Get BOINCing!

[sir_LOIN]

I don't know why many people are acting this way with boinc, it's been great for me so far. Of course they have a few glitches, but with a project that size, who can blame them. Boinc has much more potential than the old seti. They're finally gonna be able to work on different projects at the same time. The research coming out of it will be greather than before.

So you don't get instant credit stats like you used to with the new system, big deal!!! I started seti from the beginning because I love science and I wanted to do a small part, not because I wanna brag about how many credit I have! Seti1 will be dead shortly wether we want to or not, so get pass your ego's and get boincing.

[SunRedRX7]

Well put.

The working on different projects at the same time is going to be huge for distributed computing projects. Imagine the SETI and Folding team at overclockers.com together as one SUPER TEAM under BOINC.

[Steven4563]

everyone might be moaning but ive said from the start when i first started once they finish phase 1 i wont be running seti on any of my computers

[Dk Jedi Allianc]

Well put.

The working on different projects at the same time is going to be huge for distributed computing projects. Imagine the SETI and Folding team at overclockers.com together as one SUPER TEAM under BOINC.

Why not post a thread in folding forum???


Edit: Done that :p

[SunRedRX7]

Well, first we need a folding project for BOINC to be released. Once that comes out then we can have multiple projects all being taken care of by one team.

[EyelessFace]

So you don't get instant credit stats like you used to with the new system, big deal!!! I started seti from the beginning because I love science and I wanted to do a small part, not because I wanna brag about how many credit I have! Seti1 will be dead shortly wether we want to or not, so get pass your ego's and get boincing.

Why does not getting instant credit suck? Take this hypothetical example:

SETI sends you one of those new WUs made specifically for 512 MB of RAM. You take 10 hours to complete it on a p4 3.2C oc'ed to 4.0, and it would yield a large amount of credit for you. However, instead of sending the same WU to people with runtimes and a system similar to yours, it sends it to a 500 MHz P3 (with say 768 MB pc100 ram, meeting the 512 MB requirement) that only runs SETI 2-3 hours a day. So, that WU that took you 10 hours straight would take him 4-5 days. Add the fact that his machine is WAY slower than yours, and your talking WEEKS to complete the same WU. Which basically means that those 10 hours you wasted on ONE WU could have done 4-5 smaller ones that would end up netting you credit very close to what the larger one would net you. And to make it worse, SETI would never know if that triplet or pulse you may have found was actually real, and not just an error of your proc., so the science part ends up getting tossed out the window.

We should get some sort of instant credit... maybe 1/3-1/2 of what we would get as a total for one WU. It may not be as much as full, but if the above example were to ever happen, at least you would get some recognition that your WU was completed. It would not be a total waste of time on your part.

In all truth, I will more and likely end up switching to folding

[sir_LOIN]

Yeah it might take some time, but you will still get the credits eventually... In the long run alot more science will get done. I just don't get why people are so obsessed with credits and stats.

Isn't folding suppose to switch to boinc eventually?

[Fast420A]

There is going to be folding via Boinc but it's called Predidctor@Home run by a different college than Stanford University who runs Folding@Home that this forum folds for.

[TC]

That's interesting - I could have sworn there was an article posted on the main boinc page by Standford about a year ago.

[Fast420A]

I know. Stanford had said they had been talking with Berkley about Boinc but I guess nothing came about with it.

[JigPu]

SETI sends you one of those new WUs made specifically for 512 MB of RAM. You take 10 hours to complete it on a p4 3.2C oc'ed to 4.0, and it would yield a large amount of credit for you. However, instead of sending the same WU to people with runtimes and a system similar to yours, it sends it to a 500 MHz P3 (with say 768 MB pc100 ram, meeting the 512 MB requirement) that only runs SETI 2-3 hours a day. So, that WU that took you 10 hours straight would take him 4-5 days. Add the fact that his machine is WAY slower than yours, and your talking WEEKS to complete the same WU. Which basically means that those 10 hours you wasted on ONE WU could have done 4-5 smaller ones that would end up netting you credit very close to what the larger one would net you. And to make it worse, SETI would never know if that triplet or pulse you may have found was actually real, and not just an error of your proc., so the science part ends up getting tossed out the window.
OK, first off SETI 2 units take almost as long as SETI 1 units. My box can go through a WU in 3.5 hours using SETI, or a WU every 3-4.5 hours using BOINC. This means that P4@4.0 would not take 10 hours to complete. Two hours at the very most. This means the system that it was sent to would probaly end up finishing the same BOINC unit in about 11 hours. Assuming it's on 2 or 3 hours a day, that means a little over 5 days before you ever see your precious credits.

