OK, well here's my chart from BOINC manager:



As you can see the visible drop, in my case, is fully explained by time spent running Ralph (note that the dates for the two projects aren't aligned directly over each other).

I show this not to dispute Stevea. But just to contribute more data. You can see my hosts and CPU types and etc. as my machines are not hidden. Perhaps they are similar machines to Steve's? Perhaps his are different and that's the reason we're seeing different results? I've already speculated on possible explainations that cover some of what my machines have been through in the past month or so. But you really don't seem to be able to see a trend in my chart from those events.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

The only solution for stevea is to go to WCG where they still use the quorum system liek Predictor did. Though since few people use optimised clients there it will probably be found that PPD is decreased compared to predictor.

Other projects to look at are docking@home, developed by some of the predictor people, or proteins@home. Both use a similar fixed credit system, maybe it will be more stable what you see here.
Or go folding@home and be done with credits of boinc.

I would guess until BOINC try to setup the 'project benchmarking' and we all start running a standard benchmark task (for different task types) then it will never be stable with the current system.
But it is unstable for all the people running Rosetta@home.

Since stevea and similar one project people are only competing against other rosetta@home people then it doesn't really matter if the credits are dropping by 10% due to a shift in computers/clients. Since it is effecting everyone else just running rosetta@home just the same.

---

Team mauisun.org

No I don't think that is the only solution, nor the best. He was making a point about credits granted by Rosetta. I believe my chart illustraits that not everyone is seeing this impact. Which leads to one of two conclusions:

1) Either Steve's machines are changed in some way which is causing them to crunch less Rosetta models,

or,

2) some users are impacted, and others are not. Whichs tends to rule out variations between WUs as the cause.

And we've already discussed a number of ways that #1 might come about with other Windows (or other platform) tasks slipping in and pulling CPU time, WU failures or watchdog ends, which don't award credit in the normal mannar and etc. But Steve feels he has ruled ALL of these other issues out. And seems to feel that the credit decline is project-wide, and his point about how his world ranking not slipping with his RAC seems rather sound.

...on the other hand, not everyone is effected. I don't make a dent in any world rankings, and have been intentionally running screensaver on one machine to help study the problem. This caused many failed WUs and hangs and loss of credits for me, but I've still got two other machines crunching away

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

No I don't think that is the only solution, nor the best. He was making a point about credits granted by Rosetta. I believe my chart illustraits that not everyone is seeing this impact. Which leads to one of two conclusions:

1) Either Steve's machines are changed in some way which is causing them to crunch less Rosetta models,

or,

2) some users are impacted, and others are not. Whichs tends to rule out variations between WUs as the cause.

And we've already discussed a number of ways that #1 might come about with other Windows (or other platform) tasks slipping in and pulling CPU time, WU failures or watchdog ends, which don't award credit in the normal mannar and etc. But Steve feels he has ruled ALL of these other issues out. And seems to feel that the credit decline is project-wide, and his point about how his world ranking not slipping with his RAC seems rather sound.

...on the other hand, not everyone is effected. I don't make a dent in any world rankings, and have been intentionally running screensaver on one machine to help study the problem. This caused many failed WUs and hangs and loss of credits for me, but I've still got two other machines crunching away

Well thanks for pointing out the fact about the word ranking. I just thought that it would be logical that if it were only me then my world ranking would be one of the side effects of it being just me. But it has remained constant thru the credit drop?

But looking at your chart, I don't see the same drop? This is one of the things that has me baffled. The two just don't make any sense.

I'm going to go bang my head against the wall again, and see if I can come up with a logical way to blame this all on who?

---

BETA = Bahhh

Way too many errors, killing both the credit & RAC.

And I still think the (New and Improved) credit system is not ready for prime time...

I'm not sure that a fluctuation of less than 10% is really something to be worried about... ?

---

Looking at the charts provided by Stevea and Feet1st and comparing them to my own numbers, a few questions come to mind.

