Archives

September 2001

More data for S@H
It has been a while so how about some news?  First up today is the recent announcement that the S@H project is going to be collecting more data than before for analysis.  Currently when the Arecibo telescope scans the sky, the recorders record a range of radio frequencies onto tapes.  Those tapes are then split into work units which are then sent to the home users to analyze using the S@H clients.  For each spot in the sky the telescope scans, there can be up to 256 work units covering different frequencies (see this page for an explanation).  The current data recorder is kind of old and on the slow side, and limits the frequencies that can be recorded at each spot in the sky.

In the article linked above, Hewlett Packard has donated a new digital data recorder that can record data about 10 times faster than the one they are currently using.  This faster data recorder is going to allow the S@H project record a wider band of frequencies while scanning the sky leading to more data to crunch.  They are still testing the new recorder, and it may take them a couple of months to get it up and running at full speed.  How is this going to affect the home S@H cruncher?  Probably not much, just going to give us a wider variety of frequencies to crunch.  In the article they claim that they are running out of work units to crunch...but I am a bit confused on that point.  My confusion stems from what the different diagrams on the S@H pages are.

This is where the confusion lies.  This calendar shows the "status" of the data collected from the Arecibo telescope, but if you notice, it shows "no data" from March 23 to the present day.  I am not sure if this means there has not been any data collected from the Arecibo telescope since that day, or if they have not received the tapes from Arecibo from the days since March 23.  The confusion is compounded by the graphs shown on this page.  I have always thought that those graphs showed the estimated amount of data being collected by the Arecibo telescope for that day, but I am now thinking that the data shown there are the amount of work units that are created from the data tapes (the data tapes are from random days shown on the calendar linked above.).

One final thing to point out about the announcement...I am not sure how they will handle the extra data collected from the telescope.  It is fine and dandy that they will have more data to crunch, but there are two issues that they need to address with the added data.  The first issue is the equipment that they use to split the data tapes into work units.  AFAIK the splitting of the data is one of their rate limiting steps.  They can only split data so fast.  With up to 10 times more data needed to be split will the current splitter be able to handle it?  The other issue is just how they are going to distribute the new data.  Right now each work unit is crunched an estimated 3.3 times (number is from memory and may not be correct).  Having a work unit done 3 times gives them a way of error checking the data coming back to them.  if result 1 and result 2 are different, on average they have a 3rd result to check for accuracy.  If they have a faster splitter and send out a larger variety of work units then that 3.3 number above will start creeping lower.  On the other hand if they distribute the work units the way they do now, with the added data, they would end up with a huge backlog of work units and/or tapes (10 times the current data is quite a bit more than they have now).  I am not sure what they are going to do...I guess we will have to wait and see.

-Front Page