How to Have a Bad Career as a Stanford Graduate Student
Christos Kozyrakis http://csl.stanford.edu/~christos
September 2004
Outline
Advice for a Bad Career while a Graduate Student Alternatives to a Bad Graduate Career Advice on How to Survive in a Large (Systems) Project If we have time
Advice for Bad Papers/Presentations/Posters Alternatives to Bad Papers/Presentations/Papers
Please interrupt me for questions & comments
This is not a lecture There is not single correct approach to these issues
September 2004
Christos Kozyrakis
The History of this Talk
A variation of Dave Pattersons talk on How to Have a Bad Career in Industry/Academia
Initial version in 1994, targeted to junior faculty http://www.cs.berkeley.edu/~pattrsn/talks/nontech.html
Modifications
Got rid of advice for junior faculty (too early for you) Added some advice of my own
Stanford-specific & big-project advice
Who is Dave?
A computer science professor at Berkeley Research: RISC (Sparc), RAID, NOW (Clusters), IRAM, ROC Research & Teaching awards from Berkeley, ACM, IEEE, 2 seminal books on computer architecture (with John Hennessy)
September 2004
Christos Kozyrakis
Who Am I?
Assistant professor of EE & CS
I teach EE282 (Computer Systems Architecture) Research on architecture, compilation, programming models for parallel systems Stop by 304 Gates, visit http://csl.stanford.edu/~christos, or come to my CS300 talk (10/5, 4.15pm) for more info
My graduate student background
CS Ph.D. at Berkeley (2002)
Major: Computer Systems Minors: Digital Circuits, Management of Technology (Business)
Undergrad abroad (U. of Crete, Greece) Part of a large systems project (IRAM) that implemented prototypes
2-3 professors, many students, multiple companies Worked on architecture, design, simulators, and benchmarking
Took me 6 years
but I got a job offer 1 year before filling my thesis
September 2004
Christos Kozyrakis
Part I: How to Have a Bad Graduate Career
How to Have a Bad Graduate Career (1)
Concentrate on getting good grades
postpone research involvement: might lower GPA
Minimize number and flavors of courses (CS students)
Why take advantage of one of the top departments with a wide range of excellent grad courses Why take advantage of one of the best universities in the world? May affect GPA
Dont trust your advisor
Advisor is only interested in his or her own career, not yours Advisor may try to mentor you, use up time, interfering with GPA
Only work the number of hours per week you are paid!
Even less if possible Dont let master class exploit the workers!
September 2004
Christos Kozyrakis
How to Have a Bad Graduate Career (2)
Concentrate on graduating as fast as possible
Winner is first in class to Ph.D. People only care about that you have a Ph.D. and your GPA, not on what you know
Nirvana: graduating in 3.5 years with a 4.0 GPA!
Dont spend a summer in industry: takes longer
How could industry experience help with selecting Ph.D. topic?
Dont work on large projects: takes longer
Have to talk to others, have to learn different areas Synchronization overhead of multiple people
Dont do a systems Ph.D.: takes longer
Dont go to conferences
It costs money and takes time; youll have plenty of time to learn the field after graduating
Dont waste time polishing writing or talks
Again, that takes time
September 2004 Christos Kozyrakis
How to Have a Bad Graduate Career (3)
Worry too much about the Quals (EE students)
Their only purpose is to make you look stupid & fail They also take time, lower GPA, delay graduation, Make it the goal of your life for the first couple of years Dont take them your first year, more likely you will fail
Never read on your own
Takes time, Whats the advisor for if you have to pick your own readings? May lead to extra work, interaction with others,
Work in the specific area you applied for
E.g. if you applied for networking, stay in networking no matter what Your application is a binding contract Leverage your previous experience, graduate faster!
September 2004
Christos Kozyrakis
How to Have a Bad Graduate Career (4)
Dont worry at all about research funding
It's your advisor's job to keep you in the money Why waste any time with proposals, fellowship applications, etc? Fellowships just make it more difficult to write a 2-page CV
Worry a lot about research funding
If you dont have a fellowship, you are a 2nd class citizen If you are asked to TA, you are a 3rd class citizen
September 2004
Christos Kozyrakis
Part II: Alternatives to a Bad Graduate Career
Alternatives to a Bad Graduate Career (1)
Concentrate on getting good grades?
Reality: need to maintain reasonable grades
Minimize number and flavors of courses (CS students)?
