Hunt Engineering DSP; grad school advice

Hello,

1. I am a senior working on a research project involving array processing
for direction of arrival (DOA) estimation. We are using the HEPC9 carrier
board from Hunt Engineering (www.hunteng.co.uk) which contains four Spartan
FPGAs for data collection. Each FPGA module receives data from an IEEE
802.11b antenna card via an ADC. Does anybody have any experience using
tools from Hunt Engineering? Might anyone know how one can compare the data
received by all four antenna cards such that each signal can be passed into
a spectrum estimation algorithm such as MUSIC? Right now, we are using the
software API to write the data into a text file. However, I think that the
processor is so slow that the text file is missing samples. Could anybody
suggest further reading material into how DOA estimation is typically
conducted?

2. Next year, I plan on applying to graduate programs in EE that specialize
in signal processing. Over the past few months, I have been looking closely
into schools which have excellent signal processing departments. After some
research, these are some of the responses I got (in no order): MIT, Georgia
Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
UC-Boulder, and Texas-Austin. Would you attest to the quality of any of the
schools on this list? Can anyone lend some advice regarding these particular
schools as well as other lesser-known schools? Does anyone know any
resources I can use to find out which other graduate schools are
particularly suited for signal processing?

Any advice regarding either of these questions would be much appreciated.

Thank you very much,

Steve

What about University of Maryland? I don't know much about their
signal processing program. I thought they did good work in image
processing.I have worked with a few who graduated from UMD and they
are all excellent engineers. Only thing they lack (compared to those
from name brand school) is a big ego.

"Steve Tjoa." <kiemyang@umd.edu> wrote in message news:<bdtjdo$mne$1@grapevine.wam.umd.edu>...
> Hello,
> 
> 1. I am a senior working on a research project involving array processing
> for direction of arrival (DOA) estimation. We are using the HEPC9 carrier
> board from Hunt Engineering (www.hunteng.co.uk) which contains four Spartan
> FPGAs for data collection. Each FPGA module receives data from an IEEE
> 802.11b antenna card via an ADC. Does anybody have any experience using
> tools from Hunt Engineering? Might anyone know how one can compare the data
> received by all four antenna cards such that each signal can be passed into
> a spectrum estimation algorithm such as MUSIC? Right now, we are using the
> software API to write the data into a text file. However, I think that the
> processor is so slow that the text file is missing samples. Could anybody
> suggest further reading material into how DOA estimation is typically
> conducted?
> 
> 2. Next year, I plan on applying to graduate programs in EE that specialize
> in signal processing. Over the past few months, I have been looking closely
> into schools which have excellent signal processing departments. After some
> research, these are some of the responses I got (in no order): MIT, Georgia
> Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
> UC-Boulder, and Texas-Austin. Would you attest to the quality of any of the
> schools on this list? Can anyone lend some advice regarding these particular
> schools as well as other lesser-known schools? Does anyone know any
> resources I can use to find out which other graduate schools are
> particularly suited for signal processing?
> 
> Any advice regarding either of these questions would be much appreciated.
> 
> Thank you very much,
> 
> Steve

"Steve Tjoa." <kiemyang@umd.edu> wrote in message news:<bdtjdo$mne$1@grapevine.wam.umd.edu>...
> 2. Next year, I plan on applying to graduate programs in EE that specialize
> in signal processing. Over the past few months, I have been looking closely
> into schools which have excellent signal processing departments. After some
> research, these are some of the responses I got (in no order): MIT, Georgia
> Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
> UC-Boulder, and Texas-Austin. Would you attest to the quality of any of the
> schools on this list? Can anyone lend some advice regarding these particular
> schools as well as other lesser-known schools? Does anyone know any
> resources I can use to find out which other graduate schools are
> particularly suited for signal processing?

You don't mention what types of applications you are interested in. 
I would pay some attention to what applications the various schools 
are involved with. As a general rule, I would go for those who have 
long, proven records in *commercial* applications, as opposed to 
government or pure research activities. 

The reason is that there is lots of wishful thinking going on. Research 
and goverment activities don't always get the "reality checks" one 
could wish for, and examples are known where "research" has gone on 
for decades and have degenerated to utter absurdity. Google for 
"MFP/SL" in comp.dsp and you will find one account for such absurdities.

