Write a Blog >>
CC 2022
Tue 5 - Wed 6 April 2022 Online conference
Wed 6 Apr 2022 13:15 - 13:30 at CC Virtual Room - Session 6: Performance Optimizations Chair(s): Doru Thom Popovici

Graphics Processing Units (GPUs) are notoriously hard to optimize for manually. What is needed are good automatic code generators and optimizers. Accelerate, Futhark and Lift have demonstrated that using a functional approach is well suited to solve this challenge. Lift, for instance, uses a system of rewrite rules with a multi-stage approach. Algorithmic optimizations are first explored, followed by hardware-specific optimizations such as using shared memory and mapping parallelism. While the algorithmic exploration leads to correct transformed programs by construction, it is not necessarily true for the latter phase. Exploiting shared memory and mapping parallelism while ensuring correct synchronization is a delicate balancing act, and is hard to encode in a rewrite system. Currently, Lift relies on heuristics with ad-hoc mechanisms to check for correctness. This paper proposes to extract parallelization constraints automatically from a functional IR and use a solver to identify valid rewriting. Using a convolutional neural network on a mobile GPU as a use case, this approach matches the performance of the ARM Compute Library GEMM convolution and the TVM-generated kernel consuming between 2x and 3.6x less memory. Furthermore, a speedup of 12x is achieved over the ARM Compute Library direct convolution implementation.

Wed 6 Apr

Displayed time zone: Eastern Time (US & Canada) change

13:00 - 14:00
Session 6: Performance OptimizationsCC Research Papers at CC Virtual Room
Chair(s): Doru Thom Popovici Lawrence Berkeley National Lab
13:00
15m
Paper
Loner: Utilizing the CPU Vector Datapath to Process Scalar Integer Data
CC Research Papers
Armand Behroozi University of Michigan, Sunghyun Park University of Michigan, Scott Mahlke University of Michigan
DOI
13:15
15m
Paper
Mapping Parallelism in a Functional IR through Constraint SatisfactionArtifacts Evaluated – Reusable v1.1Artifacts Available v1.1Results Reproduced v1.1
CC Research Papers
Naums Mogers University of Edinburgh, Lu Li University of Edinburgh, Valentin Radu University of Sheffield, Christophe Dubach McGill University
DOI
13:30
15m
Paper
Software Pre-execution for Irregular Memory Accesses in the HBM Era
CC Research Papers
DOI
13:45
15m
Paper
Efficient Profile-Guided Size Optimization for Native Mobile Applications
CC Research Papers
Kyungwoo Lee Meta, Ellis Hoag Meta, Nikolai Tillmann Meta Platforms, Inc.
DOI