> I see! But do you know why Windows had this advantage? And how the
> developers are correcting it? I remember seeing in another thread, where a
> user complaining about low credits when using Linux, and somebody said it was
> because of the poorly optimized Linux client. So this is false then?
>

Basically the problem is the windows-compiler is so "clever" it saw the benchmark-code didn't really do anything, so just optimized-it away. Not running part of the code at all of course speeds things up, so got higher benchmark-score than should.

Other platforms compilers didn't optimize-away part of the benchmark, so gets lower score.

Doing some small changes to the benchmark-code have lead to the windows-compiler isn't optimizing it away, so now windows-computers at last must run the whole benchmark the scores goes down. These changes haven't changed anything for other platforms, so the end-restults is windows-computers now benchmarks nearer to other platforms.

Steve: "Why does it take that long on your 2500+. I typically do a WU in 2h 40m with the same processor."

Really? Strange... Is it overclocked?

Maybe it's because I have lots of other programs running at the same time in the background, which are stealing CPU cycles. But these are necesarry tasks... The 4.09 CLI is typically using about 87-92% idle time with default priority. I can boost it upto tops 97% if I change the priority to "AboveNormal".

mikey: "That is because with ver 4.25 and above the Windows benchmark is coming down towards where it really should be as far as credits go. Windows for a long time has had an advantage over other platforms and the developers are bringing the numbers more in line with each other."

I see! But do you know why Windows had this advantage? And how the developers are correcting it? I remember seeing in another thread, where a user complaining about low credits when using Linux, and somebody said it was because of the poorly optimized Linux client. So this is false then?

---

> sterling0466, it looks as if your stats are relatively a bit low. With your
> fx 51 processor, you have greater crunching power than me. I have only one
> processor; an AMD 2500+ (Barton type, not overclocked) 1 GB DDR333 RAM, and
> I'm getting about 210 credits per day (24 hours). A WU typically finishes
> after 3,5 hrs here.
>
> But I'm still running BOINC 4.19. I won't upgrade until 4.26 at least...
>
That is because with ver 4.25 and above the Windows benchmark is coming down towards where it really should be as far as credits go. Windows for a long time has had an advantage over other platforms and the developers are bringing the numbers more in line with each other. Personally I would like to see a mandatory upgrade when they do this so people like you and I, that are still running older versions, don't have an advantage over users with the newer versions. Either that or leave it out until they release a version that is mandatory.

---

> sterling0466, it looks as if your stats are relatively a bit low. With your
> fx 51 processor, you have greater crunching power than me. I have only one
> processor; an AMD 2500+ (Barton type, not overclocked) 1 GB DDR333 RAM, and
> I'm getting about 210 credits per day (24 hours). A WU typically finishes
> after 3,5 hrs here.
>
> But I'm still running BOINC 4.19. I won't upgrade until 4.26 at least...
>
>
Why does it take that long on your 2500+. I typically do a WU in 2h 40m with the same processor.

---

98SE XP2500+ @ 2.1 GHz Boinc v5.8.8

And God said"Let there be light."But then the program crashed because he was trying to access the 'light' property of a NULL universe pointer.

sterling0466, it looks as if your stats are relatively a bit low. With your fx 51 processor, you have greater crunching power than me. I have only one processor; an AMD 2500+ (Barton type, not overclocked) 1 GB DDR333 RAM, and I'm getting about 210 credits per day (24 hours). A WU typically finishes after 3,5 hrs here.

But I'm still running BOINC 4.19. I won't upgrade until 4.26 at least...

---

Well, it has been a couple of days now, and things are running, but differently. Boinc has given me a third id number for my work computer, and my credits have been cut in half of what I normially did. I used to have on average a daily credit of 300, give or take, now I am sitting at about 179 daily average. I have merged all of my three work computers together, and am hopeing that things will stabilize soon. My home computer is using the AMD Athlon 64 fx 51 processor with 1 Gig 333 DDR memory, and completes most work units in about 2 hours flat. My work unit is a Intel Celeron 1.4 with 640 Megs of Pc133 memory, completing most units in about 7 hours. Does this sound right to anyone else??? Plus, now the classic information has been added to the Boinc databases, are they going to assign a given credit amount for each of the classic hours and then add it to our Boinc credit? I thought that I read that somewhere not to long ago?

---

I'm running three projects here under Boinc 4.19 and having no problems at all in uploading/downloading WUs.....