Commercial activities get the "reality checks". Some would say too much so. 
Reality checks, if implemented and working, would ensure that wishful 
thinking is recognized early on, and that project based on such are 
abandoned. In a commercial world, this is a necessity, as any bad 
projects would have an impact on profit. The price may be that sound 
projects are dicarded along the way, but if the alternative is to 
get stuck with MFP/SL type activities, that's a very cheap price to pay.

Those who have long experience with commercial applications know that 
there indeed exist problems that may halt a project, that there are 
technicalities that may not be solvable, etc. Those are the people 
you want to get in touch with. Those are the people who are hard 
to recognize between the imposters and dilletants.

FWIW,

Rune

"Steve Tjoa." <kiemyang@umd.edu> wrote in message news:<bdtjdo$mne$1@grapevine.wam.umd.edu>...
> Hello,
<snipped>
> 
> 2. Next year, I plan on applying to graduate programs in EE that specialize
> in signal processing. Over the past few months, I have been looking closely
> into schools which have excellent signal processing departments. After some
> research, these are some of the responses I got (in no order): MIT, Georgia
> Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
> UC-Boulder, and Texas-Austin. Would you attest to the quality of any of the
> schools on this list? Can anyone lend some advice regarding these particular
> schools as well as other lesser-known schools? Does anyone know any
> resources I can use to find out which other graduate schools are
> particularly suited for signal processing?
> 
> Any advice regarding either of these questions would be much appreciated.
> 
> Thank you very much,
> 
> Steve

Steve,

I imagine you can get someone to attest to the quality of most of the
schools on the list. I will discuss my experience with Georgia Tech
and maybe give you some other things to think about.

I attended GA Tech from 1980-1981, getting an MSEE (specializing in
Signal Processing) in 1981.  At the time I applied I was supporting a
family of 3 (wife not working), had very little money, and did not
live near any of the schools. I had seen a copy of Oppenheim and
Schafer, from MIT and GA Tech, respectively, so I looked there.  MIT
was expensive, out of my price range, assuming I could get in. When I
looked at GA Tech, here was what I found (in 1980 mind you):

1) The school actually had a DSP program. (In 1980 very few schools
did. Most schools had 1 or 2 courses at most.)
2) The school had a great program with some world-class faculty (even
truer now).
3) The school paid better than most for its teaching assistantships
(TAs).
( A pointer: I worked both "1/3-time" and "1/2-time" assistantships.
It actually took significantly less hours for the "1/2-time" job than
the "1/3-time" job, I just got more responsibility (mostly handling
lectures instead of labs) ).
4) The tuition was very low, partially because as part of being a TA I
was given a waiver and allowed to pay in-state tuition (a state
school). The TOTAL cost of my tuition for my degree was $1200.00 (that
is 12 hundred dollars!!!).
5) The school was located in a very nice yet very inexpesive town
(Atlanta). Rents were very low compared to the DC suburbs where I
lived.
6) The school offered both thesis and non-thesis options. Supporting a
family while going to school full-time I needed to be in and out as
fast as possible (had 1 more child at school). I selected the
non-thesis option and was in and out in 12 months.

Another Topic: I would suggest that you look at the job market before
you proceed. The DSP job market (actually the whole hi-tech market) in
the US is very tight today, as it was in the late 1980s-early 1990s.
The Hi-Tech Jobs section of the Washington Post used to be fairly
thick, now it runs about 1 page. There are currently not many job
openings available for a country of this size, and almost all of those
seem to be defense/intelligence-related jobs. It seems that most of
the commercial work may be being sent off-shore to be done with much
cheaper labor, but I don't really know. All I know is that new
commercial jobs have all but disappeared.  The ups and downs of the
DSP job market has caused a lot of people I know a lot of financial
grief.  I feel confident that the job market will recover, but I would
guess the fluctuations will continue.

