View Full Version : Credit recalibration at PG
valyn42
07-02-11, 05:37 PM
I haven't seen anyone else post about it... from the PG message board http://www.primegrid.com/forum_thread.php?id=3467#37798:((
spingadus
07-02-11, 07:05 PM
Wow. That sucks.
I"m only a part time credit ho, but it's still a downer for me. I've been working on my CGW badges at the expense of credits. Time to switch back to PPS before the credits go down.
Just 39M to my 100M mark and then I'm off to other projects. Glad I didn't get a second GTX 580. What will be the next best, Moo, MW, Colataz???
I am going to go for the 100 million mark and move over to gpugrid. Don't know how the credits are but I will put 5 gpu's on it and see what I can do.
I'd go over now but I want to get into the top 10 for the team (and do a little stomping of joker along the way.)
As I only have a 275GTX and a 470GTX on the project, you should be able to stomp me pretty easily.
As I only have a 275GTX and a 470GTX on the project, you should be able to stomp me pretty easily.And don't think Joker won't enjoy that. :)
Welcome back on the boards J. Your absence was noticed.
YoDude9999
07-03-11, 01:01 AM
Well, that just bites ass for sure.
Fire$torm
07-03-11, 01:08 AM
Well, that just bites ass for sure.
AMEN to that bro.
I'm beginning to look at this in the positive way. I never cared about prime numbers, so now I can put my 1 Nvidia GPU on some project that might even be useful.
As I only have a 275GTX and a 470GTX on the project, you should be able to stomp me pretty easily.
In that case I will just creep by instead of doing a stomp. I just want to get to the 100 million mark and in the top 10 for the team.
Crap crap, this is what brought me out of the masses of average crunchers up into sniffing area of the heavy weights. Either way, I am full tilt on it until it is backed down, then I will add on some MooWrapper. :(
spingadus
07-03-11, 11:46 AM
It's not clear to me how the credits will eventually match up with other projects. I'm happy with my new 590 regardless and I do want to run non nvidia projects as well. So this helps with purchasing decisions...
I guess this means my next pc will have amd cards.
It's not clear to me how the credits will eventually match up with other projects.
When they take it down by the full 50% then a GTX 570 on PG will produce about the same credits (or slightly less) than a 5870 on Moo or DNETC.
When they take it down by the full 50% then a GTX 570 on PG will produce about the same credits (or slightly less) than a 5870 on Moo or DNETC.
Bryan,
I have a GTX 580 and two ATI 5870s in CF. They produce about the same. 650K on PG and 600K on Moo. Two 5870s cost the same as one GTX 580, with some 5870s to be had for sub $200.
Dan
Oh ya and won't this make it about twice as hard to catch SIC?
YoDude9999
07-04-11, 05:18 AM
As far as catching sic goes it shouldn't make any difference. They are taking the hit too, just like everyone else is.
As far as catching sic goes it shouldn't make any difference. They are taking the hit too, just like everyone else is.
I'm horrible at math problems, so help me out here. So lets say that Team X is 100 points ahead of Team Y. Team X makes 2 points a day and Team Y makes 4 points a day. In how many days will team Y catch Team X? If we half the points both teams make each day then how long will it take to catch up. I get 50 vs 100 which would be twice as long. What do you get? So in this case Sic just had their lead effectively doubled on PG.
Dan
I'm horrible at math problems, so help me out here. So lets say that Team X is 100 points ahead of Team Y. Team X makes 2 points a day and Team Y makes 4 points a day. In how many days will team Y catch Team X? If we half the points both teams make each day then how long will it take to catch up. I get 50 vs 100 which would be twice as long. What do you get? So in this case Sic just had their lead effectively doubled on PG.
Dan
Yep.
(delta credits) / (delta rate of closure) = time to overtake
100 / (4-2) = 50 days
100 / (2-1) = 100 days
So the net effect is that their lead has doubled or another way to say it ... it will take twice as long to catch them.
YoDude9999
07-05-11, 04:05 AM
Yes, it will take twice as long to catch them, but also remember, it's taking them twice as long to get things done too, so, it's not like they are going to be pulling away from us at an increased rate as the ratio still remains the same (which is what I meant by stating, "shouldn't make any difference").
