Research, Projects, Topics, and Presentations
Prepared by Professor Gita Alaghband

Purpose:
The purpose of our seminar series is to familiarize ourselves with recent and important research topics and advances in the field. This research oriented study has numerous advantages. To name only a few would be the ability to use your background knowledge for further study in an area, preparation for graduate level research, familiarity with recent advances in the field, self study in an interest area, and the opportunity to communicate research ideas to others.

Projects and their implementations will be carried out either individually or in teams (size to be determined depending on class size.)  You will be able to  develop an experiment, define its scope, define parameters and methods to be experimented with, implement , and  make expert observations.

• You may select a research topic to investigate and present to class in preparation or
  
• you may choose a project topic to implement and experiment with and present to class.  

Your proposals (both for the research and for the project) should be one typed page long describing the research/project briefly complete with references. It should describe what will be covered in your research/project. If you work in teams, then submit one proposal with your team-member names. Describe in as much as possible the areas of responsibilities of your team-members.

Research Paper :
Propose a topic to study in depth. You may investigate any area within the scope of computer architectures beyond what is covered in class.  Plan ahead to allow yourself to study a topic of your interest for research and look into available references for related material. I strongly recommend that you talk to me about your research ideas early.

Your PPT presentation must  include:

• motivation,
  
• background and related work
  
• complete description,
  
• variety of methods used, their success,
  
• results obtained, and
  
• your own conclusion/observations.
• An adequate list of references should be included and referenced in your presentation slides as appropriate.
  

Project (Related to your Research Topic):
Students may choose to do a project implementation instead of an in-depth research topic. Propose a project that you will implement. You may choose to simulate an architectural feature, model a system, compare different strategies on the same or different machines, measure performance of different implementations, develop a set of bench marks, study system affects of specific architecture on performance such as caching, paging, latency, simulate execution pipelines, simulate arithmetic pipelines, etc.  I strongly recommend that you talk to me about your project ideas early.

It is important to be able to explain and anlyze your results for the class.
Your PPT presentation must  include:

• motivation,
  
• background,
  
• complete description,
  
• explain your methods algorithmically and/or using flowcharts
• proof of correctness and a demonstration of the working of your project.
  
• explain your methods, findings, comparisons, analysis of your results and your learning experience.
  
• An adequate list of references should be included with each project.
  
• Keep all formats simple, use MS Word for you documentation.
  
• The following must be provided : copy of the program , test data sets, and instructions for running the program.

Submission to Canvas:

• All presentations PPT, code, etc. for both research and projects must be submitted to Canvas.

Potential Areas of interest: Following are a few suggested topics for you to look into:

• Memory system/hierarchy/storage systems
• Cache memory/strategies
• Parallel architectures,
• Performance modeling,
• Hardware synchronization mechanisms,
• Hardware simulations
• Cache Coherence Protocols in Multicore architectures
• Multicore Architectures
• Multicore memory hierarchy
• Use SimpleScalar tool set  to model and analyze performance. [This tool set from SimpleScalar LLC includes sample simulators ranging from a fast functional simulator to a detailed, dynamically scheduled processor model that supports non-blocking caches, speculative execution, and state-of-the-art branch prediction.]
• Simulate Cache policies/performance experimentation
• Simulate Cache coherence policies and compare performance
• Simulate Various Pipeline processors and compare performance (Superscaler, super pipeline, etc)
• Simulate Brach Prediction policies and compare performance
• Experiments on GPU Memory hierarchy and performance
• Experiments on Multicore Cache coherence and performance
• Experiment on shared memory and cache memory hit/miss and access performance and also overall execution performance due to memory access patterns, optimization, parallelism, etc.
• Experiments on the effects of OS management (process and memory) in multicore
• Memory to memory copying from server to GPU and its effect on performance
• Networking
• Clusters
• Fault Tolerant Designs

Computer Equipments:

•  Our Parallel Distributed Systems Lab, PDS Lab, is available to you for your project implementations and reports. You will have a couple of lab assignments that will require you to use the computing facilities at this lab.
• The department has both Unix and Windows machines available for educational use. On these computers, you should plan on any code that you write executing in user mode (you will only be able to run in user mode on department educational machines).

Presentations:
We will organize and schedule the presentations by topics. Be sure to present proof of correctness and a demonstration of the working of your project.

Use the seminar preparation guide (included in this hand out) to help prepare your presentation.

Seminar Preparation Guide

While preparing for your presentation, keep the following questions in mind. These are provided to give guidance for your presentation effectiveness. After you are prepared, grade yourself with a number between 0-10 on each question and give an overall letter grade (A-F) on each of the 3 areas.

I. Communication [Letter Grade =}

 1. Is the “problem” defined clearly?
 2. Are explanations clear?
 3. Is it clear how the system was developed?
             (i.e.: Language, Architecture, Commercial Software, Algorithms,....)
 4. Is the material well organized? (Both individual and team work)
 5. Effective visual aids?
 6. Creative and interesting?

II. Research [Letter Grade=]

 1. Good knowledge of the field?
 2. Clear comparison between well known related systems or topics?
 3. Has clearly identified the technical successes, failures, and limitations of their system?
 4. Relates clearly and precisely to Parallel Processing problems?
 5. Specific references to material discussed in class?
 6. Appropriate and accurate use of technical words?

III. Analysis   [Letter Grade=]

 1. Problem is well analyzed, bringing out the critical issues?
 2. Use of graphs, charts, statistics, equations, examples, etc.?
 3. Logical structure of the problem is well analyzed?
 4. Distinction made clearly between the use of logic versus the use of probability?