Welcome to MilkyWay@home

Posts by Nicolas

21) Message boards : Number crunching : New App status (Message 6393)
Posted 22 Nov 2008 by Nicolas
Post:
yep we got that working :)

Well, I came here specifically because someone told me "People at MilkyWay are wanting to set up test applications, but don't know how." Maybe he had very outdated info...
22) Message boards : Number crunching : New App status (Message 6391)
Posted 22 Nov 2008 by Nicolas
Post:
do you know what we need to do to set that up server side? i didn't know we could do that.. but for now i'll be sending out limited batches of workunits for testing until i'm confident enough to start up the assimilator and have it automatically generate work.

http://boinc.berkeley.edu/trac/wiki/BetaTest
23) Message boards : Number crunching : New App status (Message 6390)
Posted 22 Nov 2008 by Nicolas
Post:
I'll try and release the code sometime this weekend for people to look at, because I don't think there will be many more changes in terms of fixing any bugs that'll happen. I'm actually looking forward to people looking at the new code and giving some optimization suggestions, because right now i'm not quite sure how much more performance we can squeeze out of the integral loop.

You should make your version control system public, so people can immediately see every change you do.

(I will assume you are using some version control already...)
24) Message boards : Number crunching : New App status (Message 6389)
Posted 22 Nov 2008 by Nicolas
Post:
*psst* *psst* (avoicefromtheshadows)

From my sources (that don't want their names made public) I can tell you that the performance tuning of the new app is finished, but it's not ready to be released.
Right now two major items are not finished yet:
1. native support for GPU computing on the soon-to-be-released Bitboys Avalanche 3D chips
2. port to GNU Hurd (upcoming release in 2009)

This sounds suspiciously sarcastic.
25) Message boards : Number crunching : New App status (Message 6386)
Posted 22 Nov 2008 by Nicolas
Post:
*ROFL* +1...

Hey, every post has a [+] button.

(which nobody uses)
26) Message boards : Number crunching : New App status (Message 6385)
Posted 22 Nov 2008 by Nicolas
Post:
E.g. I've seen code like this in the MW sources:
x = y / z;
v = x * z;
Anyone who made it through school's Mathmatics, will notice that this is the same as v = y.

And in many cases, a compiler can figure that out too, and optimize it out. I know for sure compilers can convert this:
a = sin(x) + b * (sin(x));

into this:
t = sin(x);
a = t + b * t;

avoiding the double calculation of sin(x) (which is relatively expensive to calculate).

It's possible they won't do that in the code you quote, though. For example, if they are integers, that operation isn't v = y, because the first assignment will round off the division result. Or maybe compilers will try hard to make sure if you asked for floating point inaccuracies, you get them :)

And of course, this doesn't remove the fact that optimized-out by the compiler or not, that code is stupid.
27) Message boards : Number crunching : New App status (Message 6384)
Posted 22 Nov 2008 by Nicolas
Post:
And frankly, when looking at the source code one can get some doubts if the MW staff is actually able to put some reasonable code together. You should look for a fresh and ambitious undergraduate or PhD student!

Heh. Go see TSP or Ramsey@Home source code, you'll take your eyes out :)
28) Message boards : Number crunching : New App status (Message 6383)
Posted 22 Nov 2008 by Nicolas
Post:
Personally, I can't ever remember seeing the participants asking for longer, 'tougher' work on any project (at least the ones I run and/or follow).

It usually means making workunits twice as long but doing twice the useful work too. Basically merging multiple workunits in one. You don't lose anything, you just get less server load (the longer each workunit, the less frequently clients contact the server).


Previous 20

©2020 Astroinformatics Group