[RC5] OGR anyone?

Adam Zilinskas AZilinskas at SolutionsIQ.com
Thu Jan 21 12:26:26 EST 1999


From: Ed Wensell III aka THE StormRaiser
>... I had suggested that d.net offer it's services to companies that 
>have massive databases that need converting to Y2K
The problem there is not so much computing power limited
  as in the brute force DES searches, but data transfer limitations.

Company X has a 10 Gigabyte database, they want to tweak some fields 
  and rebuild the database to 10.01 Gigabytes.

Firstoff the company might not look kindly upon the fact that
  their data is shipped off to 100,000 odd machines to be reassembled
   later.
Second in the act of collecting the rows in the data tables to be parsed out
to 
  others, you could have just changed the fields right then and there.
Lets not even discuss the issues of keeping indexes and relationships
  in sync after a "scatter-gather" of the database.

=====
From: Clayton, Dennis
>That would be really cool and then have a stats system that tells us
>what frame and what pixels each person did on the movie so we can watch
>for our own pxiels/frames! ;-)

I can just see it now, someone jumping up in the theater in the middle
  of the movie; "Did you see that glint in her eye? That was My Pixel!" :-)

============
Ray tracing images, massive FEM (finite element modelling) simulation and
such looks 
  attractive for a massively parallel system like d.net, the problems 
  will be the large amount of data that must be transferred to the 
   clients to do their work (you can't figure out the illumination on
   a pixel without knowing all the light sources impinging on it, including
   the reflections of neighboring pixels).
 There is also problems in keeping synchronization, DES encryption, one
   machine can go check block#159 and another go check block#225 as all the 
   blocks are independent of each other, they are easily computed out 
   of sequence.
  Now for a FEM Simulation, you can't have a bunch of machines working on
    time=10second mark with other still working on time=9second as
    the state of the system at time=10seconds depends upon the previous 
    state which has not completed yet. I guess a clever dispersion of 
    information and regrouping/synchronization algorithms are needed.
    (can't have the big iron clients waiting around because one block 
     of the task was given to a '286 and it is not done yet).

                  Adam Zilinskas
                  Solutions IQ
                  azilinskas at solutionsiq.com

--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list