Fire$torm
07-05-11, 12:05 PM
Yes, it will take twice as long to catch them, but also remember, it's taking them twice as long to get things done too, so, it's not like they are going to be pulling away from us at an increased rate as the ratio still remains the same (which is what I meant by stating, "shouldn't make any difference").
Says the "Under 40" guy......... :P
The "Recalibration" is a total Crock-O-____ !!! On PG's forum, one volunteer tester (http://www.primegrid.com/forum_thread.php?id=3467&nowrap=true#37824) even states
CPUs and GPUs are much faster than in the past and a credit-recalibration is normal in all projects of the boinc community.
What a lame answer. I have never seen where PG computes credit based on time. I HAVE read PG Statements where they say their "Formula" was based on quantity of work done. The problem is as they state in this announcement that credit/wu will decrease BUT crunch time/wu will not. Sounds like a cave-in to "DA" and his Demigod mindset.
Says the "Under 40" guy......... :P
The "Recalibration" is a total Crock-O-____ !!! On PG's forum, one volunteer tester (http://www.primegrid.com/forum_thread.php?id=3467&nowrap=true#37824) even states
What a lame answer. I have never seen where PG computes credit based on time. I HAVE read PG Statements where they say their "Formula" was based on quantity of work done. The problem is as they state in this announcement that credit/wu will decrease BUT crunch time/wu will not. Sounds like a cave-in to "DA" and his Demigod mindset.
The only way to protest is with your feet as the saying goes, or GPU in this case. I'm gone after my 100M and may not participate in the challenges.
YoDude9999
07-06-11, 02:54 AM
The problem is as they state in this announcement that credit/wu will decrease BUT crunch time/wu will not. Sounds like a cave-in to "DA" and his Demigod mindset.[/Opinion]
What totally PI$$ES me off, is that, "supposedly" the projects can dish out whatever credit they want to. This was proven with Moo in the beginning of their project. They doled out very high credits in the beginning of the project and then made adjustments to it later as time progressed. Now, they are pretty much inline with Dnet as far as credit is concerned and because the projects are very similar, this make complete sense to me. Okay, that's all fine and dandy.
The thing is, PG sounds as though they are making this adjustment based on PPS Sieve WUs being converted to a different type of WU, or mathematical algorithm. Why didn't they convert (raise) the credit of the "other" calculations to match the credits given by the PPS Sieve project instead of reducing the credits to match something that gives less?
Every time we see credit, "adjustments", it's ALWAYS to a lower value then before and never an increase. Why is that? I believe F$ has made a correct statement, a "cave-in to DA"........grrrrrrrr [/Additional_Opinion]
Cruncher Pete
07-06-11, 03:31 AM
The only way to protest is with your feet as the saying goes, or GPU in this case. I'm gone after my 100M and may not participate in the challenges.
I agree completely with that statement. It is against my principal to support a project that before its cycle is finished decide to half their credit. I also suspect that they are pressured by DA of Bercley fame to reduce credit because they are so high. If that is the case, than why are they doing it mid stream, giving players who are chasing those ahead of them an unfair advantage. What some fail to see is that those ahead of you got there with credits that were twice as much as it is now (or will be), although now we are also on equal field, it will take us twice as long (in time) to make up that difference. In any case, you will not see me supporting them with cash donations or participation in their challenges. It costs me heaps to maintain this hobby of mine and I will do it according to my terms and will support those teams that are at least fair even though at a lower credit.
I agree completely with that statement. It is against my principal to support a project that before its cycle is finished decide to half their credit. I also suspect that they are pressured by DA of Bercley fame to reduce credit because they are so high. If that is the case, than why are they doing it mid stream, giving players who are chasing those ahead of them an unfair advantage. What some fail to see is that those ahead of you got there with credits that were twice as much as it is now (or will be), although now we are also on equal field, it will take us twice as long (in time) to make up that difference. In any case, you will not see me supporting them with cash donations or participation in their challenges. It costs me heaps to maintain this hobby of mine and I will do it according to my terms and will support those teams that are at least fair even though at a lower credit.
Agreed. Let's assume for a minute (I did not), but let's just assume that I won the lottery last night, and now it is my goal to catch Steve* and become the #1 Overall BOINC producer. It would now, in theory, take me twice as long to crunch the # of credits Steve* crunched, and therefore severely hamper my ability to catch him.
