2 hours, 4 hours, 8 hours, who cares? Can anyone explain to me why this is important? If Berkeley released jumbo WU's that took 5 days to process on high end CPU's, would everyone abandon Seti?


<a>

> 2 hours, 4 hours, 8 hours, who cares? Can anyone explain to me why this is
> important? If Berkeley released jumbo WU's that took 5 days to process on high
> end CPU's, would everyone abandon Seti?
>
>
> <a>
>
If the science is what's important, then efficiency is also important. People (including me) seem to be concerned that machine cycles are being needlessly wasted.

Berkeley could easily clear this up by simply answering all the queries about what's really going on!!


--Bill Z.

I see two possible causes for increased processing time - the new client is performing more calculations (very likely), or the new client indeed is less efficient. If the latter is true, perhaps the developers are busy fixing the problem, explaining their absence from these forums, and if its the former, then everyone is in a snit for no reason at all....


<a>

Size doesn't matter ;)

Actually, if all 3 systems (used to verify result is good) were similar CPUS and took same ammount of time...it would not matter.

Credit_claimed = * * cpu_time_spent.

No, the issue is that previous WUs were taking xx time on peoples machines and now they take xx+yy time.

Is the WU different, nobody in authority has said "were sending out more complex WUs...they will crunch for longer". So people suspect the newest WU crunching code has some coding flaw. And if true then less science is being done, than could be done.

It makes a difference if you are running several projects. Any wasted cycles on SETI will impact the time spent on the other projects. Science or credit, both are impacted on multi-project BOINC machines.

> It makes a difference if you are running several projects. Any wasted cycles
> on SETI will impact the time spent on the other projects. Science or credit,
> both are impacted on multi-project BOINC machines.
>

Then choose the project that maximizes your cycles and stick with it. As for me, I have a particular fondness for Seti, so I devote all CPU cycles to it. Everyone has finite CPU cycles to offer - choose the science that performs to your own expectations and quit the moaning and groaning.

BTW, my reply was not to you personally Purple Rabbit, it is a general expression of my growing annoyance at the lack of patience poeple exhibit in these forums.


<a>

>Can anyone explain to me why this is important?

All of us who don't have 36 machines and aren't in the top 25
would care, If you were penalized credit be cause you are a hi end cruncher
you would complain.


We do it for some kind of competition, bragging rights etc.
We learn from doing it.
We all are searching for E.T.
We don't do this for money.
We do it for science.
We learn from doing it
We need goals and accomplishments.
We love to crunch and complain when we don't get the credit

</img>

---

> I see two possible causes for increased processing time - the new client is
> performing more calculations (very likely), or the new client indeed is less
> efficient. If the latter is true, perhaps the developers are busy fixing the
> problem, explaining their absence from these forums, and if its the former,
> then everyone is in a snit for no reason at all....
>
>
> <a>
>
Well, what does "more calculations" mean? Berkeley certainly hasn't owned up to this. If those "more calculations" are machine cycles wasted on debugging code that was accidentally left turned on, then it is indeed your "former" scenario, but everyone is quite justified in being in a "snit" for a stupid coding mistake like that.

So again, it seems that Berkeley really owes us the courtesy of a simple answer. It wouldn't take that long for them to be polite. The developers wouldn't lose that much time from being "busy fixing the problem".

How about it Berkeley???????


--Bill Z.

> All of us who don't have 36 machines and aren't in the top 25
> would care, If you were penalized credit be cause you are a hi end cruncher
> you would complain.
>
>
> We do it for some kind of competition, bragging rights etc.
> We learn from doing it.
> We all are searching for E.T.
> We don't do this for money.
> We do it for science.
> We learn from doing it
> We need goals and accomplishments.
> We love to crunch and complain when we don't get the credit
>

I appreciate the need for goals and rewards, and I am fortunate to have many PC's at my disposal. That said, I have lost quite literally thousands of WU's since the start of BOINC. Read through every post I have made - not once have I criticized the project and/or its developers. I have simply waited patiently along with everyone else until the 'issue of the moment' is worked out.... it really isn't that hard to do.


