Predictor Stats - not right

[Ototero]

I've found a great stats page for Predictor.

http://www.setisynergy.com/stats/tea...79&project=pah

[Beerknurd]

So who all is running Predictor.....???? Should I be??

[Empty_5oul]

if u look at the numbers there beerknurd, u'll c only a few people are running it.
run what u want but i think we should have another discussion soon and chose 1 or 2 projects to all do, if we all spread out we dont get anywhere in any particular project.

[vaughan]

Hey "That Pentium Guy" I haven't been able to get any work from Predictor over the last few days. Site still says they will lpost when they are fully working again.
As mt said we need to focus. Maybe we can check possible targets again soon?

[Beerknurd]

I know AMD Users's land is a free country, and we can run whatever we want, but I agree with you. We need to have some kind of priority project.

Priority may be too strong of a word. Here are my views on the following projects.

Folding @ Home -- Never ran it.
Predictor -- I will run this once it becomes more stable. "It may already be"
Ubero -- I personally love this project. But I know "It's dead"
SOB -- This project pisses me off. I lost too many tests.
D2OL -- Ok but very boring for some reason.
Lifemapper -- I'm ok with this.
FAD -- Never ran it.
Chessbrain -- Building the world's largest chess computer does nothing for me...
Seti -- Never ran it.

I don't know if this helps or not..... I guess if I had to vote it would be...

Lifemapper
Predictor
D2OL

But what does everyone else think???

[Empty_5oul]

k
i wouldnt do predictor yet, its still jumping around and the coders are still "playing" as it were. the time limit was 3 days when i last treid it this week, though thats ok for most people its pretty quick. personally id leave this.

about SOB it doesnt really matter if you lose test, u still get credit for the amount you have processed and as it talks to the server after every time you complete 1 block, so i cant really see your prob.

also i have never yet managed to lose a save point in SOB, i tried doing a re-boot with it still running and it re-started fine, also there was powercut the other night n still ok next day, last night we blew a fuse and all upstairs circuit was down. when i booted earlier SOB still did a remember from cache or something and im now 48% through my test.

[Beerknurd]

Well for some reason, mine didn't start from cache. I'm willing to give it another try if that's what we decide to run.

[Ototero]

Whats's happening in Predictor.

We've moved from position 33 up to 22 without our points tally changing.

If this keeps up, we'll be the number 1 spot for crimpo

[AMDave]

Stu,

the way I remember it BC represented the AMD view at the Predictor rewrite that the AMD processed WUs were dissadvantaged in terms of the points allocation.

I recall BC saying that they were including in the rewrite that they were switching to a WU matching process by similar processor.

Thus, having fewer AMDs in the wild it would take longer for the same Wus to be validated by similar processors.

I also remember that at the time of the Predictor server crash there was a huge backlog of uploaded Wus to be validated.

Noteably when the server came back up our rank dropped considerably.

I would surmise that this was due in part to 2 things:
1 - it was due to the back log of AMD processed Wus waiting validation whilst the Intels were matching up WUs faster, Intels being more prevalent in the wild.
2 - the repair of the Predictor stats database (I recall it was also impacted in the crash) was lower on the agenda than getting the app back up and running.

I expect that this change in rank should be resulting from our old WUs getting matched up and validated and some repairs to the DB being made as well as the fact that the points should now also be being allocated more equitably as a result of the representations made by BC at the code change sessions.

* Nota Bene - I have not dived into the Predictor forums for any information to base my assumptions on. That is why they are assumptions pure and simple. I am merely basing this on the discussions that we had with BC at the time. Im not in a rush to dive back into preditor forums. All I can say is that this was a correction that I was expecting inthe long run but which has taken a longer time to come about than expected.

[AMDave]

ok ok.

I have had a closer look and on second thoughts...much of my surmise has got to be wrong because the points situ is not changing.

So in second cut I would put it down to the database corrections that may (or may not) be being done..

Either way the stats are in my opinion still not corect, but they are getting back towards what they should be.

I did not find much in the way of relevant info so I still don't know one way or the other what is actually going on. Lets keep plugging away and see what happens.