|Instructor(s)||Siegel, Andrew (siegela)|
|Fulfills||Core Systems Elective|
*Please note: This is the syllabus from the 2021/22 academic year and subject to change.*
This course focuses on computer achitecture from the perspective of a
high performance programmer. It is aimed at students looking to
demistify the process of coding, profiling, and optimizing algorithms,
with a particular focus on systems geared toward performance intensive
computing: clusters, Multicore CPUs, GPUs, RISC-V, ARM, TPUs, etc. The
fundamental principles of computer architecture will be covered in
depth, but an equal emphasis will be given to hands-on experience in
learning how measure, tune, and report performance on common HPC
resources. While a key outcome of the course is the ability to write
more efficient code, the more enduring goal is to educate students to
"see their code" from the hardware's perspective, broadening and
deepening their understanding of programming with potential benefits
across a broad range of disciplines.
-history of HPC architectures
-efficient programming for cache-based serial processors
-MIMD and cache coherency
-efficient programming for multicore processors
-SIMD and vector instructions
-GPU memory and cores
-TPUs, matrix engines, systolic arrays
Required: B+ or higher in Core Programming. Assignments in this class
can be done in any programming language. Some assignments will involve
manipulating basic snippets of C code, so some basic knowledge of C is
assumed (however, students do not need to be proficient in C to take
MPCS 52011 Introduction to Computer Systems recommended.
This course requires competency in Unix and Linux. Please plan to attend the MPCS Unix Bootcamp (https://masters.cs.uchicago.edu/page/mpcs-unix-bootcamp) or take the online MPCS Unix Bootcamp Course on Canvas.
This class is scheduled at a time that does not conflict with any other classes this quarter.