Apache TVM and Open Source ML Acceleration Conference

Wed-Fri, December 15th-17th 2021
Free Virtual Event


Apache TVM is an open-source deep learning compiler stack for CPUs, GPUs, and specialized accelerators. It aims to close the gap between the productivity-focused deep learning frameworks, and the performance- or efficiency-oriented hardware backends.

Thank you for joining us for the 2020 TVM Virtual Conference, December 2-4, 2020!


2020 Program

  Dec 2 - Tutorials    
Time Title Speakers Organization
9:00 Introduction to TVM Chris Hoge OctoML
10:15 TVMC: A Command-Line Driver for TVM Leandro Nunes Arm
11:30 Bring Your Own Codegen to TVM Zhi Chen, Cody Yu Amazon Web Services
1:00 uTVM: Running the TVM Stack on Bare Metal Andrew Reusch OctoML
  Dec 3 - Conference    
Time Title Speakers Organization
9:00 Keynote and Community Update Luis Ceze OctoML
    Tianqi Chen OctoML
    Wilson Yu AMD
    Jem Davies Arm
    Kavitha Prasad Sima
    Yida Wang Amazon Web Services
    Jason Knight OctoML
10:00 Break    
10:15 TVM at AWS Yida Wang, Yao Wang, Yong Wu, Haichen Shen, Wei Xiao Amazon Web Services
10:45 TVM at Imagination Ashutosh Parkhi, Jaydeep Patil Imagination Technologies
10:51 TVM at Synopsys & ITRI Kerwin Tung, Chuck Pilkington, Dexian Li, ITRI (Industrial Technology Research Institute)
10:57 A Generic Method to Utilize Vendor-specific AI Accelerator on Android Mobile for TVM Ming-Yu Hung, Ming-Yi Lai MediaTek Inc.
11:03 Extending TVM to Support Custom ML Hardware Joey Chou, Randy Allen SiMa.ai
11:10 Break    
11:20 Advances in Learning Systems Research Joey Gonzalez UC Berkeley
11:50 Lorien: A Scale-Out System and Database for Auto-Tuning Tensor Programs Cody Yu Amazon Web Services
12:10 Lunch Break    
12:50 TVM at ARM Ramana Radhakrishnan Arm
13:20 TVM at OctoML Jason Knight OctoML
13:40 Break    
14:00 Lightning Talks Day 1 Various Various
  Dec 3 - Lightning Talks    
Time Title Speakers Organization
14:00 Graph-Level Scheduling Optimization with Polyhedral Analysis for Tensor Programs Jie Wang Amazon Web Services
14:06 FeatGraph: A Flexible and Efficient Backend for Graph Neural Network Systems Yuwei Hu Cornell University
14:12 UNIT: Unifying the Compilation of Tensorization Jian Weng UCLA
14:18 RAFT:Accelerating the Tuning Process for AutoTVM HaiWen Fu, XiaoHua Shi, Yuchen Feng Beihang University
14:24 A TVM IR to MLIR Automatic Converter : Bridging TVM with MLIR Ecosystem Jinman Zhao Huawei Technologies Canada/ University of Toronto
14:30 BERT Inference Optimization Using TVM Haichen Shen Amazon Web Services
14:36 Writing Sparse Operators in TIR Tristan Konolige OctoML
14:42 Automatic Differentiation on Tensor Expression Yizhi Liu Amazon Web Services
14:48 Optimizing Automatic Tuning Process of TVM Based on Parallel Genetic Algorithm YuChen Feng, XiaoHua Shi, HaiWen Fu Beihang University
14:54 Overhauling the Onnx Importer to Support Dyanmism Matthew Brookhart, Lily Orth-Smith OctoML
15:00 AdaTune: Adaptive Tensor Program Compilation Made Efficient Menghao Li Microsoft
  Dec 4 - Conference    
Time Title Speakers Organization
9:00 Machine Learning Attacks and how to use KubeFlow Pipelines to Defend David Aronchik Microsoft
9:30 MLIR and MLIR in the TensorFlow Ecosystem Jacques Pienaar Google
10:00 Break    
10:15 Bringing Vitis-AI Hardware Acceleration to TVM for Cloud and Edge Jorn Tuyls Xilinx
10:35 TVM at Alibaba Xiaoyong Liu Alibaba
10:55 End-to-End Performance Assessment of AI Systems with TVM and Virtual Models Michael J. Klaiber Bosch Corporate Research
11:15 Break    
11:30 TVM at Qualcomm Krzysztof Parzyszek Qualcomm
11:50 TVM for Edge Inference at AMD Mei Ye, David Marques AMD
12:10 Lunch Break    
12:50 Ansor : Generating High-Performance Tensor Programs for Deep Learning (An auto-scheduler for TVM) Lianmin Zheng UC Berkeley
13:10 TVM Object System: Multi-language Support for just $19.99 Jared Roesch OctoML
13:20 Break    
13:40 Lightning Talks Day 2 Various Various
  Dec 4 - Lightning Talks    
Time Title Speakers Organization
13:40 AutoTIR - Bringing Automatic Scheduling to TIR Junru Shao OctoML
13:46 Real-time AI on Edge Servers with TVM Kazutaka Morita NTT
13:52 Integrating and Simulating Hardware Accelerators in TVM Luis Vega OctoML
13:58 Tensorization, Scheduling and Allocation for Machine Learning Inference Accelerators Hongbin Zheng, Randy Huang Amazon Web Services
14:04 AArch64 Pre-Quantized Networks Performance Giuseppe Rossini Arm
14:10 Hardware-aware Quantization in TVM Ziheng Jiang, Animesh Jain OctoML, AWS
14:16 Integrating the Arm Ethos-N NPUs into TVM Matthew Barrett Arm
14:22 Running AI WASM Model Securely in Sandbox using TVM and Wasmtime Leon Wang Huawei
14:28 Ethos-U55 : microNPU Support for uTVM Manupa Karunaratne Arm
14:34 Enable TVM QNN on RISC-V with Subword SIMD Computation Jenq-Kuen Lee, Chen Yu-Ri National Tsing-Hua University, Taiwan
14:40 A Generic Framework Based on TVM/VTA for OpenCL-Compatible Cloud Devices Zhang Hao, Li Jaishu Fourth Paradigm Southeast Asia
14:46 HAWQV3: Dyadic Neural Network Quantization Amir Gholami UC Berkely

TVMConf Partners

Community Sponsor

Follow us on Twitter

Past Conferences