The Original 'Ode To Joy' thread - Page 72

NBtarheel_33 · 2011-08-30, 05:12

Quote:

Originally Posted by Christenson

How did you do on last year's projects?

Obviously, this year's theme should be 29..29 cores, 29 tests for Operation Billion Digits, 29,292 TFs, etc.

You'll need some good CUDA cards to do this.....

Not as well as I'd have liked, I'm afraid. I started the year with something like 39 cores...but old borgs faded away, and as of now, I'm left with 13. The full report will come on Friday, but it looks like there will be about 21-22K TFs, 216 P-1's complete (60-60.01M is completely P-1'd as a result of these efforts), *none* of the 46Ms, one 82M which succumbed to TF early on, three 92M LLs (and a fourth which will finish a hair late, on the 10th). Seven of the 28 doublechecks are done - IIRC they were all matches.

I'm seriously thinking about making a first attempt at building a system this year, when the new goodies are announced in 2012 (it is rumored that Ivy Bridge - the 22nm shrink of Sandy Bridge - may be out in the Spring). My trusty 3.06GHz PIV (c. 2006) is struggling a little more with each new version of Windows and each new version of Internet Explorer...and the fans spin more and more mercilessly as the LLs march toward 60M. I will definitely want to include a CUDA-capable GPU. My folks have a 2007 Core 2 Duo with a nice GeForce GPU - alas, it is *one measly rung* below the minimum for CUDA! Hmm, maybe I should build two systems...

I can guess Davieddy's thoughts on the subject...but it has been fun P-1'ing my way through the early 60Ms. I think I will definitely extend that project towards 60.02M and possibly beyond.

29,292 TFs is certainly reachable with a little more focus on TF assignments. Though as we march on, they will be 65-66 bits instead of 64-65 bits. Maybe I will "seize" the 292M range before LMH gets there...

Once I clear out some of the LL inventory I have backed up, I might consider making an attempt on a 100M digit test. Doing this on a borged machine scares me a bit, because you never know when you will be unborged (trust me, losing >95% of a 45M test is bad enough!), but it would probably be fine with regular backups. "First to LL a 500M exponent" ought to get me into the David Hasselhoff subforum, eh?

Decisions, decisions...

Christenson · 2011-08-30, 13:41

Take the cat in your icon, and have him look at your budget.....

Do P-1 on the borged machines that you might lose control over...that leaves you no more than a few days out...though backing up your checkpoint files should be pretty easy, too....much less than even a 5 inch floppy disk, remember those?

wblipp · 2011-09-01, 03:29

I've had a series of phone calls tonight for help understanding the proof that any finite dimensional square matrix can be put into Jordan Normal Form. The greatest confusion was around proving the dimension of the range of $(A-\lambda I)$ is less than n. Most of the people I know that will know what this is and why it was fun are on this forum.

petrw1 · 2011-09-01, 04:40

Any day now it will be unplugged at work as I left there 10 days ago.

davieddy · 2011-09-01, 18:17

Quote:

Originally Posted by wblipp

I've had a series of phone calls tonight for help understanding the proof that any finite dimensional square matrix can be put into Jordan Normal Form. The greatest confusion was around proving the dimension of the range of $(A-\lambda I)$ is less than n. Most of the people I know that will know what this is and why it was fun are on this forum.

Not sure to what extent you are including me, but I can easily
understand why you found the exercise fun!

David

davieddy · 2011-09-06, 09:48

Quote:

Originally Posted by wblipp

I've had a series of phone calls tonight for help understanding the proof that any finite dimensional square matrix can be put into Jordan Normal Form. The greatest confusion was around proving the dimension of the range of $(A-\lambda I)$ is less than n. Most of the people I know that will know what this is and why it was fun are on this forum.

At Oxford, we (physicists etc) were started off with some remedial maths.
Most of which was essential learning.
Eigenvectors and diagonalizing a matrix I found quite pretty, and later
was to learn the importance thereof.

General variables and Lagrange I found fascinating.
Pity that in the only practical examples I encountered, the "Normal Modes" stuck out like a sore thumb.

OTOH as an accomplished 3D animator, I am very greatful to Euler.
I created random asteroids, found their Principal Axes, and displayed
the motion of such.
Especially interesting when the intial motion approximates rotation
about the middle axis of inertia.

David

PS My Classical Mechanics lecturer attempted to demonstrate the
latter phenomenom with a matchbox. He was too spastic and theoretical,
so it wasn't that spectacular but I remember it vividly

wblipp · 2011-09-07, 01:50

This week the review material is Numerical Methods. I am less helpful with this material - about the only classroom thing I remember is Rung-Kutta integration. The standard description uses "k" for some of the variables. In the era when Fortran was the standard for these kinds of computations, many students fell prey to the bug of failing to declare k (or rename it). This gave me an opportunity to tell the Fortran joke "GOD is Real." "Yes, but JESUS is integer." After explaining it, I a got dry sarcastic "Very funny." I said, "It's about as funny as Fortran jokes get."

davieddy · 2011-09-07, 06:37

Someone threw up this blindingly "simple" problem:
a sphere radius R attached to the ceiling via a string of length R.

