Parallel Computer Architecture and Programming : 15-418/618 Spring 2014

Parallel Computer Architecture and Programming (CMU 15-418/618)

From smart phones, to multi-core CPUs and GPUs, to the world's largest supercomputers and web sites, parallel processing is ubiquitous in modern computing. The goal of this course is to provide a deep understanding of the fundamental principles and engineering trade-offs involved in designing modern parallel computing systems as well as to teach parallel programming techniques necessary to effectively utilize these machines. Because writing good parallel programs requires an understanding of key machine performance characteristics, this course will cover both parallel hardware and software design.

[ Policies, Logistics, and Details ]

When We Meet

Mon/Wed 1:30 - 2:50pm

Baker Hall A51 (Giant Eagle Auditorium)

Instructor: Kayvon Fatahalian

Spring 2014 Schedule

Jan 13	Why Parallelism?
Jan 15	A Modern Multi-Core Processor: Forms of Parallelism + Understanding Latency and BW Assignment 1 out
Jan 20	No Class (CMU MLK holiday)
Jan 22	Parallel Programming Models and their Corresponding HW/SW Implementations
Jan 27	Parallel Programming Basics (the parallelization thought process) Assignment 1 due (on Jan 28th)
Jan 29	GPU Architecture and CUDA Programming Assignment 2 out
Feb 3	Performance Optimization I: Work Distribution
Feb 5	Performance Optimization II: Locality, Communication, and Contention
Feb 10	Parallel Application Case Studies
Feb 12	Workload-Driven Performance Evaluation Assignment 2 due (on Feb 13th) Assignment 3 out (on Feb 13th)
Feb 17	Snooping-Based Cache Coherence I
Feb 19	Snooping-Based Cache Coherence II
Feb 24	Directory-Based Cache Coherence
Feb 26	Memory Consistency (+ Course-So-Far Review)
Mar 3	Exam I Assignment 3 due (on Mar 4th)
Mar 5	Scaling a Web Site: Scale-Out Parallelism, Elasticity, and Caching Assignment 4 out (Fri Mar 7)
Mar 10-14	Spring Break
Mar 17	Designing a Basic Snooping-Based Multiprocessor
Mar 19	A More Sophisticated Snooping-Based Multiprocessor
Mar 24	Interconnection Networks
Mar 26	Implementing Synchronization Assignment 4 due (on Mar 27th)
Mar 31	Fine-Grained Synchronization and Lock-Free Data Structures
Apr 2	Transactional Memory Project Proposal Due (Fri April 4)
Apr 7	Heterogeneous Parallelism and Hardware Specialization
Apr 9	Domain-Specific Parallel Programming Systems
Apr 14	Domain-Specific Parallel Programming on Graphs
Apr 16	Addressing the Memory Wall Project Checkpoint Due (Fri April 18)
Apr 21	Scheduling Fork-Join Parallelism
Apr 23	NT Algorithm + Exam II Review
Apr 28	Exam II
May 30	Course Wrap Up and Project Presentation Tips (How to Give a Clear Talk)
May 9	3rd Annual Parallelism Competition Final Projects Due

Assignments and Projects

due Jan 28	Assignment 1: Analyzing Parallel Program Performance on a Quad-Core CPU
due Feb 13	Assignment 2: A Simple Renderer in CUDA
due Mar 4	Assignment 3: Two Algorithms, Many Cores
due Mar 27	Assignment 4: A Simple, Elastic Web Server
weekly	Guidelines and Tips for Making Good Lecture Comments
due May 13	Final Project / Parallelism Competition (proposal due Apr 4)

Acknowledgments

Special thanks to NVIDIA Corporation for their equipment donation to support GPU-computing assignments and projects. Intel Corporation has also provided financial support for course development. Todd Mowry created the original version of 15-418 and much of the structure of his innovative course persists today. Thanks to Matt Pharr and Warren Hunt for technical assistance with ISPC. Alex Reece and Manish Burman developed the course web site. Solomon Boulos is now getting the credit he deserves.