I would call it more of a trickle. My last few cobblestones were just that, going up by one each time. Unfotunately we STILL cannot see pending or results - oh yes its because of the system load.

Since this is more of a Beta test than anything else, would it not make sense to allow sight of these results (we have never had them here only on the Beta site). The users would be able to give better input to the errors that may or may not be happening if they could see the results.
<IMG><IMG><img>

Give it a rest jeeezzzzzz!!!


My name is Pascal and this message has my approval...

It is 10 oclock, do you know what your WUWUs are doing tonight...

> I would call it more of a trickle. My last few cobblestones were just that,
> going up by one each time. Unfotunately we STILL cannot see pending or
> results - oh yes its because of the system load.
>
> Since this is more of a Beta test than anything else, would it not make sense
> to allow sight of these results (we have never had them here only on the Beta
> site). The users would be able to give better input to the errors that may or
> may not be happening if they could see the results.
> [/url]

I had them all. One of our computers reached its download limit today! Yes our magnificent celeron 900Mhz (or there abouts) was downloading and crunching unis in about 30 seconds each. Somehow you know I think either one of our students slipped in a home designed quantum computer processor or the units were a bit iffy. Anyway when I saw it the thing was sitting doing nothing and SETI would not let it have anything to crunch. Oh for the day of classic when I would have crunched 50 units in under 1 hour on such a machine. Anyone else getting iffy units, this was my only machine to do it but the others were crunching earlier units and I don't know if they downloaded any dross.

> Since pendings seem to work well on predictor then maybe it would be a good
> idea to bring them back and work out the problems. I do not see the load on
> this project getting any less, only growing in the future.

George,

Predictor, LHC, and Pirates have all limited their number of participants. So, it should be no surprise that there are no issues there with pending credit. There were no problems in SETI during the beta with pending credit also.

The problem with many of these issues is that the "load" on system components is that the resultant load is usually non-linear. Add 1,000 users and have load "x", add another 1,000 and it is 1.1 * 'x' + 'x'; with the multiplier going up at each additional group.

I don't know what the actual working group size is at this time on SETI, but the last number I had any faith in was well over 30,000 ...

<p>
For BOINC Documentaion: Click Me!

> > Since pendings seem to work well on predictor then maybe it would be a
> good
> > idea to bring them back and work out the problems. I do not see the load
> on
> > this project getting any less, only growing in the future.
>
> George,
>
> Predictor, LHC, and Pirates have all limited their number of participants.
> So, it should be no surprise that there are no issues there with pending
> credit. There were no problems in SETI during the beta with pending credit
> also.
>
> The problem with many of these issues is that the "load" on system components
> is that the resultant load is usually non-linear. Add 1,000 users and have
> load "x", add another 1,000 and it is 1.1 * 'x' + 'x'; with the multiplier
> going up at each additional group.
>
> I don't know what the actual working group size is at this time on SETI, but
> the last number I had any faith in was well over 30,000 ...
>

Paul,

I am aware of the participant and load issues. My point was that the load on seti is only going to increase as time goes on so might as well do it now. Get the bugs worked out. If it goes down due to load issues it would hardly be a surprise, it does that now as I think a few have noticed. As load increases with more and more participants that type of logic would suggest that pending credit would never come back.

George

> I am aware of the participant and load issues. My point was that the load on
> seti is only going to increase as time goes on so might as well do it now. Get
> the bugs worked out. If it goes down due to load issues it would hardly be a
> surprise, it does that now as I think a few have noticed. As load increases
> with more and more participants that type of logic would suggest that pending
> credit would never come back.

George,

George, I agree with you. The problem is that right now we are having to live with a problem called "legacy code" and that is the main stumbling block. We have an infrastructure that needs a change and contrary to some of the nay-sayers, this is a tough issue.

For example, I took the design team and coders to task for the structure of the results table some time in the past. The good news is that most, if not all, of my criticisms were accepted as being valid. The problem is that we are driving down the road at 600 miles an hour... so how do we change the flat tire?

For the massive restructuring of the results table would make changes to all of the modules that read or write to that table, then we have the data that would require migration. So, do we take the project off line while we do this? How long do we test the changes before going live again? and so on, and so on ...

I have thought about this and I know that we could do a side-step into changes, but we still have the issue right now that the system is also near the edge of ability to keep up with current demand. With many other tasts still to go, I am not sure if this is the right time to do a lot of expirimentation with changes to the schema and migration of data.

Since I am only an external observer like the rest of the participants I cannot see what tasks that are being pursued at this time. Though, from appearances we are still doing a lot of client work to solve issues with stability, and enhancements to the web site. All of these are, seem to me, to be the proper things to be doing at this time.

Granted, this issue has to be solved (Paul's opinion) before the closing of the SETI@Home Classic project. But I don't think we are super close to that yet.


<p>
For BOINC Documentaion: Click Me!