My RAC is currently about 716 and climbing, but my daily output seems to be approx 2750.
Is there a relation between daily output and RAC, and if so what is it? Or is RAC merely an indication which is constantly adapting to the output? In any case, if the output becomes stable, the RAC should also, not? Then, will the the RAC be equal to the daily output, or not?
... completely confused ... :-/

What do you mean by Daily Output? Number of WUs, CPU seconds? Apples...?

---

Have a read of this wiki page

Let us know if that doesn't help.

---

What do you mean by Daily Output? Number of WUs, CPU seconds? Apples...?

Credits per day.
Which is what drives many competing DC teams.

Have a read of this wiki page

Let us know if that doesn't help.

I read it during the weekend. Too hastly perhaps, eager to grasp as much as possible in no-time ;-) And Paul's boinc wiki is so vast ...
I re-read it just now, and the bold sentence at the end saying
"The simplest way to picture it is that Total Credit is like your odometer, and Recent Average Credit is like your speedometer."
leaves me with the impression there is probably no real calculated relation between RAC and daily credits. Right?
OTOH, doing x km (or miles) per hour (speedometer) should give an idea where I will be in 24 hours (odometer). So I'm still confused.

But with RaH there is no such thing as pending credit, so credit is acquired immediately, hence the reason why the webpages update the moment a result is send.

ColdRain, I feel your pain. The wiki page is accurate, but what it doesn't talk about is how painfully slow it is to get your real RAC displayed in the Rosetta stats, despite how frightingly quick it is to lose it.

The best way to understand that is to look at boincstats. The Boincstats RAC indicator is quite accurate and 'snappy'. If you bring 700 credit a day, your boincstats rac is roughly 700 a day. But if you look at Rosetta's stats page, your RAC there will be 150 on Monday, 170 on Tuesday, 190 on Wednesday, etc... until you hit 700 and stay there.

For example today my RAC according to Rosetta is 497, but according to Boincstats it's 587. Guess which is right? Yup, boincstats.

What makes me a bit mad is how fast on the other hand you can lose your Rosetta RAC. Turn off all your machines for 3 hours, and your RAC drops dozens of points. What makes me even more mad is how at the lower levels of the competition a user with one machine can 'jumpstart' at 500 RAC, but others with 3 boxes slowly climb up the ladder, 10 points at a time. Seems random and link to how many credits you get on that first WU.

Overall though, I suppose stats heads like me are the minority and most people don't care or use the total credit as a benchmark.

---

Team CFVault.com
http://www.cfvault.com

... I suppose stats heads like me are the minority and most people don't care or use the total credit as a benchmark.

I wouldn't be too sure 'bout that, especially with the influx of new RaH but old DC people nowadays ;)
For instance the team I'm on, Dutch Power Cows (DPC for short), is involved in about 10 to 12 DC projects. DPC members (there are several thousands) tend to choose a project that is close to what they like, some like number crunching, some like seti, some medical prjects. But once they've choosen, they all want "credits for the bucks" :) People on our team tend to be quite sharp and critical when it comes to this. Which is maybe not all too scientific responsible, but nevertheless very understandable. Credits should be fair, nothing more, nothing less.
On the other hand, I know there might be (technical or architectural maybe infrastructional) reasons why the credit system maybe isn't perfect. Which won't stop me though from trying to comprehend the mechanisms behind it ;-)

RAC is also extrodinarily sensitive to the timing of the "hits" on credit. So, a person like me with many computers, lots of returns on a fairly regular basis would have a "better" RAC than someone with dial-up that only connects every day or so ...

Worse, there is a bug, though I was not able to get Dr. Anderson too interested in it, where if you have too many results credited AT THE SAME TIME, you can lose some ... :(

Basically, the results have to "hit" more than one second apart ... if they don't, well, oops ...

---

And one more thing: that 'statistics' panel in boinc is well.. not really responsive. For example it flatline yesterday all day for me, then jumped up all of a sudden today. I know it only only update itself when reporting back to the server, but still, you are better off using something like boincstats instead.

---

Team CFVault.com
http://www.cfvault.com

RAC is also extrodinarily sensitive to the timing of the "hits" on credit. So, a person like me with many computers, lots of returns on a fairly regular basis would have a "better" RAC than someone with dial-up that only connects every day or so ...

Worse, there is a bug, though I was not able to get Dr. Anderson too interested in it, where if you have too many results credited AT THE SAME TIME, you can lose some ... :(

Basically, the results have to "hit" more than one second apart ... if they don't, well, oops ...

All the more reason to get the "bug" fixed that waits !! to report (as opposed to uploading) the completed WUs.
Anyone with a queue size (connect interval) longer than the time to crunch one WU will end up reporting multiple WUs at the same time - getting royally screwed on RAC. In a system like R@H which doesn't need a quorum, they can very well get credited all at once.

---

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)

And one more thing: that 'statistics' panel in boinc is well.. not really responsive. For example it flatline yesterday all day for me, then jumped up all of a sudden today. I know it only only update itself when reporting back to the server, but still, you are better off using something like boincstats instead.

... or you could manually punch the "Update" button all day long :)

---

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)

And one more thing: that 'statistics' panel in boinc is well.. not really responsive. For example it flatline yesterday all day for me, then jumped up all of a sudden today. I know it only only update itself when reporting back to the server, but still, you are better off using something like boincstats instead.

