Paul G. Allen School of Computer Science and Engineering
University of Washington
Seattle, WA 98195
Paul G. Allen Center, Room 576
(206) 543-1896 [phone], (206) 616-3804 [fax]
I help run three groups: Sampa on hardware/software systems, SAMPL on machine learning systems and architecture, and MISL on using DNA for information technology applications.
My official UW CSE webpage, and my perpetually semi-up-to-date CV.
An overview of our work on approximate computing can be found here. And try out our language, compiler and benchmarking infrastructure for approximate computing.
Check out the videos from my Hardware/Software Interface class
Take a look at TVM, our recently-released end-to-end stack for deep learning - tvm.ai.
I work on the intersection of computer architecture, programming languages, machine learning and biology. My goals are explore new and better ways to build computing systems.
Selected, recent-ish, publications(Google Scholar Profile, full curated list):
"Puddle: A Dynamic, Error-Correcting, Full-Stack Microfluidics Platform", ASPLOS'19 (to appear).
"Learning to Optimize Tensor Programs", NeurIPS'18.
"TVM:An Automated End-to-End Optimizing Compiler for Deep Learning", OSDI'18.
"Parameter Hub: a Rack-Scale Parameter Server for Distributed Deep Neural Network Training", SoCC'18.
"Architecture Considerations for Stochastic Computing Accelerators", CODES'18.
"LightDB: A DBMS for Virtual Reality Video", PVLDB'18.
"Troubleshooting Transiently-Recurring Errors in Production Systems with Blame-Proportional Logging", USENIX ATC'18.
"Random access in large-scale DNA data storage", Nature Biotechnology, Cover Feature in Mar'18.
"Application Codesign of Near-Data Processing for Similarity Search", IPDPS'18
"MATIC: Learning Around Erros for Efficient Low-Voltage Neural Network Accelerators", DATE'18, Best Paper Award.
"Clustering Billions of Reads for DNA Data Storage", NIPS'17
"Computer Security, Privacy, and DNA Sequencing: Compromising Computers with Synthesized DNA, Privacy Leaks, and More.", Usenix Security'17
"Exploring Computation-Communication Tradeoffs in Camera Systems", IISWC'17
"Customizing Progressive JPEG for Efficient Image Storage", USENIX HotStorage'17
"A Hardware-Friendly Bilateral Solver for Real-Time Virtual Reality Video", HPG'17
"VisualCloud Demonstration: A DBMS for Virtual Reality, SIGMOD'17
"WeLight:Augmenting Interpersonal Communication through Connected Lighting, CHI-LBW'17 (try it out!)
"Approximate Storage for Encrypted and Compressed Videos, ASPLOS'17
"Enabling In-network Computation with a Programmable Network Middlebox, ASPLOS'17
"Energy-Efficient Hybrid Stochastic-Binary Neural Networks for Near-Sensor Computing, DATE'17.
"Disciplined Inconsistency with Consistency Types, SOCC'16.
"A DNA-Based Archival Storage System, ASPLOS'16.
"High-Density Image Storage Using Approximate Memory Cells, ASPLOS'16.
"Optimizing Synthesis with Metasketches”, POPL'16.
"Probability Type Inference for Flexible Approximate Programming”, OOPSLA'15.
"Hardware–Software Co-Design: Not Just a Cliche”, SNAPL'15.
"Latency-Tolerant Software Distributed Shared Memory", USENIX ATC'15.
"Debugging and Monitoring Quality in Approximate Programs", ASPLOS 2015.
"SNNAP: Neural Acceleration on Programmable Logic", HPCA 2015.
"Symbolic Execution of Multithreaded Programs from Arbitrary Program Contexts", OOPSLA 2014.
"General-Purpose Code Acceleration with Limited-Precision Analog Computation", ISCA 2014.
"Expressing and Verifying Probabilistic Assertions", PLDI 2014.
"Low-Level Detection of High-Level Data Races with LARD", ASPLOS 2014.
"Approximate Storage in Solid-State Memories", MICRO 2013.
"EnerJ, the Language of Good-Enough Computing", IEEE Spectrum Feature Article.
"Input-Covering Schedules for Multithreaded Programs", OOPSLA 2013.
"DNA-based Molecular Architecture with Spatially Localized Components", ISCA 2013.
"Cooperative Empirical Failure Avoidance for Multithreaded Programs", ASPLOS 2013.
"Neural Acceleration for General-Purpose Approximate Programs", MICRO 2012 (Selected as IEEE Micro Top Picks and CACM Research Highlights).
"EnerJ: Approximate Data Types for Safe and General Low-Power Computation", PLDI 2011.
"Deterministic Process Groups in dOS", OSDI 2010.
"DMP: Deterministic Shared Memory Multiprocessing", ASPLOS 2009. (Selected for the IEEE Micro Top Picks 2009).
We have released: Grappa runtime system for large-scale irregular applications (e.g., graph analytics), approxbench.org, and ACCEPT, a set of tools and benchmarks for approximate computing research, and TVM/VTA, an end-to-end HW/SW system for deep learning acceleration.
Take a look at the collectively written white-paper on 21st Century computer architecture research, and a vision for the next 15 years of architecture research.
I have the pleasure of working with the following incredible graduate students:James Bornholt (primarily advised by Emina Torlak)
I was born in São Paulo, Brazil.
I received my PhD in Computer Science from University of Illinois at Urbana-Champaign. I got my BEng and MEng in Electrical Engineering from University of São Paulo, Brazil.
I love to cook and eat.
I am very fortunate to have such a happy family.
My (much smarter than me) brother was freezing in Michigan but having fun with aerospace engineering, now he lives just a few miles away!
I am always happy because she