Palestra: Rethinking Memory System Design for Data-Intensive Computing

Data: 
23/10/2015 - 14:00
Local: 
Sala 351- IC-3,5
 
                       TODOS SÃO BEM VINDOS!

    ======================================================                  
                Universidade Estadual de Campinas - UNICAMP
                          Instituto de Computação - IC
                                      
                                                    
               PALESTRA DA SÉRIE DE SEMINÁRIOS'2015
                                                                     
                   Sexta-feira, 23/10/2015, às 14:00 horas
                  
                                      
                                  Sala 351- IC-3,5
                             
                             
                                                
            
         RETHINKING MEMORY SYSTEM DESIGN FOR
                    DATA-INTENSIVE COMPUTING
                             
           
                 
                             Prof. Onur Mutlu
                  Carnegie Mellon University, USA
                   http://www.ece.cmu.edu/~omutlu
            
                     
           
                                    ABSTRACT
                               


        The memory system is a fundamental performance and energy
        bottleneck in almost all computing systems. Recent system
        design, application, and technology trends that require more
        capacity, bandwidth, efficiency, and predictability out of the
        memory system make it an even more important system
        bottleneck. At the same time, DRAM and flash technologies are
        experiencing difficult technology scaling challenges that make
        the maintenance and enhancement of their capacity,
        energy-efficiency, and reliability significantly more costly
        with conventional techniques.

       In this talk, we examine some promising research and design
       directions to overcome challenges posed by memory
       scaling. Specifically, we discuss three key solution
       directions: 1) enabling new memory architectures, functions,
       interfaces, and better integration of the memory and the rest
       of the system, 2) designing a memory system that intelligently
       employs multiple memory technologies and coordinates memory
       and storage management using non-volatile memory technologies,
       3) providing predictable performance and QoS to applications
       sharing the memory/storage system. If time permits, we might
       also briefly touch upon our ongoing related work in combating
       scaling challenges of NAND flash memory.

       An accompanying paper can be found here:
       http://users.ece.cmu.edu/~omutlu/pub/memory-systems-research_superfri14.pdf

 
       SHORT BIOGRAPHY

       Onur Mutlu is the Strecker Early Career Professor at Carnegie
       Mellon University. His broader research interests are in
       computer architecture and systems, especially in the
       interactions between languages, system software, compilers,
       and microarchitecture, with a major current focus on memory
       systems. He obtained his PhD and MS in ECE from the University
       of Texas at Austin and BS degrees in Computer Engineering and
       Psychology from the University of Michigan, Ann Arbor. Prior
       to Carnegie Mellon, he worked at Microsoft Research, Intel
       Corporation, and Advanced Micro Devices. He was a recipient of
       the IEEE Computer Society Young Computer Architect Award,
       Intel Early Career Faculty Award, faculty partnership awards
       from various companies, including Facebook, Google, HP, Intel,
       IBM, Microsoft and Samsung, a number of best paper
       recognitions at various computer systems venues, and a number
       of “computer architecture top pick” paper selections by the
       IEEE Micro magazine. For more information, please see his
       webpage at http://www.ece.cmu.edu/~omutlu
            
  
          
       
        Responsável: Profa. Ariadne M. B. R. Carvalho
        Email: ariadne@ic.unicamp.br
        Fone: (19) 3521-5864
        Instituto de Computação, Unicamp