My two cents worth,

Dirk

Dirk A. Bell
DSP Consultant

Steve Tjoa. wrote:
> Hello,
> 
> 1. I am a senior working on a research project involving array processing
> for direction of arrival (DOA) estimation. We are using the HEPC9 carrier
> board from Hunt Engineering (www.hunteng.co.uk) which contains four Spartan
> FPGAs for data collection. Each FPGA module receives data from an IEEE
> 802.11b antenna card via an ADC. Does anybody have any experience using
> tools from Hunt Engineering? Might anyone know how one can compare the data
> received by all four antenna cards such that each signal can be passed into
> a spectrum estimation algorithm such as MUSIC? Right now, we are using the
> software API to write the data into a text file. However, I think that the
> processor is so slow that the text file is missing samples. Could anybody
> suggest further reading material into how DOA estimation is typically
> conducted?
> 
> 2. Next year, I plan on applying to graduate programs in EE that specialize
> in signal processing. Over the past few months, I have been looking closely
> into schools which have excellent signal processing departments. After some
> research, these are some of the responses I got (in no order): MIT, Georgia
> Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
> UC-Boulder, and Texas-Austin. Would you attest to the quality of any of the
> schools on this list? Can anyone lend some advice regarding these particular
> schools as well as other lesser-known schools? Does anyone know any
> resources I can use to find out which other graduate schools are
> particularly suited for signal processing?

I once had a boss who would not interview MIT grads.  His point was that 
a new engineer out of school took at least 5 years of experience to be 
useful. In that period of time he or she needed extensive training.  His 
primary criteria was that a candidate should be trainable.   "Trainable" 
was the ability to get along with others, including technicians, 
secretaries, and purchasing agents. "Trainable and potentially useful" 
was that one should approach boring and tedious tasks with care and 
creativity.

I suspected that he wasn't' comfortable with people smarter than he was 
at the time but I've met other smarter old timers who essentially agreed 
with the philosophy.

There are probably about 100 universities in the US with very good 
programs that have something to do with signal processing.  I'm looking 
at your list and I'm saying to myself, University of Rhode Island has 
Steve Kay, Don Tufts, Faye Boud...Bartels and a few others.  Is this guy 
saying he's too good for URI?  Ben Frielander is at UC Davis, Is this 
guy too good for UC Davis?  Tom Parks, a real nice guy, went back to his 
alma mater, Cornell, too good for Cornell? J.B Thomas, Stuart Schwartz, 
and Vincent Poor at Princeton?  Harry Van Trees at George Mason?

the list can go on and on, and I haven't touched on Europe, Israel, 
India, or China.




> 
> Any advice regarding either of these questions would be much appreciated.
> 
> Thank you very much,
> 
> Steve
> 
> 

Thank you all very much for your responses. I hope that no one finds this
discussion too tangential to the focus of this newsgroup; although this is
not a technical issue, I feel that this may be a topic of interest to many
of the people who read this newsgroup.

Ichiro: I forgot to mention that I am currently studying at the University
of Maryland, and that it will certainly be on my list of schools. My
research advisor is Dr. K.J. Ray Liu, editor-in-chief of the IEEE Signal
Processing Magazine. Indeed, communications and signal processing is a
strength within our EE department.

Rune: Sorry about that. I am primarily interested in multimedia signal
processing (image, speech, audio). I haven't really narrowed my interests
any further than that. Many of the applications of DSP fascinate me very
much, and for now I am keeping my options open. As for the commercial
relevance of research activities, I have never even considered that aspect
of graduate programs before. I will keep an eye out for that. (Does anybody
else have any comments about this? If so, please share.)

Dirk: Thank you for your input about Georgia Tech. I would also like my
entire degree to cost $1200! As for the DSP job market, I surely do not have
as much inside knowledge about the economic fluctuations as you or others,
but I would not really mind working for a defense agency, like the NSA for
example. That, of course, is a biased opinion being that I'm right here in
College Park, 15 miles from DC. I think the tightness of the job market is
actually a prime reason why I am considering graduate school, plus the fact
that an undergraduate education is inadequate for a job that really involves
DSP.

Stan: My list was certainly not a comprehensive list of "good" graduate
programs. I apologize if this came across as being haughty. In fact, I am
approaching the application process with a great deal of humility, knowing
that the number of graduate applications is higher than ever. I know that it
is unlikely I would be accepted at the top-tier schools given the current
circumstances, and therefore I welcome any suggestions for other colleges
such as UC Davis or URI with great signal processing departments, as you
mentioned.

Thank you very much. I look forward to hearing more responses.

