I am OCing a dual core Intel 3ghz cpu and I see in my results that I get between 12-20 credit more than claimed yet my RAC graph is more or less flat lined.
I should be regaining what I lost when I crashed a few tasks, but I don't see that happening. Why is this happening?

I am OCing a dual core Intel 3ghz cpu and I see in my results that I get between 12-20 credit more than claimed yet my RAC graph is more or less flat lined.
I should be regaining what I lost when I crashed a few tasks, but I don't see that happening. Why is this happening?

have a look on the boinc stats graph. Either one will take a while to change significantly though.

---

I still don't get this, I am running with only two tasks out of twenty came back with a granted total of less than the claimed total. Of the other 18 there are six still waiting to run and all the rest have a granted total of 4-10 points OVER the claimed total, yet my RAC graph shows a curve going downward. This is not logical. It should still be climbing. There have been no errors with the tasks, so why am I taking a dive? There is no pending credit either....

You're leaving the time frame out of your description. The time smoothing of the RAC may hide your recent improvement. I know from my own experience that I could suddenly set Rosetta to 100% resource share and it would be days before I notice the RAC and charts showing any effect. So rising and falling depends on how many tasks you complete, the credit granted, and over what time frame it is granted. Of course you know that, but maybe you are expecting too much too soon?

http://boincstats.com/stats/user_graph.php?pr=rosetta&id=85645

---

time frame was about 24hrs, dual core cpu run time 4hrs. just taking the one page that has the most recent completed items and looking at the score and seeing that i have a downwards curve instead of a upwards diagonal line. as stated all the credit run above the claimed credit other than a few. been running 100% cpu for over 2 months now.

You're leaving the time frame out of your description. The time smoothing of the RAC may hide your recent improvement. I know from my own experience that I could suddenly set Rosetta to 100% resource share and it would be days before I notice the RAC and charts showing any effect. So rising and falling depends on how many tasks you complete, the credit granted, and over what time frame it is granted. Of course you know that, but maybe you are expecting too much too soon?

interesting, of course my graph is shrinked to fit the window versus theirs. but it looks good.

http://boincstats.com/stats/user_graph.php?pr=rosetta&id=85645

24 hours is not long enough to see major changes on the RAC. Give it more time.

---

there is something seriously wrong with Rosies computation of RAC
here is 17 tasks with granted credit being higher than claimed.

179677986 164059192 23 Jul 2008 12:39:49 UTC 28 Jul 2008 15:54:56 UTC Over Success Done 14,264.70 94.00 111.10
179676312 164056919 23 Jul 2008 12:44:00 UTC 28 Jul 2008 15:54:56 UTC Over Success Done 14,406.73 94.94 111.74
179654801 164036811 23 Jul 2008 10:35:26 UTC 28 Jul 2008 8:38:09 UTC Over Success Done 13,875.16 91.44 108.31
179653820 164035588 23 Jul 2008 10:39:37 UTC 28 Jul 2008 11:59:24 UTC Over Success Done 14,175.08 93.41 114.09
179628736 164013608 23 Jul 2008 8:12:06 UTC 28 Jul 2008 5:18:30 UTC Over Success Done 13,938.95 91.86 101.76
179601689 163987521 23 Jul 2008 5:23:27 UTC 27 Jul 2008 13:59:13 UTC Over Success Done 13,945.28 91.90 129.71
179601686 163987518 23 Jul 2008 5:23:27 UTC 27 Jul 2008 21:12:09 UTC Over Success Done 13,951.95 91.94 105.55
179601680 163987512 23 Jul 2008 5:23:27 UTC 27 Jul 2008 17:34:52 UTC Over Success Done 14,025.16 92.42 129.04
179601100 163986385 23 Jul 2008 5:34:40 UTC 28 Jul 2008 1:17:58 UTC Over Success Done 14,053.64 92.61 111.47
179600828 163986037 23 Jul 2008 5:38:50 UTC 28 Jul 2008 1:17:58 UTC Over Success Done 14,188.73 93.50 116.20
179597242 163984216 23 Jul 2008 5:23:27 UTC 27 Jul 2008 13:59:13 UTC Over Success Done 13,853.22 91.29 110.06
179597241 163984215 23 Jul 2008 5:23:27 UTC 27 Jul 2008 17:34:52 UTC Over Success Done 14,282.30 94.12 114.49
179597210 163984187 23 Jul 2008 5:23:27 UTC 27 Jul 2008 21:12:09 UTC Over Success Done 14,384.17 94.79 100.02
179592479 163977465 23 Jul 2008 8:07:55 UTC 28 Jul 2008 5:18:30 UTC Over Success Done 14,194.59 93.54 125.05
179509771 163899132 22 Jul 2008 19:58:59 UTC 27 Jul 2008 11:18:19 UTC Over Success Done 14,005.22 92.29 125.98
179506576 163898078 22 Jul 2008 19:49:47 UTC 27 Jul 2008 11:18:19 UTC Over Success Done 14,238.03 93.83 105.47