Two machines, both crunching 24/7 -
Machine 1 is connected to Boinc all the time.
Machine 2 is connected to Boinc once a day. (by manually switching USB cable from Broadband modem; Machine 1 to Machine 2 ! )

---

> > % ./sent_result_check
> > Sent: 3 1 2 1 1 6 1 0 1 2 - 18 total
> >
> > So one out of the last ten people got nothing. Here.. I'll run it again:
> >
> > % ./sent_result_check
> > Sent: 3 1 1 1 6 1 1 5 1 1 - 21 total
> >
>
> Just curious: this seems to imply that most of the time, BOINC is requesting
> one WU at a time. Is that because the default cache size is small?
>
Most likely this is because of a change in the way the uptime is figured in the 4.25 client. In earlier versions the active_frac was the total percentage that BOINC was able to crunch and was used directly to determine how much work to send. In the new versions active_frac is the percentage of on_frac that the BOINC is able to crunch then the 2 numbers are multiplied to determine how much work to send.

Take for example numbers from a computer that is on half the time and allowed to crunch for half the time that it is on.
Old: on_frac = 0.5 active_frac = 0.25 multiplied = 0.125
New: on_frac = 0.5 active_frac = 0.5 multiplied = 0.25
From this example you can see that after the update the number reported to the server will be half of what it should be. It will return to more normal levels with time.

---

BOINC WIKI

BOINCing since 2002/12/8

> % ./sent_result_check
> Sent: 3 1 2 1 1 6 1 0 1 2 - 18 total
>
> So one out of the last ten people got nothing. Here.. I'll run it again:
>
> % ./sent_result_check
> Sent: 3 1 1 1 6 1 1 5 1 1 - 21 total
>

Just curious: this seems to imply that most of the time, BOINC is requesting one WU at a time. Is that because the default cache size is small?

---

I have now switched both of my work and home computers to Ver. 4.25, and I must confess that that I have been able to get a few download files. I have also upped my cashe to 5 days at home and 2.5 days at work since I was only getting one or two packets per download. Since my computer at home completes about 5 or 6 packets a night, and the one at work does about 3 or 4 a day, I should be alright now. Still have the question though, why do I have to manually connect each time, several times, to get any of the available downloads?

---

Matt,

> Well.. for the record when the status page says "0 results to send" that doesn't mean we're out of work to send and we should all give up and go home. That means the splitters aren't getting ahead of demand, but most everybody is getting hot, fresh work as soon as it is generated at this point in time.

This is an example of why I think the Server status page should be a little more explanatory. As I said in the thread "What would you like to see on Status page?":

"But I'd like a better explanation of the data already present there."

So, if you please can be a little more verbose in explaining the numbers and terms? It would perhaps also be fewer questions in these forums then. :)

---

> Check it out. I have a little script that checks the last ten requests for
> work and how many workunits were handed out to these clients:
>
> % ./sent_result_check
> Sent: 3 1 2 1 1 6 1 0 1 2 - 18 total
>
> So one out of the last ten people got nothing. Here.. I'll run it again:
>
> % ./sent_result_check
> Sent: 3 1 1 1 6 1 1 5 1 1 - 21 total
>

It might be cool to put something like this on the status page....

---

> I may have had 20 download attempts, but only three of them actually worked,
> the others all stated download failure, and did nothing for credit! I really
> do enjoy working on projects for SETI, but the continued and mounting problems
> and excusses are driving people away. I know of five of my friends who have
> already quit using SETI because of Boinc, and if things don't get better soon,
> I may be #6

So, does this mean that SETI can't ever have a download failure?

... because if the downloads succeeded later, why does it matter?

---

> > Thanks for the heads-up, Matt. We appreciate it greatly, though I must
> admit
> > I'm one of those that upped my cache to 10 days just so I can keep my
> computer
> > crunching. Having mistakenly detaching my computer from this
> project(and
> > losing 13 perfectly crunched WUs in the process), I'll leave this idling
> if
> > SETI runs out of work. My laptop's handling all of this just fine.
> >
> > Okay, sterling?(My brother's name!) BTW, I forget which one of your
> computers
> > still hasn't upgraded to 4.19 but just a suggestion to check it. Glad to
> see
> > you're still willing to crunch.
> >
> >
> Alright, that does explain things a bit more. I have just finished updating
> my work computer to the new Ver. 4.25, and I lost the ten or tweleve work
> units that I had, so I guess I have to wait a few minutes until the BOINC
> servers produce more work. Hopefully this will solve the problems, and I will
> switch my home computer tonight to the new version as well. Thanks for all
> the advice.......
>

Home computer now switched over to Ver 4.25, lets see if I can get a few work units to test out the new software??? Hopefully this will solve my problems.