Now, I have nowhere near the cash or even desire to do something like that. . .so it's a theoretical argument for me. But I am SURE there IS someone out there, that this is NOT a theoretical argument for. There is a lot of money and resources out there, and someone has them...and that someone who may be a new cruncher, might want to become #1 Overall...this is a severe handicap being dished out now.
On that note, I am getting every PG credit I can right now - CPU and GPU both while the getting is good. What day do I need switch???
I would STRONGLY suggest NOT participating in the PG Challenges. The only thing the admins will understand is diminished production on the project. In reality, I don't think many will actually pull off because even at 1/2 credit there is nowhere to park your Cuda that pays more.
I will pull off at 250M which was my goal anyway. Two weeks ago I pulled 1 card off to run GPUGrid and even with 3 it should have taken me 30 days to get there. A week later they start dropping credits so I put the card back on ... still gonna take 30 more days. Now here we are almost a week later and the new wu take 10% longer ... so still need 30 more days. That is gonna be the longest 30 days of my life :))
zombie67
07-07-11, 11:21 PM
[...]In reality, I don't think many will actually pull off because even at 1/2 credit there is nowhere to park your Cuda that pays more.
Yep.
I will pull off at 250M which was my goal anyway.
Ditto (http://en.wikipedia.org/wiki/Ditto_machine). (Anyone else old enough to know where the term (and that addictive smell) comes from?)
Fire$torm
07-08-11, 01:05 AM
Yep.
Ditto (http://en.wikipedia.org/wiki/Ditto_machine). (Anyone else old enough to know where the term (and that addictive smell) comes from?)
Yep, used to help a few of the teachers at school 6th ~ 8th grade because the smell didn't bother me. I like the smell of gasoline too, Go figure.
The best time to dump would be all at once, but with all our individual goals, I'm sure it will be a gradual loss of support, so they won't even notice. I figure my 100 mil won't be reached for 2 months.
trigggl
07-08-11, 06:45 AM
The best time to dump would be all at once, but with all our individual goals, I'm sure it will be a gradual loss of support, so they won't even notice. I figure my 100 mil won't be reached for 2 months.
Well, I have a couple 4870's coming and I'm nowhere near my next milestone, so I may be stopping real soon.
My goal has been 100 million in PG. I should hit it within a month barring any GPU or computer problems, then I am going to put everything back on Medical scientific projects again. I went with PG just to be able to say that I got 100,000,000.
Hmmm...well, I'm at 111Mil today; looks like I'm down to 1.2Mil PG credits per day now. Is there a 150Mil milestone, or does it jump from 100Mil to 200Mil?:confused: I suppose I'll probably be on it until they really drop the credits at least - just trying to get every bit in that I can for the team now. The more we can gain on LAF in overall points, the better in my book.
zombie67
07-08-11, 03:05 PM
Hmmm...well, I'm at 111Mil today; looks like I'm down to 1.2Mil PG credits per day now. Is there a 150Mil milestone, or does it jump from 100Mil to 200Mil?:confused: I suppose I'll probably be on it until they really drop the credits at least - just trying to get every bit in that I can for the team now. The more we can gain on LAF in overall points, the better in my book.
They each increase my 2x or 2.5x:
10, 25, 50, 100, 250, 500, 1000, etc
(There are some 750 MMs listed, but I consider those false MMs. ;)
krackedpress
09-13-11, 12:56 PM
I would STRONGLY suggest NOT participating in the PG Challenges. The only thing the admins will understand is diminished production on the project. In reality, I don't think many will actually pull off because even at 1/2 credit there is nowhere to park your Cuda that pays more.
I will pull off at 250M which was my goal anyway. Two weeks ago I pulled 1 card off to run GPUGrid and even with 3 it should have taken me 30 days to get there. A week later they start dropping credits so I put the card back on ... still gonna take 30 more days. Now here we are almost a week later and the new wu take 10% longer ... so still need 30 more days. That is gonna be the longest 30 days of my life :))
I made 2 million for PG with PPS Sieve. Now the credits are low and time it takes is high. I used to crunch these in a few hours - under 8 - but now it takes about 20 or more for less credits. I was making 10K to 20K daily from PG PPS Sieve. Sorry, but for those loosing millions per day running GPU, but the loss of 10K to 18K per day with my CPU-only system is big news for me. I was making a million credits every 6 to 8 months, now that I have my quad. 2 Million out of 2.8 million of PG's credits were PPS Sieve. Now it will take me over a year to do a Million credits. I am at a total of 5.3 million and a few weeks ago it was going to take till the end of October to make 6M, but now it is the end of November. Before the credit/time switch, I was to have 6 million by now.