Steve




"Steve Tjoa." <kiemyang@umd.edu> wrote in message
news:bdtjdo$mne$1@grapevine.wam.umd.edu...
> Hello,
>
> 1. I am a senior working on a research project involving array processing
> for direction of arrival (DOA) estimation. We are using the HEPC9 carrier
> board from Hunt Engineering (www.hunteng.co.uk) which contains four
Spartan
> FPGAs for data collection. Each FPGA module receives data from an IEEE
> 802.11b antenna card via an ADC. Does anybody have any experience using
> tools from Hunt Engineering? Might anyone know how one can compare the
data
> received by all four antenna cards such that each signal can be passed
into
> a spectrum estimation algorithm such as MUSIC? Right now, we are using the
> software API to write the data into a text file. However, I think that the
> processor is so slow that the text file is missing samples. Could anybody
> suggest further reading material into how DOA estimation is typically
> conducted?
>
> 2. Next year, I plan on applying to graduate programs in EE that
specialize
> in signal processing. Over the past few months, I have been looking
closely
> into schools which have excellent signal processing departments. After
some
> research, these are some of the responses I got (in no order): MIT,
Georgia
> Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
> UC-Boulder, and Texas-Austin. Would you attest to the quality of any of
the
> schools on this list? Can anyone lend some advice regarding these
particular
> schools as well as other lesser-known schools? Does anyone know any
> resources I can use to find out which other graduate schools are
> particularly suited for signal processing?
>
> Any advice regarding either of these questions would be much appreciated.
>
> Thank you very much,
>
> Steve
>
>

I went to Rice about 8 years back and don't remember much research
like what you describe. The signal processing guys were very
theoretical, proving theorems and all of that. I'd be careful you like
what they are doing. With small schools like Rice and Caltech you
don't have many options. I assume you're talking about PhD research
btw.

If you haven't already, I would suggest you look on the department web
pages to find professor(s) that look interesting and ask them about
what you have in mind as a research area. They are perfectly willing
to give you opinions on other schools as well.

Sorry I've never heard of Hunt nor fooled with fpga's for signal
processing. Having said that I'll take some wild shots. If you are
using printf's for file IO that would be about the slowest method
possible. You could try using fwrite or some binary function if they
provide that or use the debugger to save data, uh if there is a
debugger. You might also try buffering the data so you write more
samples at one time, assuming they use some slow interrupt to write
the data each time.

"Stan Pawlukiewicz" <stanp@nospam_mitre.org> wrote in message
news:bdvf69$r8k$1@newslocal.mitre.org...
> Steve Tjoa. wrote:
> > Hello,
> >
> > 1. I am a senior working on a research project involving array
processing
> > for direction of arrival (DOA) estimation. We are using the HEPC9
carrier
> > board from Hunt Engineering (www.hunteng.co.uk) which contains four
Spartan
> > FPGAs for data collection. Each FPGA module receives data from an IEEE
> > 802.11b antenna card via an ADC. Does anybody have any experience using
> > tools from Hunt Engineering? Might anyone know how one can compare the
data
> > received by all four antenna cards such that each signal can be passed
into
> > a spectrum estimation algorithm such as MUSIC? Right now, we are using
the
> > software API to write the data into a text file. However, I think that
the
> > processor is so slow that the text file is missing samples. Could
anybody
> > suggest further reading material into how DOA estimation is typically
> > conducted?
> >
> > 2. Next year, I plan on applying to graduate programs in EE that
specialize
> > in signal processing. Over the past few months, I have been looking
closely
> > into schools which have excellent signal processing departments. After
some
> > research, these are some of the responses I got (in no order): MIT,
Georgia
> > Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
> > UC-Boulder, and Texas-Austin. Would you attest to the quality of any of
the
> > schools on this list? Can anyone lend some advice regarding these
particular
> > schools as well as other lesser-known schools? Does anyone know any
> > resources I can use to find out which other graduate schools are
> > particularly suited for signal processing?
>
> I once had a boss who would not interview MIT grads.  His point was that
> a new engineer out of school took at least 5 years of experience to be
> useful. In that period of time he or she needed extensive training.  His
> primary criteria was that a candidate should be trainable.   "Trainable"
> was the ability to get along with others, including technicians,
> secretaries, and purchasing agents. "Trainable and potentially useful"
> was that one should approach boring and tedious tasks with care and
> creativity.
>
> I suspected that he wasn't' comfortable with people smarter than he was
> at the time but I've met other smarter old timers who essentially agreed
> with the philosophy.
>
> There are probably about 100 universities in the US with very good
> programs that have something to do with signal processing.  I'm looking
> at your list and I'm saying to myself, University of Rhode Island has
> Steve Kay, Don Tufts, Faye Boud...Bartels and a few others.  Is this guy
> saying he's too good for URI?  Ben Frielander is at UC Davis, Is this
> guy too good for UC Davis?  Tom Parks, a real nice guy, went back to his
> alma mater, Cornell, too good for Cornell? J.B Thomas, Stuart Schwartz,
> and Vincent Poor at Princeton?  Harry Van Trees at George Mason?
>
> the list can go on and on, and I haven't touched on Europe, Israel,
> India, or China.
>
>
>
>
> >
> > Any advice regarding either of these questions would be much
appreciated.
> >
> > Thank you very much,
> >
> > Steve
> >
> >
>