---

> Thanks for the heads-up, Matt. We appreciate it greatly, though I must admit
> I'm one of those that upped my cache to 10 days just so I can keep my computer
> crunching. Having mistakenly detaching my computer from this project(and
> losing 13 perfectly crunched WUs in the process), I'll leave this idling if
> SETI runs out of work. My laptop's handling all of this just fine.
>
> Okay, sterling?(My brother's name!) BTW, I forget which one of your computers
> still hasn't upgraded to 4.19 but just a suggestion to check it. Glad to see
> you're still willing to crunch.
>
>
Alright, that does explain things a bit more. I have just finished updating my work computer to the new Ver. 4.25, and I lost the ten or tweleve work units that I had, so I guess I have to wait a few minutes until the BOINC servers produce more work. Hopefully this will solve the problems, and I will switch my home computer tonight to the new version as well. Thanks for all the advice.......

---

Thanks for the heads-up, Matt. We appreciate it greatly, though I must admit I'm one of those that upped my cache to 10 days just so I can keep my computer crunching. Having mistakenly detaching my computer from this project(and losing 13 perfectly crunched WUs in the process), I'll leave this idling if SETI runs out of work. My laptop's handling all of this just fine.

Okay, sterling?(My brother's name!) BTW, I forget which one of your computers still hasn't upgraded to 4.19 but just a suggestion to check it. Glad to see you're still willing to crunch.

---

Well.. for the record when the status page says "0 results to send" that doesn't mean we're out of work to send and we should all give up and go home. That means the splitters aren't getting ahead of demand, but most everybody is getting hot, fresh work as soon as it is generated at this point in time. Maybe not as much as they'd like, but something.

Check it out. I have a little script that checks the last ten requests for work and how many workunits were handed out to these clients:

% ./sent_result_check
Sent: 3 1 2 1 1 6 1 0 1 2 - 18 total

So one out of the last ten people got nothing. Here.. I'll run it again:

% ./sent_result_check
Sent: 3 1 1 1 6 1 1 5 1 1 - 21 total

..anyway, as other users pointed out, this in not a problem with BOINC. In fact, we fully expect in the future (once we are caught up and have all servers in line) that it will be quite normal for work to be unavailable since we have a surplus of computing power - which is why we highly recommend users join other projects to keep their CPUs busy (and their credits increasing).

- Matt

---

-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude

> I am using ver. 4.19, but I just checked the server's status
> page and it says that there are 0 packets ready for download at the time of
> this posting...we'll see later tonight.......

That's what Boinc is for:
Seti has no work? No problem, just crunch some other project.

The Classic version was this: We don't have new work? Send them some old crap, just to please them. It's been crunched 30x already? So what!

I prefer the Boinc solution!

---

Gruesse vom Saenger

For questions about Boinc look in the BOINC-Wiki

> > I havn't counted the DL-failures, just the ~20, that have either reported
> back
> > yet or are still green. ~1/3 has reported so far, one even granted
> credit. And
> > just for 2 days, 08 and 09.
>
> I agree with Saenger,(Same guy from Boincstats? If so, `sup!) You tend to
> have a moderate amount of d/l errors but for the most part, you got at least
> 12 WUs successfully d/l onto your second computer and work on the first one
> seems to be going fine. You did a lot of work for SETI and I hope, as a
> fellow cruncher, you keep at it. It's minor problems that SETI has.(What's
> your client at? If it's 4.13, think about upgrading to 4.19 or even 4.25,
> that might help mitigate your d/l errors.)
>
>
>
You are correct, my work computer did recieve several downloads today, I just checked, hopefully my home computer which is three times faster will recieve some tonight. I am using ver. 4.19, but I just checked the server's status page and it says that there are 0 packets ready for download at the time of this posting...we'll see later tonight.......

---

> I havn't counted the DL-failures, just the ~20, that have either reported back
> yet or are still green. ~1/3 has reported so far, one even granted credit. And
> just for 2 days, 08 and 09.

I agree with Saenger,(Same guy from Boincstats? If so, `sup!) You tend to have a moderate amount of d/l errors but for the most part, you got at least 12 WUs successfully d/l onto your second computer and work on the first one seems to be going fine. You did a lot of work for SETI and I hope, as a fellow cruncher, you keep at it. It's minor problems that SETI has.(What's your client at? If it's 4.13, think about upgrading to 4.19 or even 4.25, that might help mitigate your d/l errors.)

---