Second off, amount of RAM is not one of the criteria for how to distribute WUs AFAIK. The WUs are distributed based on how the computer scores in benchmarks and your own personal settings (how much disk space to use, etc). The benchmarks test integer, floating point, and memory bandwidth performance are tested. This makes it unlikely for that PIII to ever see your WU.

In fact, it's usually NOT slow computers that delay your credit. In fact, it's the WU distrabution system itself. Since your WU has to be completed by 3 different computers before you recieve credit, that means your WU has be be sent out three different times. Combined with the caches people set (mabey 3 or 4 days), THAT is where the slowdown in credit acceptance occurs.


Delayed credit is no real problem my friend. All it means is a lag between what Berkeley says you've done, and what you've actually done. I was a member of our forum's BOINC beta team (and #3 on it with only one box crunching :attn: ), and credit would show up with nearly every WU I sent. It may not have been credit for the WU I just finished, but the result was the same. I send in WUs, I get credit "instantly". Unless you're a credit pincher (I was for the first week or so), you'll never notice WUs taking an abnormally long time to gain credit. The vast majority will be returned in a few days, and you'll end up pleasently supprised when a WU you sent out appears a week later in your credit totals. Even if you are a credit pincher, you can still see just what results are waiting on credit, letting you know who else has crunched the unit, how long it took him, etc, just so you know that your WU is being worked on.



Finally, your comment about SETI not knowing wether a triplet or pulse was an error or real.... This is absolutely false. The validator compares several results (three or four in most cases) and checks to see if they agree with each other. If they do, each person gets credit, and everything is happy. If the results do NOT agree with each other, the validator sees which result dosen't agree ("which one of these things is not like the other?... which one of these things dosen't belong?"), and then provides credit to the machines who agree on the result, and gives nothing to the ovbiously erroring computer. This is a safeguard against patched clients which just return results without ever doing a computation on them (this was a big problem earlier in SAH1), and against computers who are delivering erronous calculations.


I'm a very happy BOINC user. Been with BOINC since May of last year, and I must say it's a lot better (especially for credit) than naysayers make it out to be.

JigPu

[sir_LOIN]

That was very well said JigPu, pretty much sums it all.

Now all of you, get BOINCing already!

[Horribleron]

Well..I would get to BOINCing if the freakin Berkeley servers would give me something to BOINC with.

BOINCless in Seattle

[sir_LOIN]

BOINCless in Seattle, lol ¦¬)

yeah, seems they've been having their share of problems. Hopefully it'll be up and running soon.

[elec999]

Whats the main different betweem boinc seti and just plain seti.

[eaglescouter]

When does the Boinc move from a beta to a final release?

[SkyHook]

When does the Boinc move from a beta to a final release?

Well if you can believe Berkeley IT IS OFFICIALLY RELEASED! As of last week they went public and almost immediately fell flat on their(anatomy of your choice). They couldn't foresee that suddenly flinging the doors open to a public that had been teased and tantalized for over a year would result in people actually signing up and expecting to do work. And I didn't misspeak when I said "suddenly", as it has been stated that as late as 6 hours prior to their announcement there had been no serious discussion of the current state of things and when they might be ready to start the process of bringing the general public onboard.

SkyHook

Here is one of the announcements taken from the Phase 1 Homepage:

June 22, 2004
A new version of SETI@home based on BOINC, is now available. We'll be transitioning to this new version (details are here). For now, you can use the new version or stick with SETI@home Classic.

[TC]

What puzzles me is this - they already had trouble managing the system loads and bandwidth under phase1, yet they released phase 2 knowing full well that it requires even more resources on their end than phase 1. They don't appear to have taken that into consideration until after they went public. That and the fact SkyHook mentioned about not even having a serious discussion about launching it just a few hours prior to the actual launch. I guess they stopped in at the local watering hole and said "Hey lets go put up a note on the website and see if anybody signs up." Anyway I'm back to crunching phase 1 at the moment. My machines finally ran out of boinc bits.

[sir_LOIN]

I thought I would run out of units but it seems to download a couple here and there, it keeps saying there's no scheduler though. I can't send my finished units so they're just pilling up until they go back online. At least I can still crunch until they sort it out.

