llvm dev - Sep 2004 - [LLVMdev] Hey, who said you could use Obsequi's code!

Just kidding. :-)

Hi, I was just doing a Google search on my name and I
ran across a site that actually had a reference to
Obsequi a program I wrote almost three years ago now.
I thought the program had been forgotten a long time
ago (forgotten might not be the right word, I'm not
sure if it was ever remembered).

Anyway I was just wondering how John Criswell ever ran
across such a low profile program, and why he decided
to include it into the llvm source tree? I would be
interested to know. I hope it wasn't just for a
laugh...

By the way I ran across an article on llvm a couple of
months ago on slashdot. Kudos to the entire group
working on it, it sounds quite interesting.

Nathan Bullock


ps. What is the Ph.D. program like at the University,
what sort of funding is available, etc? I have thought
about doing a Ph.D. but have never been brave enough
to give up my full time job.

====Visit my website at http://bullock.mooo.com

______________________________________________________________________ 
Post your free ad now! http://personals.yahoo.ca

Nathan Bullock wrote:
> Just kidding. :-)
My gosh.  Your subject line nearly gave me a heart attack!
:)
> 
> Hi, I was just doing a Google search on my name and I
> ran across a site that actually had a reference to
> Obsequi a program I wrote almost three years ago now.
> I thought the program had been forgotten a long time
> ago (forgotten might not be the right word, I'm not
> sure if it was ever remembered).
> 
> Anyway I was just wondering how John Criswell ever ran
> across such a low profile program, and why he decided
> to include it into the llvm source tree? I would be
> interested to know. I hope it wasn't just for a
> laugh...
Not at all.

Several years ago, I became very interested in abstract board games. 
This led me to the University of Alberta GAMES Group, among other web 
sites.  As I'm sure you know, Obsequi is available for download from there.

I originally compiled Obsequi with LLVM in hopes that we could use it 
for testing an optimization called Automatic Pool Allocation 
(http://llvm.cs.uiuc.edu/pubs/2002-06-AutomaticPoolAllocation.html). 
Automatic Pool Allocation, we believe, will work very well with programs 
that heavily use linked data structures.  While I haven't checked 
Obsequi's code specifically to see if it uses lots of linked data 
structures, I figured there was a good chance that it (and other board 
game programs) did, as many of these programs use mini-max to find 
solutions, and mini-max can be implemented using linked data structures.

So far, we haven't used Obsequi with Automatic Pool Allocation.  But we 
did add it to the LLVM test suite as a regression test (to ensure LLVM 
can keep compiling it).  Additionally, it's adjustable problem size is 
useful as we often want to get long runtimes out of programs for use in 
benchmarking (it's hard to make comparisons when a program takes 0.01 
seconds to run).
> 
> By the way I ran across an article on llvm a couple of
> months ago on slashdot. Kudos to the entire group
> working on it, it sounds quite interesting.
Thank you for the kudos, and another thank you for writing Obsequi and 
letting us use it!
:)
> 
> Nathan Bullock
> 
> 
> ps. What is the Ph.D. program like at the University,
> what sort of funding is available, etc? I have thought
> about doing a Ph.D. but have never been brave enough
> to give up my full time job.
That I can't really answer (I'm a research programmer here).  Perhaps 
someone else in the group can?

Thanks again for writing.  It's very cool to hear from you.

-- John T.

-- 
*********************************************************************
* John T. Criswell                         Email: criswell at uiuc.edu *
* Research Programmer                                               *
* University of Illinois at Urbana-Champaign                        *
*                                                                   *
* "It's today!" said Piglet. "My favorite day," said
Pooh.          *
*********************************************************************

John Criswell,
> So far, we haven't used Obsequi with Automatic Pool
> Allocation.  But we 
> did add it to the LLVM test suite as a regression
> test (to ensure LLVM 
> can keep compiling it).  Additionally, it's
> adjustable problem size is 
> useful as we often want to get long runtimes out of
> programs for use in 
> benchmarking (it's hard to make comparisons when a
> program takes 0.01 
> seconds to run).
> 
I am glad to be able to help in my own little way.
> Thank you for the kudos, and another thank you for
> writing Obsequi and 
> letting us use it!
> :)
> 
I am a big fan of the GPL. And I am glad that some of
my code is making it into circulation to help pay back
the millions of times that I have used and learned
from others GPLed code. (Although 1 vs 1,000,000 I am
still way behind.)
> Thanks again for writing.  It's very cool to hear
> from you.
> 
> -- John T.
> 
It is cool to see my code being used. Thanks for
putting a smile on my face yesterday.

Nathan Bullock

ps. If you ever have questions about the code feel
free to drop me a line. Although I warn you I haven't
looked at the code in 2-3 years.


====Visit my website at http://bullock.mooo.com

______________________________________________________________________ 
Post your free ad now! http://personals.yahoo.ca