First, what I see:

It looks like Stevea's RAC was on par with what he was getting in late November until about December 12. We might be able to attribute anything before that to normal fluctuations. But then after the 12th, the bottom fell out. It has not recovered.

Feet1st's RAC took a similar nosedive around the same time, but it recovered beginning around December 20.

My overall RAC means nothing, because of other influences, so I looked at my steadiest cruncher, which happens to be a Mac (handy for purposes of comparison). I saw a similar nosedive around December 13 or 14 but it began to recover around December 20, the same as with Feet1st.

Now, the questions that this makes me ask (mostly obvious ones):

What happened around December 12-14? This was just after the results were released from CASP7. I don't remember what the specific changes were (maybe someone can help) but I do recall that the mix of WUs changed around the same time. Did something else happen too?

What happened around December 20? It seems likely that this marked another shift in the WUs that were being sent out. Was this the end of a series that was introduced around the 12th? Did something else happen too?

Why did Feet1st's RAC rebound at the same time that mine did, but Stevea's did not? It seems platform independent since they are using the same platform and I am using something different. Could it be hardware specific somehow? Since Stevea is running four of the same setup, we can't rule this out.

It might be helpful if others could post the charts from their individual machines, so we can have a look. Please also tell us what type of machine it is that we are looking at.

I tend to agree that the RAC has been dropping this month.
I have two machines that run Rosetta and nothing else.

The first is a Inel based laptop with a 1.70Ghz CPU. The first few days of
Dec it had a RAC of 110, the RAC has declined slowly everyday and is at 99 today.
It has not show any recovery around the 20th as some have noticed on their machines. I haven't made a single change to this machine.

The other machine that only runs Rosetta is a AMD 3400+based CPU BUT I didn't
start it until the 12 th of Dec. It is showing an increase everyday but I think this is normal at this point.

The other machines run several other projects in addition to Rosetta and they do
show the RAC decreasing around the first of Dec and increasing around the 20 of Dec. The other projects are SETI and WGC on both Windows and Linux platforms.

I'm not sure what all the results on my machines are telling me. I haven't concluded there is a common factor at his point. Anyway thats my 2 cents worth.

---

Puzzling--we will look into this next week. do you think this could be related to the overall drop in credits per day earlier this month?

---

These are the dates of new rosetta apps since the introduction of docking.
(editted to add descriptions)

rosetta_5.32 10-Oct-2006
Initial docking release

rosetta_5.34 24-Oct-2006
Bug fixes, Lower energy decoys from same amount of computation, Allow more of the protein's geometry (more specifically, the bond angles) to vary during the run

rosetta_5.36 31-Oct-2006
Fixes to varying bond energy (too slow) and to fix uneven creditting this caused (accelerate the functhion), Improved accuracy to energy functions, New compressed output format for designed proteins (HIV etc). Graphics fixes

rosetta_5.40 13-Nov-2006
Graphics fixes, Bug Fixes, Added flexibility feature to test new shapes, Added new mode (for Alzheimer's), Added feature related to non ideal backbone and sidchain

rosetta_5.41 30-Nov-2006
Bug fixes

rosetta_5.43 13-Dec-2006
Lockdown of Graphics, Added fix for croyelec. microsocopy

---

Team mauisun.org

Humm i just looked at my RAC also i used to get mid 40s for my granted. Now my machine still says mid 40s for requested but the granted is between 28 to 35 tops that tends to be more than 10 % drop for the last 3 weeks. So I think Stevea truey is onto something. My machine runs 24/7 tho there was 5 days near the begining of the month i had it off. And i do game during the day withthis machine but still I have not t changed anything i been doing for over 6 months.

---

Who?:

...
I guess, the thousands of Core 2 Computers can be responsable for other lower performance RAC drops ... if i understood well.

who?

