rights reserved. Presenter: Haichen Shen, Yao Wang Amazon SageMaker Neo, Deep Engine Science Dynamic Model in TVM AWS AI© 2019, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Models with while loop Limitation of TVM/graph runtime ● Cannot compile and run dynamic models© 2019, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Support dynamic model in TVM ● Support Any-dim new runtime for Relay ● Dynamic codegen (WIP) ○ Kernel dispatch for a single op ○ Graph dispatch for a (sub-)graph In collaboration with Jared Roesch, Zhi Chen, Wei Chen© 2019, Amazon Web Services,

[tile] DB libraries, and large-scale graph analytics. Open-source software projects resulting from his work include the Matrix Template Library, the Boost Graph Library and Open MPI. 。Phil Ratzloff *。Distinguished Graphs Are Fundamental Abstractions * Graphs model relationships between elements of a data set *，Without regard to what the data set actually is *。 Graph theoretical (abstract) results can be applied cvMuNKCATIOWS 有二有 灿瑟必帮 和一是/第 手太则光大 A，训四 昱 站 -本| 厅 是，主 半一司、|全之体 ~ 对 /人-看 奋才全 和 全原 和 刁曾 The Future ls Big Graph: on meetalThngs soneaiadman MD] Basic Principles *。The C++ standard library (nee STL) provides

Scott, Principal Engineer at CMU SEI Graph/ML/AI algorithms for large- and small- scale parallel systems. Working on GBTL, a linear algebra-based C++ library for graph analytics.[DISTRIBUTION STATEMENT GraphBLAS community, C API Overview of our draft C++ API How might this interoperate with standard C++, graph library proposal? 4[DISTRIBUTION STATEMENT A] This material has been approved for public release GraphBLAS community, C API Overview of our draft C++ API How might this interoperate with standard C++, graph library proposal? 5[DISTRIBUTION STATEMENT A] This material has been approved for public release

tuples - (Possibly some additional CPOs for ordered multi-dimensional iteration) Shout out to Graph Library Proposal P1709Distributed Dense and Sparse Matrices - Distributed matrix data structures

Memory Model C++11 – C++23About Me: alex.dathskovsky@speedata.io www.linkedin.com/in/alexdathskovsky https://www.cppnext.comAlex Dathskovsky | alex.dathskovsky@speedata.io | www.linkedin.com/in/a

Agenda • Understand the motivation behind Taro • Learn to use the Taro C++ programming model • Dive into the Taro’s coroutine-aware scheduling algorithm • Evaluate Taro on microbenchmarks and a real-world Conclusion 2Agenda • Understand the motivation behind Taro • Learn to use the Taro C++ programming model • Dive into the Taro’s coroutine-aware scheduling algorithm • Evaluate Taro on microbenchmarks and top-down task graph What is Task Graph-based Programming System (TGPS) Code 4• TGPS encapsulates function calls and their dependencies in a top-down task graph What is Task Graph-based Programming

guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2 New graph model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2 files examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 7.6 Graph examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 718 8.8 The binary model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 720 8.9

guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2 New graph model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2 files examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 7.6 Graph examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716 8.8 The binary model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 718 8.9

guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2 New graph model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2 files examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 7.6 Graph examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 706 8.8 The binary model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 708 8.9

guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.2 New graph model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2 files examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 7.6 Graph examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 702 8.8 The binary model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 704 8.9