<a>

OK, lets look at it another way. Say your computer is capable of crunching a work unit in 3 hours. In one year, you could crunch 8 WU's per day in 365 days or 2920 WU's per year. Now lets say the same work unit takes 6 hours. You could now crunch 1460 WU's per year. What's the difference, you say? Well,, you could leave your computer powered off for half a year at the 3 hour rate and still accomplish the same amount of science. And you would use half the electrical energy to do that work.

Now keep in mind, if the new WU's under 4.05 do more intense or different processing for a scientific reason, fine. That's a whole different situation. We haven't seen anything from the Boinc Set Admin. that says that. If it's a mistake or just a strategy to relieve the network load, it's wrong. Just tell us to back off until hardware/software problems are solved. Or stop the down loadable WU's.

We were told Boinc Seti was ready to transfer to (from Seti Classic). I still have Seti Classic on my machines and Will be glad to switch back if that would help. Only feedback from the project is what is needed. It doesn't seem that complex to me. Maybe I'm missing something?

I guess it's like using a 120 watt light bulb where a 60 watt bulb would suffice.

> I see two possible causes for increased processing time - the new client is
> performing more calculations (very likely), or the new client indeed is less
> efficient. If the latter is true, perhaps the developers are busy fixing the
> problem, explaining their absence from these forums, and if its the former,
> then everyone is in a snit for no reason at all....
>
>

If it is indeed the "former" than why don't they tell us this. There would be a lot less complaining if the dev team would keep us up-to-date of what is going on. If v4.05 is indeed doing more calculations, why not tell us so that we don't ask "Why are my WUs taking so much longer....?".

L8R....

---




Rick A. - BOINCing right along now.... It can only get better!

"There is no fate except that which we create for ourselves."

Live Long and Prosper....

>
> Only feedback from the project is what is needed. It doesn't seem that
> complex to me. Maybe I'm missing something?
>

That's all everyone is asking for.

CyberGoyle,

I hope you took my post as why I do it not to complain about you
I envy you and wish I could have a farm like yours.

I have grumbled in the past but I did not call Berk/Seti team names, I did use the s**t word once but not at S/B. I roll with the punches and yes it is disappointing to loose credits.

I do get irritated when I read the posts and hear the whines rants and threats.
I have defended this project and the teams.

If a user wants to quit I wish them well but JUST GO!
Don’t bother with snide comments, Come back some other time.

Timmy

</img>

---

> OK, lets look at it another way. Say your computer is capable of crunching a
> work unit in 3 hours. In one year, you could crunch 8 WU's per day in 365 days
> or 2920 WU's per year. Now lets say the same work unit takes 6 hours. You
> could now crunch 1460 WU's per year. What's the difference, you say? Well,,
> you could leave your computer powered off for half a year at the 3 hour rate
> and still accomplish the same amount of science. And you would use half the
> electrical energy to do that work.
>
That's the best answer yet. Additionally we could crunch another project for that half year.
The point is Berkeley needs to keep us informed. Lack of communication seems to be the biggest reoccuring complaint.
Was the upgrade precipitated by the errors being experienced by our win98 users? If so then have a seperate Boinc release for XP and Win98, otherwise they are fixing something that isnt broken.
If this latest version takes longer because the WUs are more complex then thats fine. If this latest version takes longer because it analyzes the data in a more complex and detailed way then fine. If this latest version takes longer because of additional debugging and error checking to be put forward into a future release then fine. BUT AT LEAST TELL US ABOUT IT!

Would it be nice if there was a developer on call to answer everyones questions? Sure. Would this be enough to sate everyone's desire for information evertime there is a glitch or change of any kind? I doubt it. Allow me to ask this: exactly why is a steady information flow so critical? And how often should we get news updates? Hourly, daily, or weekly?

Consider what would happen if Berkeley simply folded the Seti project. Nothing. Every Seti user would simply move on with their lives, find other things to do with their 'CPU cycles'. My point is, none of this is as important as some would make it seem in the grander scale of things....