Interesting, actually - your other box (the overclocked quad) doesn't show such an extreme effect. The claimed credit (based on benchmark) is around 100, and the granted credit is around 160 (versus 70 and 130 on the octocore). Granted credit per hour per core is around 26-27 which is pretty impressive. As an aside your octocore should reach 4100 RAC after a month or so.

I do note that the Boinc benchmark is quite often erratic, so you might find your octocore box gives you a better benchmark the next time it runs (and hence higher claimed credit).

Other core2 systems also seem to show less of a divergance between claimed and granted, but still tend to show
https://boinc.bakerlab.org/rosetta/results.php?hostid=317079&offset=60 - 120 claimed, 150 granted
https://boinc.bakerlab.org/rosetta/results.php?hostid=359325 - 54 claimed, 65-70 granted
https://boinc.bakerlab.org/rosetta/results.php?hostid=282631 - 55-59 claimed, 65-72 granted
https://boinc.bakerlab.org/rosetta/results.php?hostid=346572&offset=20 - 57-59 claimed, 70-80 granted


You're quite right to say that if the % of these machines in the pool running Rosetta becomes significant, it'll affect the overall balance between granted and claimed. (The same is of course also true for Linux, Mac, ...).

---

I have a combination of old and new computers running Rosetta full time. I have noticed that all computers have had a steady decline that began in early December. The 1.8ghz Duo Core was running about 400 credits, and now it runs about 338 credits. The Tyan dual Athlon MP 2000+ was running near the high 300's and now it is running in the low 200's. I have two quad Poweredge (700 & 900mhz) that have been least affected of the 10 or so computers I run for Rosetta.

Poweredge 6450, Xeon 4 x 700mhz, 4 gig PC133 Reg, 10K 18 gig hd perc 3 raid 5.
Poweredge 6450, Xeon 4 x 900mhz, 2 gig PC133 Reg, 10K 18 gig hd SCSI.
Tyan S2462, 2 x Athlon MP 2000+, 2 gig PC2100 Reg, 320 gig IDE. (7 cooling fans)
Tyan S2510, 2 x PIII 1 ghz, PC133 1 gig PC133 Reg, 10K 9 gig hd perc 3 raid 0.
HP Kayak, 2 x PIII 800mhz, PC800 512meg, 20 gig IDE.
Asus K7XX, Sempron 3000+, PC2700 1.2 gig, 250 gig IDE.
Dell 2400, Pentium 4, 2.53 ghz, PC2700 1.5 gig. 60 gig IDE.
Dell D810, Pentium M, 1.8 ghz, 2 gig ram, 60 gig IDE.
Dell D820, T2500 Core Duo, 1.8 ghz, 2 gig ram, 60 gig sata.
Abit K7AN, Athlon 2000+ , 1.88 ghz, PC2700 512meg, 60 gig IDE.

I thought the VNC network I set up was the problem until I read this thread.

---

Well I have not hit bottom yet, my 4 machines are still dropping. As of the new year they are down to 1215 from 1330.

Just so the new people don't have to scroll down to the bottom.

1 = XP2600 M @2705mhz
2 = XP2600 M @2640mhz
3 = XP2600 M @2596mhz
4 = XP2600 M @2565mhz

All Water-Cooled running Windows XP, crunching 24/7.

Seems to have slowed down a bit, but I have dropped a few places in the World RAC ratings up to 312 from 303.

But my teammates x2 boxes have started to climb, or at least flatten out, so there may be an end in sight?

Seems some people were hit harder than me by this. Still would like to know what caused this.

---

BETA = Bahhh

Way too many errors, killing both the credit & RAC.

And I still think the (New and Improved) credit system is not ready for prime time...

I'm just searching for an answer... Does this coincide with the release of 5.43?

Yes it does, as my RAC went down after 5.43 was released. I even let them know in the 5.43 problem thread. I figure that the work units that are being sent out with this version of Rosetta are not going to be very high credit ones.

Yes Matt Davis, you are an Ax, but which ax you are I'm not sure, a Battle Ax or a Woodsman Ax...;-)

---

Well I have not hit bottom yet, my 4 machines are still dropping. As of the new year they are down to 1215 from 1330.