Also, what is going on with the credit estimates? The 1.5 hour estimates are run in less than half that time. Riesel is telling me it will take 45 hours, when a machine I host for the team is doing it with an estimate of 5-6 hours but the actual time is about 8 hours. Still 8 hours is no where near 45 hours. I usually do about 5-10% longer than the hosted machine, since I am doing a lot more than just crunching [like the dual core is doing]. So the estimates are totally messed up. I had one PG project [forget which one] gave me an estimate of over 800 hours, but ran in about 10 hours.
So, since PG is too wacky, what CPU-only project would be good to go to? Which PG project is now giving the best credits, since I want to make my 3 million goal for that one.
I have 6 non-PG project between 100K to 334K [low to high - SIMAP, DiRT, Collatz, PrimaBoinca, MW, DNETC], and FreeHal at 776K. The rest [22] are below 50K and 16 below 25K. I would like to run the higher credit ones. I run DiRT, PG, WUProp, and FreeHal right now. I want to get my 3M goal for PG, but with the current project/credit/time for PPS LLR [focus project] it will take about 70 days. So I would like to know which PG project is now the best daily credits [maybe best credits/crunch-time] to finish my goal before I loose my interest in PG, my once best credit maker.
I prefer to run two main project in my queue, instead of only PG or DiRT for all 4 cores on my system.
Any Help, or advice?
Yes I know the last post for this thread was in July, but it was the one that looked the best fit for my question.
Tim, I think that will depend on your GPU. MW credits are still decent, I have not run the others. Maybe look at MooWrapper. Some guys on the team are getting BONKERS numbers from that project. I think I have one little lappy pecking away at it from time to time because that is the only GPU project it can do.
Slicker
09-14-11, 12:51 PM
DistrGen (http://boinc.freerainbowtables.com/distrrtgen/) has pretty good credit right now. I'm getting 27K per day from my I7-920 -- and that's only with 7 cores since I have the last core reserved for Moo Wrapper.
krackedpress
09-14-11, 01:59 PM
Tim, I think that will depend on your GPU. MW credits are still decent, I have not run the others. Maybe look at MooWrapper. Some guys on the team are getting BONKERS numbers from that project. I think I have one little lappy pecking away at it from time to time because that is the only GPU project it can do.
I do not have any GPU. I am doing everything on a CPU. So all I have is a quad AMD Phenom 9650 @ 2.3 GHz.
As for Moo, it does not work on my CPU system.
krackedpress
09-14-11, 02:09 PM
DistrGen (http://boinc.freerainbowtables.com/distrrtgen/) has pretty good credit right now. I'm getting 27K per day from my I7-920 -- and that's only with 7 cores since I have the last core reserved for Moo Wrapper.
If I run only DiRT on the quad I get under 10K daily. Some days I get nothing, then I get a backlog of credits all at once. I earned 47K once, but that was 2.5+ days of reporting instead of a standard daily credits. Right now, I am earning about 4.7K daily now, with some PG running in the queue with DiRT.
To be honest, I did not know that the i7 was an 8 core system. Last I heard, the "i" series was only at 6 cores. I know that the AMD has a 6 core CPU [maybe 8 by now], but I do not know if my motherboard will support it.
Fire$torm
09-14-11, 07:54 PM
If I run only DiRT on the quad I get under 10K daily. Some days I get nothing, then I get a backlog of credits all at once. I earned 47K once, but that was 2.5+ days of reporting instead of a standard daily credits. Right now, I am earning about 4.7K daily now, with some PG running in the queue with DiRT.
To be honest, I did not know that the i7 was an 8 core system. Last I heard, the "i" series was only at 6 cores. I know that the AMD has a 6 core CPU [maybe 8 by now], but I do not know if my motherboard will support it.