What matters on graduation is letters of recommendation from 3-4 faculty/Ph.D.s who have known you for 5+ years Your last chance to be exposed to new ideas before have to learn them on your own (re: compilers and me) Search for interesting courses beyond your area/department
For minimum school/department requirements Do PhD students ever get a B?
Dont trust your advisor?
EECS, Bioengineering, IP/cyber law, management of technology, etc
Primary attraction of campus vs. research lab is getting to work with grad students Faculty career is judged in large part by success of his or her students Try taking advice of advisor!
September 2004
Christos Kozyrakis
Alternatives to a Bad Graduate Career (2)
Concentrate on graduating as fast as possible?
Your last chance to learn; most learning will be outside the classroom Considered newly minted when finish Ph.D.
Judged on year of Ph.D. vs. year of birth To a person in their 40s/50s, 1-2 more years is roundoff error (27 == 29)
Dont go to conferences?
Chance to see firsthand what the field is like, where its going There are student rates, you can share a room Talk to people in the field in the halls! If your faculty advisor wont pay, then pay it yourself; almost always offer student rates, can often share rooms
Dont waste time polishing writing or talks?
In the marketplace of ideas, the more polish the more likely people will pay attention to your ideas Practice presentation AND answering tough questions
September 2004
Christos Kozyrakis
Alternatives to a Bad Graduate Career (3)
Only work the number of hours per week you are paid?
Industrial Experience? Worry about Quals?
Faculty average is 65-70 hours/week Students should be in that range Organize each day: when most alert? nap? exercise? sleep? When/how often/how long: write, read, program, email? To-do lists: daily, weekly, semester Keep good notes of your ideas, problems, Work hard, play hard
1st or 2nd summer get work experience, or 1 semester off If we did not think you are smart enough, you would not be here Designed to improve breadth See it as an interesting challenge, not as an obstacle Take it your first year, pass it, and forget about it
September 2004
Christos Kozyrakis
Alternatives to a Bad Graduate Career (4)
Never switch field?
Technology changes, opportunities change, interests change Explore opportunities across the department Use classes and class projects to experiment with other fields It gets more difficult to switch fields later on
Dont participate in a large project?
Do it, if your find an interesting project See later part of the talk
Funding
Help your advisor secure funding
Learning how to present research to sponsors is a useful skill A fellowship in your CV looks very good so apply for one if you can
RA/TAships are also good
At the end, your research is what matters If teaching was not important, your advisor would not be here
September 2004
Christos Kozyrakis
The Highlights
1. Swim or Sink
2. Read/learn on your own
Success is determined primarily by the student Faculty set up opportunity, but up to you to leverage it Show initiative! Show initiative! Show initiative! Selecting your PhD topic is as important as your results Fast moving field, dont expect Prof to be at forefront everywhere Read papers, got to seminars/conferences, talk to colleagues Once you know about something, teach your advisor
3. Ask Questions
4. Be honest about your work
Lots of smart people in grad school (and even on the faculty), but dont be intimidated. Either they know and you will learn, or they dont know and you will all learn by trying to determine the answer Its OK to ask a stupid question every now and then Your reputation as a researcher is valuable asset Dont promise something you know you cannot deliver Be honest about your results and progres
Long term, your colleagues or advisor will find out about it
September 2004
Christos Kozyrakis
Choosing an Advisor
Very important decision , dont have to rush into it Tips
Get to know your potential advisor
Talk to advisor, her students, attend their group meetings
A good advisor for is the one
Has the right project or works on right area Has time for you (not too many students) Has the right research style Can provide you with funding (sooner or later) Your characters are compatible
Shop around You can start on a trial basis
Class or independent projects are a good way to do this
Remember: you can switch advisors, but its usually expensive
September 2004
Christos Kozyrakis
Working with your advisor
Your advisor is your academic parent
Most of what you learned growing up applies
Tips
Talk with your advisor
What you are working on and how, needs and worries (short & long term)
Dont be afraid to disagree with them If you keep your advisor happy, she will keep you happy!
For better or worse, you depend on each other
Watch out for
Advisors are also humans
They can make mistakes or be harsh, especially under lots of pressure
Computer engineers/scientists are not known for their social skill Dont lose the respect/trust of your advisor!
