Folding Forum

At the 21dec 0:00 UTC stats update, 3 WUs for 5187, 5061 and 5584 points where awarded to 3 of my clients (Martin_i7-875k_GTX560Ti, Martin_T7250 and Andre_i3-330).
Thanks for the extra points.

I'm not complaining, I'm just reporting the issue with the stats system.

Please dont confuse this with the stats outage and recredit of last weekend. Two of those clients had allready been credited with extra points during the recredit operation.

I'm a bit confused. Were you overcredited somehow? If so, please post the PRCG numbers for the WUs and they might be able to investigate further...

Yes I was overcredited.
How do you expect me to come up with PRCGs ? I dont know where the points come from.

Mactin wrote:Yes I was overcredited.
How do you expect me to come up with PRCGs ? I dont know where the points come from.

Interesting. I've never heard of anyone being overcredited before. How do you know that you were? The points are generated from the F@h servers, but your client keeps track of the WUs that you recently completed. I'm not sure what client you are using, but there should be a log file that has that information in it. It should record the Project Run Clone Generation numbers when it first starts the calculations on the Work Units.

For the latest EOC update I was expecting ~20K but received ~38K. Too many clients to analyze but I suppose it could have been from 1 or more servers that got missed during the big re-credit which occurred on the 14th or just a server that hadn't been updating stats properly.

bollix47 wrote:For the latest EOC update I was expecting ~20K but received ~38K. Too many clients to analyze but I suppose it could have been from 1 or more servers that got missed during the big re-credit which occurred on the 14th or just a server that hadn't been updating stats properly.

That might be likely, or you could have another person under your name or something like that. If your watching the graphs like that you'd know that sometimes things can just spike if a lot of WUs get completed on that day. If that's the case perhaps tomorrow will average things out. I mean I'm glad you got that many points (I wish I could ) but if there really is a problem the PG is going to need some more information. So far I'm not convinced there's a real issue here, since a lot of other things could be responsible. In order to prove it, you'd have to get the info on what WUs you completed and when, add up their points including bonus point calculations and whatnot, and then look at what the official stats gave you. It's my understanding that EOC looks at the .txt files that contain all the teams and all the usernames, so you'd probably have to look at that as well. The PG has a lot of redundancy and whatnot in their servers, which is why there were able to get everyone's points back accurately and in a timely manner.

I got two work units worth a total of 24,794 points in the 1:00 21 December update...

I'm averaging 1 work unit worth ~4,600 points every six hours folding regular SMP and one was about 'due' at that time (Project: 6951 (Run 0, Clone 78, Gen 368)). Therefore, my deduction would be that I also received a work unit worth around 20,000... I've only rarely folded SMP works units of that size (the last one worth 20,000 points was a p10127 on November 12) so I'm not sure whether the 20,000 points was a manual 'collation' of a number of work units that were missed during the downtime, although from my estimation the recredit that occurred earlier looked 'about right'...

Either way, happy for the points and a nice surprise... Maybe an early Christmas present?

Jesse_V wrote:That might be likely, or you could have another person under your name or something like that.

There isn't enough data to know whether somebody donated some work using your username or if there was an overcredit that was somehow related to the network outage and recredit last week.

If you know how much work should have been credited using a passkey (and you have not revealed that passey to anyone) you can go to the Stanford stats site and find out how many points were credited with your name AND your passkey, but that's not particularly useful unless you have a similar report from a couple of weeks ago. The Stanford site only reports current values, not historic differences.

There have been a few reports of extra credits being added during previous recrediting operations. That can't be ruled out this time. I have no idea when the recrediting actually finished (if, in fact, it is finished).

Project: 6951 (Run 0, Clone 78, Gen 368):

Hi k1wi (team 171641),
Your WU (P6951 R0 C78 G368) was added to the stats database on 2011-12-20 16:08:24 for 3032.61 points of credit.

Another possibility is that very few WUs from donors may not have been recorded (completely independent from the Server Outage). Since they are only a handful, no one notices it so every once in while, PG Members do a manual recredit to update those few WUs that were missed.

One possible source for credit of points is the server I reported in this post several days ago, viewtopic.php?f=18&t=20273. It has been accumulating returns since sometime last week, and had over 5200 waiting to be processed as of yesterday afternoon. Yesterday evening they had finally been sent to their work servers and the queue is now almost empty. I had at least a couple waiting there, and got a small uptick in points overnight from what I was expecting from ones completed yesterday.

Two of the overcredited WUs to me include a bonus, so if someone was folding under my name, he/she would also need my passkey which is highly unlikely. That is why I initialy concluded in an stats reporting glitch.
Again, thanks for the gift.

Mactin wrote:Two of the overcredited WUs to me include a bonus, so if someone was folding under my name, he/she would also need my passkey which is highly unlikely. That is why I initialy concluded in an stats reporting glitch.

If there's ANY additional information you can provide, I'd like to hear it. Nothing will change the previous credits, but I, for one, would like to track down this bug so that the next time there's a recredit, it's more accurate. I'm willing to do some additional research but don't know exactly what I can use. I also don't know what you can provide.

Which two bonus projects MIGHT have been part of your recredit? Are you certain that they were folded with the name "Mactin" and for your team and with your passkey? Presumably they've all been credited SINCE the network outage. Can you narrow the credit time-date any closer than that? Were any of these WUs returned to a Collection Server or did they all go directly to a Work Server? Can you think of anything else that might help?

At this point, my working assumption is going to be that you received double credit for these two WUs. [If you consider any of your answers confidential, send me a PM, but please don't overload my PM inbox.] Right now, I do NOT need "me too" data from anyone else.

Bruce, Mactin is my team name.
In my original post, I mentioned :
5187 points for user Martin_i7-875k_GTX560Ti, probably a 7620, 7621 or 7622 series WU, conmsistent with what the client folds.
5061 points for user Martin_T7250, a "bonus" WU, this user runs a single SMP client.
5584 points for user Andre_i3-330, a "bonus" WU, this user runs a single SMP client.

Upon further investigation this is a case of double recrediting.
I folded dozens of the 762x WUs during the outage.
I got cretided 5061 points for 7809-6-400-15 that completed folding on 2011-12-11 19:00:56 and was recredited allready on 14dec 23U00.
I got credited 5584 points for 7809-2-223-20 that completed folding on 2011-12-11 02:14:48 and was recredited allready on 14dec 23U00.

I think we have some miscommunication :
These two,I think,where credited twice

I got cretided 5061 points for 7809-6-400-15 that completed folding on 2011-12-11 19:00:56 and was recredited allready on 14dec 23U00.
I got credited 5584 points for 7809-2-223-20 that completed folding on 2011-12-11 02:14:48 and was recredited allready on 14dec 23U00.

Yes, they were. I've forwarded that information to the folks that worked on the recredit to see what might be learned so that the next time a recredit is done, it might be more accurate. There were some subtle differences in the first and second credit of each pair but I don't know what they mean.

Mactin wrote:I folded dozens of the 762x WUs during the outage.

So it appears. I found several of them but it would take a lot of searching to find them all. Even if I did, I don't know an easy way to find a specific pair that is duplicated.