Just so the new people don't have to scroll down to the bottom.

1 = XP2600 M @2705mhz
2 = XP2600 M @2640mhz
3 = XP2600 M @2596mhz
4 = XP2600 M @2565mhz

All Water-Cooled running Windows XP, crunching 24/7.

Seems to have slowed down a bit, but I have dropped a few places in the World RAC ratings up to 312 from 303.

But my teammates x2 boxes have started to climb, or at least flatten out, so there may be an end in sight?

Seems some people were hit harder than me by this. Still would like to know what caused this.

>> Stevea, unsure if this is the same problem you are having as I have not been recoding RAC due to my changing project bias but I did do a check back in November and recorded the result average over 5 results for each computer, this I have done again to see if I am also dropping.
5 Result Ave 5 Result Ave
Computer 8/11/06 2/1/07

No.1 Celeron 2.4GHz 7.61 7.02 -7.75%
No.2 Celeron 2.4GHz 7.08 6.98 -1.41%
No.3 Celeron 2.4GHz 7.43 6.56 -11.71%
P4 2.53 @2.75 GHz 9.04 8.58 -5.09%
AMD X2 4800+ 2.4GHz 14.43 14.13 -2.08%
Opteron 275 2.2 GHz 14.81 15.00 +1.28%
Opteron 285 2.6 GHz 15.43 16.09 +4.28%

So it appears that on this small sample my slower machines have dropped but my faster machines have (4800,275,285), stayed the same or increased.
A larger sample would be needed to double check this, perhaps last 10 results per machine (if I get the time).

---

Just a stab in the dark....

There is another thread here about how BOINC quits for no apparent reason over the bast month or so, leaving BOINCmgr running connected to nothing. This has been happening to me a LOT. I am sure it's dragging my RAC down, as I check on my machines only a couple times a day...so there is up to 12 hours that the machines could be sitting idle until I notice and then restart BOINC.

Perhaps this could be leading to the drop in RAC for others as well.

---

Reno, NV
Team: SETI.USA

Hello,

My RAC also dropped in all the machines since the last month.
I don't know if the reason are this new work units or the version of rosetta, but the true is that the RAC dropped.

---

Checking my machine, I see that I've got several days of the client not producing anything starting on Dec 14. And it last uploaded data on dec 30, and hasn't turned anything in since. So it seems to be that I'm getting the communication disconnect problems that Feet1st was talking about. I've got my system setup with a 24hr setting; and am not sure what's causing the system to regain communication - since the system was last rebooted on the 24th. i.e. if it restarts communication when it decides the system is overcommitted - or something else, because I haven't been checking it regularly.

---

The stdoutdae.txt log file will probably contain a hint. If Boinc lost contact with the server for more than 6 hours or so it could have gone to 'communications deferred' for a week in the worst case.

---

I have no communication problems, or stuck machines. My last wu that errored out was on Dec 19th, and thats on all 4 machines.

My RAC has seemed to at least be remaining around 1215-1225 for the last couple of days. So I don't know if will start to rise again or remain at this point, only time will tell. And I am continuing to rise in the world RAC from 303 to 318th now.

But there is something that's happened or happening thats caused it to drop from its 8 month average of over 1325.

If it's found that the drop is caused by the recent addition of Conroe's, or some other CPU's, then they are going to have to adjust the credit system. Because my machines are doing the same amount of work that they have done for the last 8 months. They are just receiving less credit for the same amount of work.

Just seems kinda weird that it all happened to some people around the same time, and others seem unaffected, and some that were affected have now recovered, while others have not and continue to receive low credit.

We will have to see what the project managers have to say, if its going to remain the way it is now, I'll be folding soon.

---

BETA = Bahhh

Way too many errors, killing both the credit & RAC.

And I still think the (New and Improved) credit system is not ready for prime time...