Sounds like your ex-boss didn't want to take on the expense of training up a
graduate. Expense comes in because the grad obviously can't be expected to
be as productive as an experienced engineer.

That all said, however in my experience all jobs require training, and those
which don't train the new engineer (regardless of experience) will cause
that engineer to be a lot less productive than anticipated until he/she gets
up to speed with how things are done at the new job.

I don't think i would want to work for that ex-boss of yours: engineering is
life-long learning, and bosses who fail to realise this cause their
companies to fall behind.

John Smith wrote:
> "Stan Pawlukiewicz" <stanp@nospam_mitre.org> wrote in message
> news:bdvf69$r8k$1@newslocal.mitre.org...
> 
>>Steve Tjoa. wrote:
>>
>>>Hello,
>>>
>>>1. I am a senior working on a research project involving array
> 
> processing
> 
>>>for direction of arrival (DOA) estimation. We are using the HEPC9
> 
> carrier
> 
>>>board from Hunt Engineering (www.hunteng.co.uk) which contains four
> 
> Spartan
> 
>>>FPGAs for data collection. Each FPGA module receives data from an IEEE
>>>802.11b antenna card via an ADC. Does anybody have any experience using
>>>tools from Hunt Engineering? Might anyone know how one can compare the
> 
> data
> 
>>>received by all four antenna cards such that each signal can be passed
> 
> into
> 
>>>a spectrum estimation algorithm such as MUSIC? Right now, we are using
> 
> the
> 
>>>software API to write the data into a text file. However, I think that
> 
> the
> 
>>>processor is so slow that the text file is missing samples. Could
> 
> anybody
> 
>>>suggest further reading material into how DOA estimation is typically
>>>conducted?
>>>
>>>2. Next year, I plan on applying to graduate programs in EE that
> 
> specialize
> 
>>>in signal processing. Over the past few months, I have been looking
> 
> closely
> 
>>>into schools which have excellent signal processing departments. After
> 
> some
> 
>>>research, these are some of the responses I got (in no order): MIT,
> 
> Georgia
> 
>>>Tech, Stanford, Berkeley, Purdue, Michigan, Illinois, Cal Tech, Rice,
>>>UC-Boulder, and Texas-Austin. Would you attest to the quality of any of
> 
> the
> 
>>>schools on this list? Can anyone lend some advice regarding these
> 
> particular
> 
>>>schools as well as other lesser-known schools? Does anyone know any
>>>resources I can use to find out which other graduate schools are
>>>particularly suited for signal processing?
>>
>>I once had a boss who would not interview MIT grads.  His point was that
>>a new engineer out of school took at least 5 years of experience to be
>>useful. In that period of time he or she needed extensive training.  His
>>primary criteria was that a candidate should be trainable.   "Trainable"
>>was the ability to get along with others, including technicians,
>>secretaries, and purchasing agents. "Trainable and potentially useful"
>>was that one should approach boring and tedious tasks with care and
>>creativity.
>>
>>I suspected that he wasn't' comfortable with people smarter than he was
>>at the time but I've met other smarter old timers who essentially agreed
>>with the philosophy.
>>
>>There are probably about 100 universities in the US with very good
>>programs that have something to do with signal processing.  I'm looking
>>at your list and I'm saying to myself, University of Rhode Island has
>>Steve Kay, Don Tufts, Faye Boud...Bartels and a few others.  Is this guy
>>saying he's too good for URI?  Ben Frielander is at UC Davis, Is this
>>guy too good for UC Davis?  Tom Parks, a real nice guy, went back to his
>>alma mater, Cornell, too good for Cornell? J.B Thomas, Stuart Schwartz,
>>and Vincent Poor at Princeton?  Harry Van Trees at George Mason?
>>
>>the list can go on and on, and I haven't touched on Europe, Israel,
>>India, or China.
>>
>>
>>
>>
>>
>>>Any advice regarding either of these questions would be much
> 
> appreciated.
> 
>>>Thank you very much,
>>>
>>>Steve
>>>
>>>
>>
> 
> Sounds like your ex-boss didn't want to take on the expense of training up a
> graduate. Expense comes in because the grad obviously can't be expected to
> be as productive as an experienced engineer.