September 2004
Christos Kozyrakis
Remember that
September 2004
Christos Kozyrakis
Picking a research topic
Selecting the problem is as important as the research on it You dont have to select it now
Get started with something fun soon! Good research training (techniques, important problems etc) Thesis topic will be clear in ~2-3 years (hopefully)
Dont be surprised if it is very different that initial plan
You have to like your research topic
Only way to keep sane during the dark periods
Make sure that your problem has intermediate milestones
Good for publishing papers Good for checking progress & feasibility
September 2004
Christos Kozyrakis
From research to publishing in few easy steps
September 2004
Christos Kozyrakis
When to Graduate?
Knowledge Expectations Time
September 2004
Christos Kozyrakis
Some Good Reading
Technology And Courage, Ivan Sutherland (SUN)
http://research.sun.com/techrep/Perspectives/smli_ps-1.pdf Be bold; Take chances on hard topics
You & Your Research, Richard Hamming (Bell Labs)
Richard Hamming (Bell Labs) Why do so few researchers make significant contributions? http://www.cs.wisc.edu/~remzi/Postscript/hamming.ps
The Task of a Referee, A. Smith
How to read & evaluate research papers
Tomorrows Professors, Richard Reis (Stanford)
Great advice on Grad School & Future Career Not limited to academics
Peopleware, Demarco & Lister
On teams, projects, & management Is your advisor a bad manager? Very entertaining
September 2004
Christos Kozyrakis
Part III:
How to Survive a Large (Systems) Project
Advantages of Large Projects
Get exposed to interdisciplinary research
Multiple fields: architecture, design, compilation, etc You become an expert in a couple and knowledgeable in 3-4
You develop a working system/prototype
Larger and faster impact, fame Learn what it take to get something to really work
Intellectual stimulation from more people
Faculty and students
Learn to work in a team
Thats how the real world works Make life-long friends & partners
Both industry and universities look for these qualities when they hire
September 2004
Christos Kozyrakis
(Assumed) Disadvantages of Large Projects
Long implementation period Overhead of group interaction Identifying yourself within a large group My opinion and experience: The benefits greatly outweigh the disadvantages Do it, if you find a project that you like Key to surviving a large project:
Show initiative Keep a positive attitude and be flexible Understand the stages the project will go through
Your are neither the first, nor the last one to go through this It all works out in the end
September 2004
Christos Kozyrakis
1. The BrainStorming Stage
Whats going on
All get to contribute to shaping the new idea New idea papers based on initial studies on potential Hot topic of discussion at conferences A lot of excitement
How you feel
We will change the world My famous advisor listens to me I got a paper
Problems
None
My advice
Enjoy it! Prepare for the next phase
September 2004
Christos Kozyrakis
2. The Implementation Stage
Whats going on
Start building the prototype/system/infrastructure Everybody in the group has an assigned task now Start realizing the practical issues/limitations of the ideas Few papers (if any) Some newer project is the hot topic at the conferences/department This is not research Not a single publication for 2 years now There is only 1 thesis in this project for 10 students! I wish I was doing an individual project
How you feel:
My advise
Patience Keep notes of what is difficult, could do better, want to change
September 2004
Christos Kozyrakis
2b. The Stretch or The Really Dark Stage
Whats going on
Final debugging before tape-out or software release, or the first period of operating the system A huge amount of work under the pressure of deadlines
How you feel
(same a in implementation stage) How long has it been since the last time I left the lab? I just want to make it work and never see it again in my life After this, my advisor will ask me to go find a PhD topic!
