Class
|
Topic
|
Reading
|
Milestones
|
Dynamic Branch Prediction 3/29, 3/31
|
Review of pipelining
|
Dynamic branch prediction
|
Sections 2.3 and 2.9; pp. 121-127
|
Predicated execution
|
Appendix G.4
|
Execution Cores 4/5, 4/7, 4/12
|
Superscalars and static scheduling
|
|
|
Overview of dynamic scheduling
|
Section 2.4 to p.92, Section 2.8
Project report guidelines,
and sample project report (available in Homework section)
|
Tomasulo's algorithm
|
Section 2.4, p.92 to its end;
Section 2.5
|
R10000-style dynamic scheduling (a physical register pool)
|
The
Smith/Sohi article for superscalars in a nutshell.
In the R10000 article
read from register mapping,
p. 32, through Register files, p. 35.
|
VLIW Processors 4/14, 4/19
|
Software techniques to exploit ILP
|
Section G.3 covers compiler techniques that we will discuss briefly.
|
|
VLIW machines
|
Section 2.7 and G.6 through p.G-40
It's not necessary to know all the details of this architecture.
Let the lecture be your guide.
|
|
Memory Hierarchy 4/21
|
Advanced caching techniques
|
Section 5.2
|
Entire branch prediction homework due before class, Tuesday, April 19
|
Preparing for the Midterm 4/21
|
|
|
Discuss the midterm and cover any outstanding material.
|
|
|
Midterm 1 4/26
|
Exam is 1.5 hours
|
|
|
Multiprocessors 4/28, 5/3, 5/5, 5/10, 5/12
|
Overview of multiprocessing
|
Section 4.1
|
Cache coherency
|
Section 4.2, Section 4.4
|
First coherency milestone due, Tuesday, May 3, before class.
|
Synchronization
|
Section 4.5
|
|
Consistency models Brandon Lucia, our TA
|
Section 4.6
|
|
Deterministic Parallel Execution
Joe Devietti, CSE Grad Student
|
This paper,
"DMP: Deterministic Shared-Memory Multiprocessing",
International Conference on Architectural Support for
Programming Languages & Operating Systems, 2009, is just FYI.
You're not responsible for reading it.
|
|
Wrap-up multiprocessing
|
|
|
No class 5/17
|
|
|
Second coherency homework due, Tuesday, May 17, at 10:30 am.
|
Multithreading 5/19
|
An Overview of Multithreading and Simultaneous Multithreading
|
Section 3.5 and the SMT paper
|
|
Dataflow Computers 5/24, 5/26
|
Dataflow Machines
|
After reading them over, I don't think any of the papers on the
early dataflow machines are appropriate for classroom use. There
are no general overview papers. So just listen to the lecture.
|
Coherency homework report due Thursday, May 26, before class.
|
Wavescalar architecture and implementation
|
Skim The WaveScalar Architecture and
An overview of the WaveScalar implementation.
Use this to reinforce what we discuss in lecture; don't pay much attention to any
new material covered in these papers.
|
Prepare for the second midterm. 5/31
|
Cover any outstanding material. Course evaluations.
|
|
|
The Wave of the Future 6/2
|
Andrew Putnam, Microsoft Research, formerly a CSE Grad Student
|
This paper,
"Performance and Power of Cache-Based Reconfigurable Computing", is just FYI.
You're not responsible for reading it.
|
Concurrency homework due Thursday, June 2, before class.
|
Adrian Sampson, CSE Grad Student
|
This paper,
EnerJ: Approximate Data Types for Safe and General Low-Power Computation>, is just FYI.
You're not responsible for reading it.
|
Midterm 2 6/6, 10:30
|
Exam is 2 hours
|
|
|
|
|