Systems Design and Implementation Seminar
— 1:00pm
Location:
In Person
-
Newell-Simon 3305
Speaker:
OLIVIA HSU
,
Ph.D. Student, Department of Computer Science, Stanford University
https://cs.stanford.edu/~owhsu
Mapping sparse applications to accelerated computing systems
Sparse applications have become increasingly popular as a way to improve storage and performance, and they have tremendous opportunities for hardware acceleration. However, leveraging sparse hardware acceleration has a host of programmability and usability problems.
This talk presents approaches to solve this problem through the automatic mapping from a high-level sparse programming model to accelerated computing systems. We first address the problem for the entire CPU-accelerator system, by introducing a compiler that maps host sub-computation to external functions (in both hardware and software). Next, we zoom in and introduce two approaches for programming sparse accelerators.
The first approach configures sparse accelerators at the level of their compute and memory units. And the second approach, the Sparse Abstract Machine (SAM), zooms in even more to configure sparse accelerators at the level of detailed sparse dataflows and filtering mechanisms.
Specifically, SAM is an intermediate representation and abstract machine model for targeting sparse tensor algebra to reconfigurable and fixed-function spatial dataflow accelerators. We develop a front-end compiler from high-level languages to SAM, and show how SAM can be leveraged to develop new hardware for sparse tensor algebra.
—
Olivia Hsu is a 5th year computer science PhD student at Stanford University advised by Professor Kunle Olukotun and Professor Fredrik Kjolstad. She currently works on mapping sparse applications to domain-specific architectures, reconfigurable dataflow hardware, and CPUs through compilation. Her research interests broadly include computer architecture, computer and programming systems, compilers, programming models and languages, and digital circuits/VLSI.
Faculty Host: Nathan Beckmann
Event Website:
https://pdl.cmu.edu/SDI/2024/041124.html