now explain to me why in the past 48 hours that with these kinds of totals, my RAC is dropping like a stone? From a high of about 1400 on the 26th to today the 28th I have lost approximately 20 RAC points to 1380. But the totals shown above do not support that loss at all. This is why I say something is wrong.
My system runs 24/7 and there have been no errors in the past 48 hours or longer to make this kind of drop. Going back farther in my history I find a task that did not make a run for some reason that reported no reply today. I don't think this has anything to do with the loss as there was never any credit asked for or received. Anyone got any ideas?

A 20 point dip in RAC is hardly "dropping like a stone".

RAC is meant to provide a smooth "average" over a long period, measured in weeks, not hours of days.

It was shown a long time ago that sporadic and grouped reporting of results has a negative effect on RAC. Those that are rabid about their RAC should attempt to report every result as it is finished, and not wait for a number to be reported at once. If I remember right (and I'm not looking it up for you in the wiki), one of the factors in the RAC computation is "time since last report". Reporting in groups makes that almost 0:00 for all but one of the results reported. The "report results immediately" feature in some of the 3rd party clients eliminates that issue.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

that's less than 2% variance, and as Angus says, you have to take the sporadic reporting of results into account - no matter how you calculate it it will always fluctuate.

---

ok guys, thanks

RAC is based on average credit over a 7 day period, plus some projects have tasks larger then that and so they use an expoential formula to graduate the peak and valley of a single task being reported. Note the 604800 seen in the formula is the number of seconds in a week. So it takes a full 14 days for RAC to be fully reflected.

In your question, you provided about 1 day of data. Not enough information to compute your RAC.

---

Rosetta Moderator: Mod.Sense

RAC is based on average credit over a 7 day period, plus some projects have tasks larger then that and so they use an expoential formula to graduate the peak and valley of a single task being reported. Note the 604800 seen in the formula is the number of seconds in a week. So it takes a full 14 days for RAC to be fully reflected.

In your question, you provided about 1 day of data. Not enough information to compute your RAC.

I did not want to flood the text box with data, but you can look at my profile to see ALL the tasks that were processed and their totals. Typically I get something like 7-10 points more than claimed. So on that basis I expect my RAC to climb at a steady rate and not this zig zag pattern I have seen lately. The RAH RAC differs in its pattern than the http://boincstats.com/stats/user_graph.php?pr=rosetta&id=85645 graph which does show a steady climb with barely descernable valleys, which is what I figured RAH to be doing. What gets my attention is that for 2 days in a row the RAC of RAH has dropped (according to the local machine graph. their (BS) graph shows for the past 2 days a steady increase in RAC vs a drop according to RAH.

So I would gather from this difference that two different formulas are being used by each group to compute RAC vs one agreed upon standard.

RAC is based on average credit over a 7 day period, plus some projects have tasks larger then that and so they use an expoential formula to graduate the peak and valley of a single task being reported. Note the 604800 seen in the formula is the number of seconds in a week. So it takes a full 14 days for RAC to be fully reflected.

In your question, you provided about 1 day of data. Not enough information to compute your RAC.

I did not want to flood the text box with data, but you can look at my profile to see ALL the tasks that were processed and their totals. Typically I get something like 7-10 points more than claimed. So on that basis I expect my RAC to climb at a steady rate and not this zig zag pattern I have seen lately. The RAH RAC differs in its pattern than the http://boincstats.com/stats/user_graph.php?pr=rosetta&id=85645 graph which does show a steady climb with barely descernable valleys, which is what I figured RAH to be doing. What gets my attention is that for 2 days in a row the RAC of RAH has dropped (according to the local machine graph. their (BS) graph shows for the past 2 days a steady increase in RAC vs a drop according to RAH.