Actually he did believe in training.  He felt that graduates of "elite" 
schools tended toward the very smart but socially inept type, which he 
thought made training more difficult, particularly in attributes that 
required interacting with non engineers.

> 
> That all said, however in my experience all jobs require training, and those
> which don't train the new engineer (regardless of experience) will cause
> that engineer to be a lot less productive than anticipated until he/she gets
> up to speed with how things are done at the new job.
> 
> I don't think i would want to work for that ex-boss of yours: engineering is
> life-long learning, and bosses who fail to realise this cause their
> companies to fall behind.
> 
Well it was a government job ;)
> 

Stan Pawlukiewicz <stanp@nospam_mitre.org> wrote in message news:<be9ln6$f7h$1@newslocal.mitre.org>...
> >>I once had a boss who would not interview MIT grads.  His point was that
> >>a new engineer out of school took at least 5 years of experience to be
> >>useful. In that period of time he or she needed extensive training.  His
> >>primary criteria was that a candidate should be trainable.   "Trainable"
> >>was the ability to get along with others, including technicians,
> >>secretaries, and purchasing agents. 

Sounds reasonable. Work in any company is a team effort. Some people 
belong in ivory towers. Ivroy towers are only found at universities.

> >>"Trainable and potentially useful"
> >>was that one should approach boring and tedious tasks with care and
> >>creativity.

Attention to detail, knowing the business... this is OK as long as 
the MIT/whatever grad is not supposed to be confined to low level/
boring tasks all his or her carreer. 

> >>I suspected that he wasn't' comfortable with people smarter than he was
> >>at the time but I've met other smarter old timers who essentially agreed
> >>with the philosophy.

Nobody are comfortable having supervising responsibilities for people 
that are obviously smarter than themselves. Making sure that whoever 
they hire will eventually be able to do a good job is a completely 
different thing, though.

> > Sounds like your ex-boss didn't want to take on the expense of training up a
> > graduate. Expense comes in because the grad obviously can't be expected to
> > be as productive as an experienced engineer.

Training graduates is expensive. In university (at least the engineering 
type I went through) the students have all the hard parts already prepared 
for them: There is a listed curriculum, the professors have evaluated and 
chosen appropriate literature, taking a course merely resembles checking 
items in a list. Find the key how to prepare exam problems from a given 
curriculum, and you will do good in anticipating exams. Discovering how 
to answer problems efficiently is even better. Learn both, and you get 
within the top 2% of your class. It takes a lot of un-learning to get 
anything useful out of exam-oriented grads in their first jobs.
 
> Actually he did believe in training.  He felt that graduates of "elite" 
> schools tended toward the very smart but socially inept type, which he 
> thought made training more difficult, particularly in attributes that 
> required interacting with non engineers.

Ivory towers, attention to their own grade sheets... seeds of hubris. 
Actually, I have not known a single person with exceptional grades that 
I would trust with anything except writing exams. I have seen very little
correlation between exceptional grades and capabilities to solve real-world 
problems. The best engineers and students I have worked with, have been in 
the top 20%-40% interval. Below that, and they don't have the basic skills. 
Above that they are too specialized in exam-type problems and situations
to be useful.

> > That all said, however in my experience all jobs require training, and those
> > which don't train the new engineer (regardless of experience) will cause
> > that engineer to be a lot less productive than anticipated until he/she gets
> > up to speed with how things are done at the new job.
> > 
> > I don't think i would want to work for that ex-boss of yours: engineering is
> > life-long learning, and bosses who fail to realise this cause their
> > companies to fall behind.

True. 

>  Well it was a government job ;)

Yes? Do you imply that it's OK that government activities fall behind 
current state-of-the-art?

Rune