<a>

> OK, lets look at it another way. Say your computer is capable of crunching a
> work unit in 3 hours. In one year, you could crunch 8 WU's per day in 365 days
> or 2920 WU's per year. Now lets say the same work unit takes 6 hours. You
> could now crunch 1460 WU's per year. What's the difference, you say? Well,,
> you could leave your computer powered off for half a year at the 3 hour rate
> and still accomplish the same amount of science. And you would use half the
> electrical energy to do that work.
>
> Now keep in mind, if the new WU's under 4.05 do more intense or different
> processing for a scientific reason, fine. That's a whole different situation.
> We haven't seen anything from the Boinc Set Admin. that says that. If it's a
> mistake or just a strategy to relieve the network load, it's wrong. Just tell
> us to back off until hardware/software problems are solved. Or stop the down
> loadable WU's.
>
> We were told Boinc Seti was ready to transfer to (from Seti Classic). I still
> have Seti Classic on my machines and Will be glad to switch back if that would
> help. Only feedback from the project is what is needed. It doesn't seem that
> complex to me. Maybe I'm missing something?
>
> I guess it's like using a 120 watt light bulb where a 60 watt bulb would
> suffice.


EXACTLY

> Would it be nice if there was a developer on call to answer everyones
> questions? Sure. Would this be enough to sate everyone's desire for
> information evertime there is a glitch or change of any kind? I doubt it.
> Allow me to ask this: exactly why is a steady information flow so critical?
> And how often should we get news updates? Hourly, daily, or weekly?
>
> Consider what would happen if Berkeley simply folded the Seti project.
> Nothing. Every Seti user would simply move on with their lives, find other
> things to do with their 'CPU cycles'. My point is, none of this is as
> important as some would make it seem in the grander scale of things....
>
>
>
> <a>
>


I am with you CyberGoyle, or as I put it "Much ado about nothing", by W. Shakespear hehehehehehe




My name is Pascal and this message has no meaning, but still has my approval...

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

> Allow me to ask this: exactly why is a steady information flow so critical?
So we arent operating blindly. When youre at work and theres a change of procedure wouldnt you ask why that change is necessary? The WHY imparts an understanding that you could never gain simply by knowing the WHAT.

> And how often should we get news updates? Hourly, daily, or weekly?
You'll need to be more specific with what you mean by news. In regards to this topic "Why does the speed of processing a WU matter at all?" it couldve been posted on the main page WHY we were having to upgrade instead of this:
"September 30, 2004 UPDATED: A new version (4.05) of the windows SETI@home client is available (and will be automatically picked up by your core clients next time you contact our servers). Changes include: the "heads up" mode now works and the progress bar advances in a more linear fashion. Earlier today we released a similar version 4.04 which gave users checksum errors and failed to run. This has been fixed in 4.05."
This says nothing about WHY we were upgrading from 4.03, nor so much as WHAT was different.

> Consider what would happen if Berkeley simply folded the Seti project.
> Nothing. Every Seti user would simply move on with their lives, find other
> things to do with their 'CPU cycles'.
Or I could go to the other extreme and ask what would happen if every Seti user simply uninstalled the application. Considering Seti has made international news I would expect a news report calling Boinc a dismal failure.

> My point is, none of this is as
> important as some would make it seem in the grander scale of things....
Unfortunately we live in the here and now and the grander scale of things is beyond our control, maybe even our influence. We all have a common interest/goal, and that is to find out if intelligent life really does exist elsewhere in our galaxy. In sports terms we are all on the same team. And as teammates go we'd like to get some information from the head coach once in a while, especially when the team (or sport) seems headed in the wrong direction. Our complaints dont originate from a need to read our own ascii but come from a desire to see this project succeed. I dont think anyone is asking Berkeley/Boinc/Seti to be perfect. We are only asking them to provide us with info in a timely manner.

The speed per wu is not really that important due to the credit calculation method.

The problem is that if the estimated time for completion is 30% below actual, then it screws up the number of wus in the cache.

I run 10 day caches to avoid down time during outages. I am in danger of over running the 14 day deadline due to there now being more wus in my cache than I have time to process.

Apparently there is no extra work (science) being performed in these wus, just a bunch of extra cycles.

It is being investigated as far as I am aware. I imagine that a fix will be forthcoming :-)

Dunc

Maybe someone out of UC could answere the question, please.



Greetings Basti