So I would gather from this difference that two different formulas are being used by each group to compute RAC vs one agreed upon standard.

Boincstats does not use the BOINC RAC formula.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

RAC is based on average credit over a 7 day period, plus some projects have tasks larger then that and so they use an expoential formula to graduate the peak and valley of a single task being reported. Note the 604800 seen in the formula is the number of seconds in a week. So it takes a full 14 days for RAC to be fully reflected.

In your question, you provided about 1 day of data. Not enough information to compute your RAC.

I did not want to flood the text box with data, but you can look at my profile to see ALL the tasks that were processed and their totals. Typically I get something like 7-10 points more than claimed. So on that basis I expect my RAC to climb at a steady rate and not this zig zag pattern I have seen lately. The RAH RAC differs in its pattern than the http://boincstats.com/stats/user_graph.php?pr=rosetta&id=85645 graph which does show a steady climb with barely descernable valleys, which is what I figured RAH to be doing. What gets my attention is that for 2 days in a row the RAC of RAH has dropped (according to the local machine graph. their (BS) graph shows for the past 2 days a steady increase in RAC vs a drop according to RAH.

So I would gather from this difference that two different formulas are being used by each group to compute RAC vs one agreed upon standard.

Boincstats does not use the BOINC RAC formula.

kind of what i figured, so how does one compare RAH to the outside world of stats when it comes to RAC?

kind of what i figured, so how does one compare RAH to the outside world of stats when it comes to RAC?

RAH uses the standard BOINC RAC calculation. It's just the BOINCStats website uses a different calculation. Having said that, it's impossible to have cross-project parity on RAC as different projects suit different architectures differently (i.e. effect of RAM/Cache/FPU/GPU etc). Having said that, RAH does seem to grant lower credit that many other BOINC projects - don't have any figures to back that up though...

---

kind of what i figured, so how does one compare RAH to the outside world of stats when it comes to RAC?

RAH uses the standard BOINC RAC calculation. It's just the BOINCStats website uses a different calculation. Having said that, it's impossible to have cross-project parity on RAC as different projects suit different architectures differently (i.e. effect of RAM/Cache/FPU/GPU etc). Having said that, RAH does seem to grant lower credit that many other BOINC projects - don't have any figures to back that up though...

lower credit is a old discussion that has floated around here before. don't recall where. i am just trying to find a way to compare RAH to the outside world, which appears to not be possible when it comes to RAC.

now while they say that a 20 point drop is not that much. when you look at a graph of my RAC over 3 days now, it has done nothing but drop where as the task totals refute that and say it should be climbing at a steady rate. which leads me to this discussion. If totals of granted credit are equal to or greater than claimed credit over the past 3 days or even longer the line should hold steady or go up and not down as it is at the moment. While 20 points does not seem that much, when you are doing numbers 20 is allot. Now, how long will it take to get the graph back up to where it dropped and then get back to climbing and making up for lost RAC?

I want to compare the graph of RAH to a outside source to see if the two match up. BS comes close, but their RAC calculation is lower the RAH. Anyway....it would seem that RAH is accurate, but only over a long period of time and not in 1 day increments like I thought it would be.

Greg, one point you seem to be confused about is that RAC has nothing to do with your credit granted exceeding your credit claim. If you had previously been exceeding it by more, that is why your RAC is now tapering off.

ex: if 2 weeks ago all of your tasks were seeing credit granted exceed the claim by 10 points, and this week they only exceed by 8, your RAC is going to drop. The credit claimed has no effect on RAC.

I think boincstats and others use statistics that are published by the BOINC projects. But these statistics are generally only published daily, whereas the figures in your user profile are updated every time you report back completed tasks. So, depending on which day your work was recorded by any given reporting system, it can effect the figures.

---

Rosetta Moderator: Mod.Sense

I think boincstats and others use statistics that are published by the BOINC projects. But these statistics are generally only published daily, whereas the figures in your user profile are updated every time you report back completed tasks. So, depending on which day your work was recorded by any given reporting system, it can effect the figures.

I'm fairly sure BS uses a different formula to calculate RAC (but not 100%)...

---