A regular "tutor" said you should calculate the moment of inertia about
the point of attachment to the ceiling (i.e. assume string/sphere are
a rigid body) etc.

My response got me banned for a week.
I am eternally grateful to Lavalamp for PMing me with the reason he
thought I got barred: he quoted an excerpt from my response
which contained language to make RDSilverman blush

When I returned, I PMed said "tutor". He responded (suitably
apologetically) "I suppose you would have suggested employing the Langrangian".

I didn't reply. Simply gave up Physicsforum as a lost cause!

David

petrw1 · 2011-09-11, 18:54

50,000 Life time GhzDays

100,000 total completions (granted most are TF / LMH)

xilman · 2011-09-11, 20:30

Quote:

Originally Posted by wblipp

This gave me an opportunity to tell the Fortran joke "GOD is Real." "Yes, but JESUS is integer." After explaining it, I a got dry sarcastic "Very funny." I said, "It's about as funny as Fortran jokes get."

The version I learned goes "GOD is REAL, except when declared INTEGER". Equally funny.

Another one went something like

Code:

      INTEGER X11H2I=1
10    FORMAT(X11H2I)=SIN(1.0)
      WRITE(5,10) FORMAT(X11H2I)

which is valid FORTRAN-IV but what happens in practice is entirely up to the compiler. The FORTRAN-IV standard was completely uncompilable because the line 10 above is intrinsically ambiguous.

There were lots of other fun things you could do in that language, including redefining the values of constants such as "1". The program snippet

Code:

      SUBROUTINE FOO(X)
      X=2
      RETURN

      FOO(1)
      WRITE(5,10) 1
10    FORMAT(X2I)

would output "2" on many systems.

Apologies if the examples above aren't completely correct. It's 35 years since I last wrote FORTRAN-IV and I don't have a compiler to hand to test them.

Paul

Christenson · 2011-09-12, 03:49

1986 on the VAX was the last time I used a FORTRAN compiler...I was writing a small FORTRAN compiler in VAX FORTRAN-77 plus...recursion was a bit tricky..

2011-09-01, 04:40	#785
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 11127₈ Posts	My first 5,000 GhzDay Core-2 Duo Any day now it will be unplugged at work as I left there 10 days ago.

2011-09-07, 06:37	#789
davieddy "Lucan" Dec 2006 England 1100101001010₂ Posts	Physicsforum homework Someone threw up this blindingly "simple" problem: a sphere radius R attached to the ceiling via a string of length R. A regular "tutor" said you should calculate the moment of inertia about the point of attachment to the ceiling (i.e. assume string/sphere are a rigid body) etc. My response got me banned for a week. I am eternally grateful to Lavalamp for PMing me with the reason he thought I got barred: he quoted an excerpt from my response which contained language to make RDSilverman blush When I returned, I PMed said "tutor". He responded (suitably apologetically) "I suppose you would have suggested employing the Langrangian". I didn't reply. Simply gave up Physicsforum as a lost cause! David Last fiddled with by davieddy on 2011-09-07 at 06:56

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Original or Remake, which is better?	Uncwilly	Lounge	6	2007-09-15 22:20
Original magic tetrahedron problem (0mtp)	mfgoode	Puzzles	5	2006-04-10 15:51
The original paper on the Crandall/Fagin DWT	Barry Fagin	Math	2	2006-01-04 19:46
monster prime hunting (original in German) Jagd nach der Monsterprimzahl	cyrix	Lounge	6	2004-06-11 08:29
Deutscher Thread (german thread)	TauCeti	NFSNET Discussion	0	2003-12-11 22:12

2011-08-30, 13:41	#783
Christenson Dec 2010 Monticello 5·359 Posts	Take the cat in your icon, and have him look at your budget..... Do P-1 on the borged machines that you might lose control over...that leaves you no more than a few days out...though backing up your checkpoint files should be pretty easy, too....much less than even a 5 inch floppy disk, remember those?

2011-09-01, 03:29	#784
wblipp "William" May 2003 New Haven 2·7·13² Posts	I've had a series of phone calls tonight for help understanding the proof that any finite dimensional square matrix can be put into Jordan Normal Form. The greatest confusion was around proving the dimension of the range of $(A-\lambda I)$ is less than n. Most of the people I know that will know what this is and why it was fun are on this forum.

2011-09-07, 01:50	#788
wblipp "William" May 2003 New Haven 2·7·13² Posts	This week the review material is Numerical Methods. I am less helpful with this material - about the only classroom thing I remember is Rung-Kutta integration. The standard description uses "k" for some of the variables. In the era when Fortran was the standard for these kinds of computations, many students fell prey to the bug of failing to declare k (or rename it). This gave me an opportunity to tell the Fortran joke "GOD is Real." "Yes, but JESUS is integer." After explaining it, I a got dry sarcastic "Very funny." I said, "It's about as funny as Fortran jokes get."

2011-09-11, 18:54	#790
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 3×5×313 Posts	50,000 Life time GhzDays 100,000 total completions (granted most are TF / LMH)

2011-09-12, 03:49	#792
Christenson Dec 2010 Monticello 5·359 Posts	1986 on the VAX was the last time I used a FORTRAN compiler...I was writing a small FORTRAN compiler in VAX FORTRAN-77 plus...recursion was a bit tricky..