My advise
Hold on
September 2004
Christos Kozyrakis
3. When it All Comes Together
Characteristics
Working prototype/system Recognition of accomplishment by colleagues Tons of data, a lot of interesting answers, many papers Tons of data, many interesting observations on current/future implementations of various aspects of the system, many new questions and approaches, many theses Individual students follow their own way for the last year(s)
How you feel
I got a thesis; I will graduate; Everybody wants to hire me I know how to make things work (Looking at the prototype) My baby!!!! I dont have enough time to explore all the new ideas/questions I got some great friends
September 2004
Christos Kozyrakis
Part IV: How to Prepare Bad Papers/Presentations/Posters
Writing Commandments for a Bad Career
1. Always make it sound complicated 2. Never define terms, never explain anything 3. Always replace will do with have done
Always publish before you implement
4. Never mention drawbacks of your approach 5. Never reference any papers (besides your own) 6. Never pay attention to the reviewers comments
September 2004
Christos Kozyrakis
In case you did not receive the memo
September 2004
Christos Kozyrakis
In case you did not receive the memo
September 2004
Christos Kozyrakis
7 Talk Commandments for a Bad Career
1. 2. 3. 4. 5. 6. 7. 8. Always make it sound complicated Never illustrate Never be brief Never print large Never use color Never skip slides in a long talk Always include all your equations and figures from the paper Never, ever, practice
September 2004
Christos Kozyrakis
Following all the commandments
We describe the philosophy and design of the control flow machine, and present the results of detailed simulations of the performance of a single processing element. Each factor is compared with the measured performance of an advanced von Neumann computer running equivalent code. It is shown that the control flow processor compares favorably in the program. We present a denotational semantics for a logic program to construct a control flow for the logic program. The control flow is defined as an algebraic manipulator of idempotent substitutions and it virtually reflects the resolution deductions. We also present a bottom-up compilation of medium grain clusters from a fine grain control flow graph. We compare the basic block and the dependence sets algorithms that partition control flow graphs into clusters. A hierarchical macro-control-flow computation allows them to exploit the coarse grain parallelism inside a macrotask, such as a subroutine or a loop, hierarchically. We use a hierarchical definition of macrotasks, a parallelism extraction scheme among macrotasks defined inside an upper layer macrotask, and a scheduling scheme which assigns hierarchical macrotasks on hierarchical clusters. We apply a parallel simulation scheme to a real problem: the simulation of a control flow architecture, and we compare the performance of this simulator with that of a sequential one. Moreover, we investigate the effect of modeling the application on the performance of the simulator. Our study indicates that parallel simulation can reduce the execution time significantly if appropriate modeling is used. We have demonstrated that to achieve the best execution time for a control flow program, the number of nodes within the system and the type of mapping scheme used are particularly important. In addition, we observe that a large number of subsystem nodes allows more actors to be fired concurrently, but the communication overhead in passing control tokens to their destination nodes causes the overall execution time to increase substantially. The relationship between the mapping scheme employed and locality effect in a program are discussed. The mapping scheme employed has to exhibit a strong locality effect in order to allow efficient execution Medium grain execution can benefit from a higher output bandwidth of a processor and finally, a simple superscalar processor with an issue rate of ten is sufficient to exploit the internal parallelism of a cluster. Although the technique does not exhaustively detect all possible errors, it detects nontrivial errors with a worst-case complexity quadratic to the system size. It can be automated and applied to systems with arbitrary loops and nondeterminism.
September 2004
Christos Kozyrakis
7 Poster Commandments for a Bad Career
1. 2. 3. 4. 5. 6. 7. Never illustrate Never be brief Never print large Never use color Never try to attract attention Never prepare a short oral overview Never prepare in advance
September 2004
Christos Kozyrakis
Following all the commandments
We describe the philosophy and design of the control flow machine, and present the results of detailed simulations of the performance of a single processing element. Each factor is compared with the measured performance of an advanced von Neumann computer running equivalent code. It is shown that the control flow processor compares favorably in the program. We present a denotational semantics for a logic program to construct a control flow for the logic program. The control flow is defined as an algebraic manipulator of idempotent substitutions and it virtually reflects the resolution deductions. We also present a bottom-up compilation of medium grain clusters from a fine grain control flow graph. We compare the basic block and the dependence sets algorithms that partition control flow graphs into clusters. We apply a parallel simulation scheme to a real problem: the simulation of a control flow architecture, and we compare the performance of this simulator with that of a sequential one. Moreover, we investigate the effect of modeling the application on the performance of the simulator. Our study indicates that parallel simulation can reduce the execution time significantly if appropriate modeling is used.
How to Do a Bad Poster David Patterson University of California Berkeley, CA 94720
Our compiling strategy is to exploit coarse-grain parallelism at function application level: and the function application level parallelism is implemented by fork-join mechanism. The compiler translates source programs into control flow graphs based on analyzing flow of control, and then serializes instructions within graphs according to flow arcs such that function applications, which have no control dependency, are executed in parallel.
A hierarchical macro-control-flow computation allows them to exploit the coarse grain parallelism inside a macrotask, such as a subroutine or a loop, hierarchically. We use a hierarchical definition of macrotasks, a parallelism extraction scheme among macrotasks defined inside an upper layer macrotask, and a scheduling scheme which assigns hierarchical macrotasks on hierarchical clusters.
