This shows you the differences between two versions of the page.

Link to this comparison view

seminars:seminar_10_27_15 [2017/09/20 22:02] (current)
Line 1: Line 1:
 +====== Performance Matters ======
 +Tuesday Oct. 27th, 2015\\
 +Location: CIC Panther Hollow Room\\
 +Time: 4:30PM\\
 +**[[http://​emeryberger.com/​|Emery Berger (UMass)]]**\\
 +Performance clearly matters to users. The most common software update
 +on the AppStore *by far* is "Bug fixes and performance enhancements."​
 +Now that Moore'​s Law Free Lunch has ended, programmers have to work
 +hard to get high performance for their applications. But why is
 +performance so hard to deliver?
 +I will first explain why our current approaches to evaluating and
 +optimizing performance don't work, especially on modern hardware and
 +for modern applications. I will then present two systems that address
 +these challenges. Stabilizer is a tool that enables statistically
 +sound performance evaluation, making it possible to understand the
 +impact of optimizations and conclude things like the fact that the -O2
 +and -O3 optimization levels are indistinguishable from noise
 +(unfortunately true).
 +Since compiler optimizations have largely run out of steam, we need
 +better profiling support, especially for modern concurrent,
 +multi-threaded applications. Coz is a novel "​causal profiler"​ that
 +lets programmers optimize for throughput or latency, and which
 +pinpoints and accurately predicts the impact of optimizations. Coz's
 +approach unlocks numerous previously unknown optimization
 +opportunities. Guided by Coz, we improved the performance of Memcached
 +by 9%, SQLite by 25%, and accelerated six Parsec applications by as
 +much as 68%; in most cases, these optimizations involved modifying
 +under 10 lines of code.
 +This talk is based on work with Charlie Curtsinger published at ASPLOS
 +2013 (Stabilizer) and SOSP 2015 (Coz), where it received a Best Paper
 +{{:​papers:​stabilizer-asplos13.pdf|Stabilizer ASPLOS 2013}} \\
 +{{:​papers:​coz-p184-curtsinger.pdf|Coz SOSP 2015}}
 +Emery Berger is a Professor in the College of Information and Computer
 +Sciences at the University of Massachusetts Amherst, the flagship
 +campus of the UMass system. He graduated with a Ph.D. in Computer
 +Science from the University of Texas at Austin in 2002. Professor
 +Berger has been a Visiting Scientist at Microsoft Research (7 times)
 +and at the Universitat Politecnica de Catalunya (UPC) / Barcelona
 +Supercomputing Center (BSC).
 +Professor Berger’s research spans programming languages, runtime
 +systems, and operating systems, with a particular focus on systems
 +that transparently improve reliability,​ security, and performance. He
 +is the creator of a number of influential software systems including
 +Hoard, a fast and scalable memory manager that accelerates
 +multithreaded applications (used by companies including British
 +Telecom, Cisco, Crédit Suisse, Reuters, Royal Bank of Canada, SAP, and
 +Tata, and on which the Mac OS X memory manager is based); DieHard, an
 +error-avoiding memory manager that directly influenced the design of
 +the Windows 7 Fault-Tolerant Heap; and DieHarder, a secure memory
 +manager that was an inspiration for hardening changes made to the
 +Windows 8 heap.
 +His honors include a Microsoft Research Fellowship, an NSF CAREER
 +Award, a Lilly Teaching Fellowship, a Most Influential Paper Award at
 +OOPSLA 2012, a Google Research Award, a Microsoft SEIF Award, a Best
 +Artifact Award at PLDI, and Best Paper Awards at FAST, OOPSLA, and
 +SOSP; he was named an ACM Senior Member in 2010. Professor Berger is
 +currently a Member of the SIGPLAN Executive Committee and an Associate
 +Editor of the ACM Transactions on Programming Languages and Systems,
 +and will serve as Program Chair for PLDI 2016.
 +**[[seminars| Back to the seminar page]]**