[SunRedRX7]

Most of my machines are 8-9hr work machines, so I'm still good in most cases.

I ran out of work units on my 24hr machines though, so I've added
Predictor@home to them http://predictor.scripps.edu/ to keep them busy.
Ratio of 99% SETI, 1% Predictor.

No machine of mine is getting a break!

[Horribleron]

My other concern is how fast is the new GUI BOINC compared to the Linux and MAC OS CLI versons? We all know that the SETI 1 GUI version sucked for speed (I stopped using it long ago) and the x86 CLI is much faster. Are Windows users going to be put at a severe disavantage because the GUI version (BOINC 3.19) is the only "easy" to install version? I have read about "compiling" a "Windows Service" version but I'm not into compiling stuff. I'm not a programer...just a stupid MCSE.

- Horribleron

[sir_LOIN]

From what I've been reading, it seems there's no real performance gain over the gui client. But I have not personally tried it yet so I'm not 100% sure.

[SkyHook]

My other concern is how fast is the new GUI BOINC compared to the Linux and MAC OS CLI versons? We all know that the SETI 1 GUI version sucked for speed (I stopped using it long ago) and the x86 CLI is much faster. Are Windows users going to be put at a severe disavantage because the GUI version (BOINC 3.19) is the only "easy" to install version? I have read about "compiling" a "Windows Service" version but I'm not into compiling stuff. I'm not a programer...just a stupid MCSE.

- Horribleron

All reports I have been able to get my hands on indicate that the BOINC GUI Client is not a system resource hog. I can personally say that I have had the taskmanager open and watching processes while doing various things in the GUI window and have not seen it make any demands on the CPU % in the taskmanager. Some even state that having the pretty graphic displayed doesn't make any demands on the CPU as it is written to do everything through the GPU on the video card, this I can't varify since I personally find the graphic annoying and don't have it on. For my own tastes I do like the information the GUI provides about what is going on and can't see a whole lot of difference between my times with the BOINC GUI and the times I was doing in Classic with the 3.03 CLI.

Hope that helps,
SkyHook

[Dk Jedi Allianc]

Most of my machines are 8-9hr work machines, so I'm still good in most cases.

I ran out of work units on my 24hr machines though, so I've added
Predictor@home to them http://predictor.scripps.edu/ to keep them busy.
Ratio of 99% SETI, 1% Predictor.

No machine of mine is getting a break!

Here is the OC.team: LINK (http://predictor.scripps.edu/team_display?teamid=10)

Looks at bit small - dont you think??

[SunRedRX7]

yeah, now that I joined the predictor@home overclockers.com team its up to 5 members, and I'm not going to really be a huge contributor since I'll probably only be doing WUs for it when SETI goes down. Also only 2-3 of my machines will be running it, as they are the only 24hr/7day machines.

But still, I figure why waste CPU cycles when SETI has its issues.

[zexmarquies01]

Yeah, for the fun of it, till Seti at home boinc is up, i'll do some predictor.

my machine is going to crunch SOMETHING!!!

i'll join, as soon as i get my ID mailed to me.

[Fast420A]

Sorry that you guys are having problems with Boinc. There's always a welcome place on the other DC team of the forums if you continue to have trouble Boincing away.

[TC]

I just went back to regular Seti - it works fine and I wanted to hit some more milestones anyway.

[sir_LOIN]

Well it seems to be back online for the moment, my credits are slowly rising.

[JigPu]

I've been back to SETI 1 for the past few days since neither of my boxes can get a scheduler to reply. I thought about singing up for Predictor, but I previously put up a request for inclusion in the Climateprediciton.net BOINC beta, and am still waiting for a reply. If that falls through, I may still split my time 3 ways between SETI, Predictor, and BOINC Beta, but I'm still not sure.

In light of the lack of WUs from BOINC, this seems rather appropriate :D
Crunch on!! :attn:

JigPu

[sir_LOIN]

It's weird, I can't get units for my main rig but my others are fine....

[JigPu]

Hmm... I had the same problem (I just decided to try BOINC again), and I wound up having to uninstall BOINC and manually remove the directory, then reinstalling before BOINC would ever work. It still didn't connect on it's first try, but it did manage eventually unlike my last install for some reason.

JigPu

[sir_LOIN]

Same here, I reinstalled it and now it's fine... Go figure!