We have demonstrated that to achieve the best execution time for a control flow program, the number of nodes within the system and the type of mapping scheme used are particularly important. In addition, we observe that a large number of subsystem nodes allows more actors to be fired concurrently, but the communication overhead in passing control tokens to their destination nodes causes the overall execution time to increase substantially.
The relationship between the mapping scheme employed and locality effect in a program are discussed. The mapping scheme employed has to exhibit a strong locality effect in order to allow efficient execution. We assess the average number of instructions in a cluster and the reduction in matching operations compared with fine grain control flow execution.
Medium grain execution can benefit from a higher output bandwidth of a processor and finally, a simple superscalar processor with an issue rate of ten is sufficient to exploit the internal parallelism of a cluster. Although the technique does not exhaustively detect all possible errors, it detects nontrivial errors with a worst-case complexity quadratic to the system size. It can be automated and applied to systems with arbitrary loops and nondeterminism.
September 2004
Christos Kozyrakis
Part V: (last, I promise) Alternatives to Bad Papers/Presentations/Posters
Alternatives to Bad Papers (1)
Do the opposite of Bad Paper Commandments
Define terms, distinguish will do vs have done, mention drawbacks, real performance, reference other papers.
Find related work through on-line library catalogs
Most papers available on-line ACM and IEEE http://library.stanford.edu
Read the Elements of Style
Best small book on writing Read it often (every 2 years?) Longer alternative Bugs in Writing by Lyn Durpe
September 2004
Christos Kozyrakis
Alternatives to Bad Papers (2)
Follow these steps:
1. 1-page paper outline, with tentative page budget/section 2. Paragraph map
1 topic phrase/sentence per paragraph, hand-drawn figures w. captions
3. (Re)Write draft
Long captions/figure can contain details ~ Scientific American Uses Tables to contain facts that make dreary prose
4. Read aloud, spell check & grammar check
(MS Word; Under Tools, select Grammar, select Options, select technical for writing style vs. standard; select Settings and select)
5. Get feedback from advisor, friends, and critics on draft; go to 3.
September 2004
Christos Kozyrakis
Alternatives to Bad Talks
Do the opposite of Bad Talk Commandments Allocate 2 minutes per slide, leave time for questions Dont over animate Do dry runs with friends/critics for feedback,
Including tough audience questions
Tape a practice talk (audio tape or video tape)
Dont memorize speech, but have notes ready
Bill Tetzlaff, IBM:
Giving a first class job talk is the single most important part of an interview trip. Having someone know that you can give an excellent talk before hand greatly increases the chances of an invitation. That means great conference talks.
September 2004
Christos Kozyrakis
Alternatives to Bad Posters (from Randy Katz)
Do opposite of Bad Poster commandments
Poster tries to catch the eye of person walking by
Answer Five Heilmeier Questions
1. What is the problem you are tackling? 2. What is the current state-of-the-art? 3. What is your key make-a-difference concept or technology? 4. What have you already accomplished? 5. What is your plan for success?
9 page poster might look like
Problem Statement
State-ofthe-Art
Key Concept Accomplish -ment # 2 Summary & Conclusion
Christos Kozyrakis
Accomplish Title and -ment # 1 Visual logo Accomplish Plan for -ment # 3 Success
September 2004
ROC: Recovery-Oriented Computing
ROC Research Group, EECS Division, University of California at Berkeley
Aaron Brown and David Patterson
People are the biggest challenge
Num b e r o f Outa g e s
For more info: http://roc.cs.berkeley.edu
AME is the 21st Century Challenge
Availability
systems should continue to meet quality of service goals despite hardware and software failures
Minute s o f Fa ilure
Recovery-Oriented Computing (ROC) Hypothesis
If a problem has no solution, it may not be a problem, but a fact, not to be solved, but to be coped with over time Shimon Peres
Maintainability
systems should require only minimal ongoing human administration, regardless of scale or complexity: Today, cost of maintenance = 10X cost of purchase
Hu m a n -c om p a n y Hu m a n -e xte rn a l HW fa ilu re s Ac t of Na tu re S W fa ilu re Va n d a lis m
Failures are a fact, and recovery/repair is how we cope with them Improving recovery/repair improves availability
Availability = MTTF (MTTF + MTTR) Since MTTF >> MTTR, 1/10th MTTR just as valuable as 10X MTBF
Evolutionary Growth
systems should evolve gracefully in terms of performance, maintainability, and availability as they are grown/upgraded/expanded