He is referring to threads not cores. For the Intel CPUs WITH Hyper Threading (HT) a quad core can run 8 threads simultaneously (4 physical cores and 4 "virtual" cores).
For Intel Core series based on Nehalem microarchitecture
Core i3 (Clarkdale): Dual Core w/ HT
Core i3 (Arrandale): Mobile Dual Core w/ HT
Core i5 (Lynnfield): Quad Core w/o HT
Core i5 (Clarkdale): Dual Core w/ HT
Core i5 (Arrandale): Mobile Dual Core w/ HT
Core i7 (Bloomfield): Quad Core w/ HT Socket LGA 1366
Core i7 (Lynnfield): Quad Core w/ HT Socket LGA 1156
Core i7 (Clarksfield): Mobile Quad Core w/ HT Socket G1
Core i7 (Arrandale): Mobile Dual Core w/ HT Socket G1
For Intel Core series based on Westmere microarchitecture
Core i7 (Gulftown): Hex Core w/ HT Socket LGA 1366
Wikipedia:
List of Core i3 CPUs (http://en.wikipedia.org/wiki/List_of_Intel_Core_i3_microprocessors)
List of Core i5 CPUs (http://en.wikipedia.org/wiki/List_of_Intel_Core_i5_microprocessors)
List of Core i7 CPUs (http://en.wikipedia.org/wiki/List_of_Intel_Core_i7_microprocessors)
Edit: Oooops Forgot this....
My Dirt stats for Monday and Tuesday
ID # ___ Tuesday __ Monday ___ Avg _____ CPU
22006 (http://stats.free-dc.org/stats.php?page=host&proj=drt&hostid=22006) --- 11,400 --- 16,000 --- 100,000 --- Phenom II 810 (Quad Core)
22005 (http://stats.free-dc.org/stats.php?page=host&proj=drt&hostid=22005) --- 28,200 --- 27,000 --- 052,829 --- Core i7 920
23048 (http://stats.free-dc.org/stats.php?page=host&proj=drt&hostid=23048) --- 33,000 --- 30,600 --- 033,943 --- Core i7 860
22004 (http://stats.free-dc.org/stats.php?page=host&proj=drt&hostid=22004) --- 01,800 --- 01,800 --- 006,286 --- P4 @2.8Ghz
23049 (http://stats.free-dc.org/stats.php?page=host&proj=drt&hostid=23049) --- 03,000 --- 02,400 --- 002,657 --- Pentium M @1.7Ghz
Slicker
09-15-11, 09:44 AM
He is referring to threads not cores. For the Intel CPUs WITH Hyper Threading (HT) a quad core can run 8 threads simultaneously (4 physical cores and 4 "virtual" cores).
Correct. Sorry about the confusion.
krackedpress
09-15-11, 09:57 AM
"He is referring to threads not cores. For the Intel CPUs WITH Hyper Threading (HT) a quad core can run 8 threads simultaneously (4 physical cores and 4 "virtual" cores)."
Correct. Sorry about the confusion.
So how do you access these "virtual cores" on an quad AMD Phenom 9650? I am running 4 cores, plus 2 FreeHal and 1 WUProp.
If I could run a few more "cores", it might help with my lack of daily credits I use to get from PG PPS Sieve. AP26 is gone, and now this one has cut the credits and take much more time to run for those credits. DiRT is not giving me much "per core" as PPS Sieve did.
But, if I could get my virtual cores to run DiRT, then it might work out in the long run.
Fire$torm
09-15-11, 01:03 PM
"He is referring to threads not cores. For the Intel CPUs WITH Hyper Threading (HT) a quad core can run 8 threads simultaneously (4 physical cores and 4 "virtual" cores)."
So how do you access these "virtual cores" on an quad AMD Phenom 9650? I am running 4 cores, plus 2 FreeHal and 1 WUProp.
If I could run a few more "cores", it might help with my lack of daily credits I use to get from PG PPS Sieve. AP26 is gone, and now this one has cut the credits and take much more time to run for those credits. DiRT is not giving me much "per core" as PPS Sieve did.
But, if I could get my virtual cores to run DiRT, then it might work out in the long run.
You Cannot. The virtual core exists via Intel's HyperThreading technology which AMD does not have.
Powered by vBulletin® Version 4.2.4 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.