Graduate Exam Abstract
PRERANA GHALSASI
M.S. Final
February 8, 2019, 3:00 pm - 5:00 pm
Computer Science Building room number 452
MAX-PLUS MATRIX MULTIPLICATION LIBRARY FOR GPUS - MPMML
Abstract: Max-Plus algebra finds its applications
in discrete event simulations, dynamic
programming, biological sequence
comparisons etc. Although there exist
highly tuned libraries like BLAS[1] and
CUDA Linear Algebra Subprograms
(CuBLAS) [2] for matrix operations,
they implement the standard linear
algebra matrix-multiplication (multiply-
add) for floating points. We found no
standardized library for Max-Plus-
Matrix-Multiplication (MPMM) on
integers. Hence,we developed a highly
tuned parallelized MPMM library
kernel. We chose GPUs as hardware
platform for this work because of their
significantly more parallelism and
arithmetic functional units as
compared to CPUs. We designed this
kernel to be portable across three
successive Nvidia GPU architectures
and it achieves close to 80% of
theoretical machine peak performance
on all of these architectures . We
closely followed the micro-
benchmarking approach described by
Volkov et al. [3] when they contributed
to cuBLAS. This MPMM kernel can be
part of a tropical algebra library for
GPUs and can help speed up
Biological Sequence comparison
applications like BPMax.
References:
By order in abstract
1.http://www.netlib.org/blas/
2.https://docs.nvidia.com/cuda/cublas/i
ndex.html
3.http://www.cs.colostate.edu/~cs675/
a31-volkov.pdf
Adviser: Dr. Sanjay Rajopadhye
Co-Adviser: N/A
Non-ECE Member: Dr. Wim Bohm, Computer Science Dept.
Member 3: Dr. Sudeep Pasricha, Electrical and Computer Engineering Dept.
Addional Members: N/A
Publications:
N/A
Program of Study:
ECE441 Optical Electronics
ECE501 Foundations of Systems Engr
ECE531 Engineering Risk Analysis
ECE560 Found. Fine-Grain Parallelism
ECE561 Design of embedded systems
ECE567 Systems Engr. Architecture
ECE569 MEMS Devices
ECE699 Thesis
