FPGAs and Reconfigurable Computing (TCFPGA) Hall of Fame for the class of 2023. The paper is "Application-Specific Instruction Generation for Configurable Processor Architectures", by J. Cong, Y. Fan, G. Han, and Z. Zhang (2004, February), and is published in Proceedings of the 2024 ACM/SIGDA 12th International Symposium on Field Programmable Gate Arrays. (http://hall-of-fame.tcfpga.org/)

 
 
 
Muhammad Shahbaz  
 

IEEE Micro Top Pick Honorable Mention for ACE PI

An IEEE Micro Top Picks (Honorable Mention) was awarded to ACE PI Muhammad Shahbaz in March 2023 for the paper: "Taurus: A Data Plane Architecture for Per-Packet ML", by Tushar Swamy, Alexander Rucker, Muhammad Shahbaz, Ishan Gaur, and Kunle Olukotun from the Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS ’22). Muhammad Shahbaz is the Kevin C. and Suzanne L. Kahn New Frontiers Professor at Purdue University.

 
     
 

Upcoming ACE Events

May 16, 2023   All Hands Meeting (only PIs and students)
June 22, 2023  Spring ACE Meeting. All industry collaborators and SRC personnel are invited
October 4-5, 2023 ACE Review
 
Upcoming Events to Meet ACE Researchers
 
April 23-25, Raleigh NC, IPASS, https://ispass.org/ispass2023/
Amin Mamandipoor
Johnson Umeike
William Won
 
June 4-8, Miami Beach, MLSys, https://mlsys.org/
Christos Kozyrakis
Tushar Krishna
Jianming Tong
 
Jun 11-16, Kyoto, Japan, Symposium on VLSI Technology and Circuits, https://www.vlsisymposium.org/
Zhengya Zhang
 
June 17-21, Orlando, FL,   ISCA/PLDI/FCRC  https://fcrc.acm.org/
Muhammad Alian, Raveeesh Garg, Gerasimos Gerogiannis, Tushar Krishna, Amin Mamandipoor, Jose F. Martinez, Charith Mendis, Antonis Psistakis, Muhammad Shahbaz, Jovan Stojkovic, Radu Teodorescu, Josep Torrellas, Johnson Umeike, Neil Zhao
 
July 9-13, San Francisco, DAC https://www.dac.com/
Jose F. Martinez
Niansong Zhang
Zhiru Zhang
 
July 10-12, Boston, MA, OSDI, https://www.usenix.org/conference/osdi23
Adam Belay
 
Sept. 10-12 Austin, TX, Techcon (https://www.src.org/calendar/e006829/)
Many ACE Attendees
 
 
 
 
footer logo