I recall that as versions fo seti@home were released, the amount of "science" done by the client increased. More processing was done to each work unit as the client and the PCs it was running on improved.

Has that continued? Does BOINC perform more processing of the work unit c.f. S@H?

---

<img border="0" src="http://www.boincstats.com/signature/user_28893.gif" />

I don't know if there are different signals that are detected (ie spikes, guassians, tripplets). I'm guessing the client isn't doing more science. However, that doesn't seem to be the purpose of the system.

Boinc was designed to be flexible, E@H is a better example here. They have 2-3 different runs (variations on the app that crunch in different ways), and it's transparent to the user. The program checks/updates the app with every download, so if a new test is needed (or different app all together), every active user is updated within 2 weeks of it going live. Never mind the multiple projects running through a common front-end.

Second Seti under Boinc is better at moving raw data to validated results. WU's are processed 4-6 times on average (boinc) compared to 8+ times under classic (IIRC 10-12x was more the norm).

---

Still looking for something profound or inspirational to place here.

Perhaps it was something else then... I was using command-line client S@H on an old 266MHz PC with processing times from 26-33 hours. My first SETI BOINC on that machine was 69 hours! I have just returned to SETI BOINC with Manager 4.30 (hidden) and processing times look like they will be about the same as the old CL. (I should do a few more WU before I make such a prediction!)

Right now the seti/BOINC app is doing basically the same science as the seti classic. When the astropulse app is up and running it will be using the same data but processing it differently. Also different sources will be possible without user intravention (Parkes, the multibeam reciever) this may have required seperate programs under the old system.

The main change in 'science' that has occured was when seti classic went from 2.xx to 3.xx.

---

BOINC WIKI

BOINCing since 2002/12/8

> Perhaps it was something else then... I was using command-line client S@H on
> an old 266MHz PC with processing times from 26-33 hours. My first SETI BOINC
> on that machine was 69 hours! I have just returned to SETI BOINC with Manager
> 4.30 (hidden) and processing times look like they will be about the same as
> the old CL. (I should do a few more WU before I make such a prediction!)
>
I found that BOINC Seti is a little faster than Classic. On the same host it took about 3:10h average on Classic, with BOINC about 2:35h.
But that doesn´t mean there is no more room for improvement in other aspects of the BOINC client... :]

---