... or you could manually punch the "Update" button all day long :)

On my Linux boxes I have this in my crontab:
# BOINC
*/15 * * * * /opt/BOINC/boincupd &>/dev/null
which does nothing more then:
#!/bin/bash
cd /opt/BOINC
./boinc_cmd --project https://boinc.bakerlab.org/rosetta/ update
exit

On my Windows boxes I downloaded an optimized client from http://boinc.truxoft.com/ which has the option "return_results_immediately" in some config file (you have to create if it doesn't exist, see the webpage).

for every solution there's a problem ;-)

On my Windows boxes I downloaded an optimized client from http://boinc.truxoft.com/ which has the option "return_results_immediately" in some config file (you have to create if it doesn't exist, see the webpage).

I just tried this approach, as detailed on truxoft.

I've been running the 5.3.1 BOINC Windows client for a few days now.

I created a remote_hosts.cfg file (that didn't already exist) and added a single line

# return_results_immediately

Saved the file ( no - it does NOT have any hidden .txt extensions or anything of the like).

Stopped and re-started BOINC, and the first WU to finish did the usual - uploaded as soon as it was done, and is now sitting there in "Ready to report" status.

So - what more does it take to make it work, or are you describing some other process?

edit :

This thread on SETI/BOINC board I guess the 5.3.1 client has been fixed to get this functionality working again. I had a Nov 1 date file, and it was reported to be patched on Nov 6. I'm off to download it again and try.

Results to follow.....


edit #2 :

Works fine with the correct patched client!!

[size=10]
11/9/2005 7:53:13 PM|rosetta@home|Computation for result 1hz7A_abrelaxmode_random_gauss_length20_jitter02_28463_0 finished
11/9/2005 7:53:13 PM|rosetta@home|Starting result 1hz7A_abrelaxmode_random_gauss_length20_jitter02_28459_0 using rosetta version 478
11/9/2005 7:53:15 PM|rosetta@home|Started upload of 1hz7A_abrelaxmode_random_gauss_length20_jitter02_28463_0_0
11/9/2005 7:53:20 PM|rosetta@home|Finished upload of 1hz7A_abrelaxmode_random_gauss_length20_jitter02_28463_0_0
11/9/2005 7:53:20 PM|rosetta@home|Throughput 25592 bytes/sec
[b]11/9/2005 7:53:24 PM|rosetta@home|Sending scheduler request to https://boinc.bakerlab.org/rosetta_cgi/cgi
11/9/2005 7:53:24 PM|rosetta@home|Reason: To report results
11/9/2005 7:53:24 PM|rosetta@home|Reporting 1 results[/b]
11/9/2005 7:53:29 PM|rosetta@home|Scheduler request to https://boinc.bakerlab.org/rosetta_cgi/cgi succeeded
[/size]

---

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 meaningless for a number of reasons, so don't worry...

1) It's driven by when the project grants the credit, not when the work is completed. There's a major gap possible when either the project validates (not an issue here), or the work is actually reported (not consistant based on dial-up vs BB connection, or how often the "update" button is pressed).

2) is a "project centric" view, in that RAC doesn't include work for other projects. And after all, BOINC is designed to be "multi project", so those that only crunch for one project may have a higher RAC than someone that does muich more work, but for other projects.

RAC should just be dropped from all Boinc projects, as it's a point of contention, it seems, on every Boinc project, and in reality, it means nothing.

---

RAC is meaningless for a number of reasons, so don't worry...

1) It's driven by when the project grants the credit, not when the work is completed. There's a major gap possible when either the project validates (not an issue here), or the work is actually reported (not consistant based on dial-up vs BB connection, or how often the "update" button is pressed).

2) is a "project centric" view, in that RAC doesn't include work for other projects. And after all, BOINC is designed to be "multi project", so those that only crunch for one project may have a higher RAC than someone that does muich more work, but for other projects.

RAC should just be dropped from all Boinc projects, as it's a point of contention, it seems, on every Boinc project, and in reality, it means nothing.

Ah - but within this single project it should work fine :)
Not everyone is a David Anderson BOINC disciple - running every project that comes along.

1) No quorum to wait for credit

2) With the return_results_immediately flag enabled, the results are reported in a timely, consistent manner. RAC should improve - I'll be waiting to see.

After all, "Credits are everything!" and you can quote me.

Angus
Free-DC
#15 in RAC and looking for improvement

---

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)

Ah - but within this single project it should work fine :)
Not everyone is a David Anderson BOINC disciple - running every project that comes along.

1) No quorum to wait for credit

Actually, no... Because of the delay when results are actually reported!

2) With the return_results_immediately flag enabled, the results are reported in a timely, consistent manner. RAC should improve - I'll be waiting to see.

Doesn't function in the 5.x.y clients. It doesn't complain about the flag, it just doesn't use it!


RAC is meaninless! I've been dealing with Boinc for 2 years, and it's only the newbees that seem to not understand that!

I see you have 91k credits here, but have you joined other projects? Try CPDN to see how well RAC fits into your score....

Trust me.. Dr. A from Seti might have a pic of me on his dartboard. I'm not a fan of his by any means. But I've been around enough BOINC projects to know that RAC is meaningless, even if you attach to one other project as a backup for when your primary project is down. Rosetta at 99%, CPDN 1% will show this..

And right now, today, the seti site as the top news item, tells people to join this project!

RAC is a joke!

---

Ah - but within this single project it should work fine :)
Not everyone is a David Anderson BOINC disciple - running every project that comes along.

1) No quorum to wait for credit

Actually, no... Because of the delay when results are actually reported!

2) With the return_results_immediately flag enabled, the results are reported in a timely, consistent manner. RAC should improve - I'll be waiting to see.

Doesn't function in the 5.x.y clients. It doesn't complain about the flag, it just doesn't use it!

If you had actually taken the time to read my previous post, it does indeed work just fine in 5.2.x. with Trux's client.

I don't think anybody really asked for opinions on the worth RAC - we were having a polite discussion on how it's calculated and what affects it.

---

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)