People > 50% outages/minutes of failure
Sources of Failure in the Public Switched Telephone Network, Kuhn; IEEE Computer, 30:4 (Apr 97) FCC Records 1992-1994; Overload (not sufficient switching to lower costs) + 6% outages, 44% minutes
Performance was the 20th Century Challenge
1000X Speedup suggests problems are elsewhere
Since major Sys Admin job is recovery after failure, ROC also helps with maintenance
ROC Principles: (1) Isolation and redundancy
System is partitionable
to isolate faults to enable online repair/recovery to enable online HW growth/SW upgrade to enable operator training/expand experience on portions of real system Techniques: Geographically replicated sites, Sharednothing cluster, Separate address spaces inside CPU
ROC Principles: (2) Online verification
System enables input insertion, output check of all modules (including fault insertion)
to check module operation to find failures faster to test correctness of recovery mechanisms
insert faults and known-incorrect inputs also enables availability benchmarks
ROC Principles: (3) Undo Support
ROC system should offer Undo
to recover from operator errors
undo is ubiquitous in productivity apps should have undo for maintenance
to recover from inevitable SW errors
restore entire system state to pre-error version
to test if proposed solution fixed the problem
discover whether need to try another solution
System is redundant
sufficient HW redundancy/data replication => part of system down but satisfactory service still available enough to survive 2nd failure or more during recovery Techniques: RAID-6; N-copies of data
to discover if warning systems are broken to expose and remove latent errors from each system to train/expand experience of operator Techniques: Global invariants; Topology discovery; Program checking (SW ECC)
to recover from operator training via fault-insertion to replace traditional backup and restore Techniques: Checkpointing; Logging; and time travel (log structured) file systems
ROC Principles: (4) Diagnosis Support
System assists human in diagnosing problems
root-cause analysis to suggest possible failure points
track resource dependencies of all requests correlate symptomatic requests with component dependency model to isolate culprit components
Lessons Learned from Other Fields
1800s: 25% railroad bridges failed! Techniques invented since:
Learn from failures vs. successes Redundancy to survive some failures Margin of safety 3X-6X times calculated load to cover what they dont know
Recovery-Oriented Computing Conclusion
New century needs new research agenda
(and its not performance)
health reporting to detect failed/failing components
failure information, self-test results propagated upwards
Safety now in Civil Engineering DNA
unified status console to highlight improper behavior, predict failure, and suggest corrective action Techniques: Stamp data blocks with modules used; Log faults, errors, failures and recovery methods
Have we been building the computing equivalent of the 19th Century irontruss bridges?
Structural engineering is the science and art of designing and making, with economy and elegance, structures that can safely resist the forces to which they may be subjected
Embrace failure of HW, SW, people and still build systems that work ROC: Significantly reducing Time to Recover/Repair => much greater availability + much lower maintenance costs
What is computer equivalent of safety margin?
Legendary great bird of Arab folklore, the Roc is known to be of such huge size that it can carry off elephants and other great land beasts with its large feet. Sinbad the Sailor encountered such a bird in The Thousand and One Nights.
September 2004
Christos Kozyrakis
Links to More Advice on Papers/Presentations/Posters
Pattersons tips on writing
http://www.cs.berkeley.edu/~pattrsn/talks/writingtips.html
Armandos paper writing & presentation page
http://swig.stanford.edu/~fox/paper_writing.html
The How to Give a Talk Talk
http://www.eecs.harvard.edu/~margo/slides/
The Task of the Referee, A.J Smith
Advice on How to Review Papers http://www.cs.wisc.edu/~markhill/the_task_of_the_referee.pdf
Jason Hongs very extensive list of links
Links on research, writing, presentation, thesis http://www.cs.berkeley.edu/~jasonh/advice.html
September 2004
Christos Kozyrakis
Overall Summary: Enjoy Graduate School
Show Initiative!
Dont wait for advisor (or senior students) to show you what to do Selecting your PhD topic is as important as your results (some people say it is even more important)
Ask questions!
Lots of smart people in grad school (and even on the faculty), but dont be intimidated. Either they know and you will learn, or they dont know and you will all learn by trying to determine the answer Its OK to ask a stupid question every now and then
When to graduate
Knowledge Expectations
September 2004
Time
Christos Kozyrakis
