| Author |
Message |
|---|
John  Honorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
                 
|
|
Last year a thread was provided for feedback on the Challenge Series. A lot of good suggestions came from it. Therefore, here is another opportunity to offer your ideas, comments, complaints, etc. and we'll do our best to incorporate as much as possible.
The 2011 Challenge Series starts 3 February 2011 and will consist of 9 challenges again. That allows for about 2-3 weeks to offer your feedback before the full schedule is revealed.
Thanks!
____________
|
|
|
mfbabb2 Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 21,825,457
RAC: 0
                    
|
|
How about a 6 month long "parallel" challenge on SoB?
____________
Murphy (AtP)
 |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1292
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
I look forward to whatever PPS challenge PrimeGrid has this year. The PPS challenge is the one that stresses the servers the most due to the smaller workunits. I'm hoping that the PPS challenge lasts at least two weeks. |
|
|
|
|
|
I think that that would be fun and would also complete a significant amount of work on the PPS search, granted that is a project that doesn't exactly have an end goal like PSP/SoB, but it would move the leading edge significantly higher with how much computing power goes into the challenges.
____________
 |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1292
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
The problem right now with PPS is that it covers all k < 10000. I think that is too high because it has led to multiple concurrent PPS projects that cover different ranges. My personal preference is to target PRPNet port 10000. IIUC, that project has an end goal, testing all k < 10000 to n = 500,000. I think that port 11000 could go away, but I don't know what that port will be used for if it comes back on-line. My second preference is to target BOINC to finish all k < 10000 to n = 1000000. That is a much bigger task that will take many years. |
|
|
|
|
|
I do all of my PRPnet work on port 10K (apart from the work that I have done for the challenge), so I would definitely look forward to that being completed also. At between ten and forty seconds a work unit on my machine depending on the n value, I think that if we could focus more computing power on it it could finish up nice and quick.
____________
|
|
|
Honza Volunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 2078
ID: 352
Credit: 9,494,715,421
RAC: 3,366,694
                                                  
|
|
PPS LLR needs a big push fi we want to hold Top5000, 10 days or 2 weeks challenge if server can hold.
We are entering 600k, long way to 1M.
(not so long to 666666 where SGS reside).
____________
My stats |
|
|
|
|
How about a 6 month long "parallel" challenge on SoB?
Funny idea. Sounds good to me.
____________
There are only 10 kinds of people - those who understand binary and those who don't
 |
|
|
|
|
|
A challenge on SoB might be a really good thing, since it's so error-prone. With more people working nonstop, we could theoretically run through the existing buffer much more quickly than with the current amount of people.
PPS LLR is always a good one. I'd like that especially since I run it on all my machines and it never takes more than 27 to 28 minutes. Perhaps another idea for a parallel challenge would be PPS LLR and SGS LLR, since they test similar ranges (only SGS is higher, or am I wrong) - we could bump the upper shelf by a huge amount if everyone was doing those two LLRs...
And of course the GPU sieve. If another gets ported to GPU we could do that, it's just that PPS sieve is going pretty much indefinitely, whereas we have an "optimal depth" for all the others.
Hey, any subproject you feel like finishing really fast (cough cough PSP), have a challenge for it.
Perhaps 2011 will see more PRPnet projects coming to/being integrated with BOINC and/or PrimeGrid...I'd like that too.
____________
 |
|
|
 warddr Volunteer tester
Send message
Joined: 7 Feb 08
Posts: 254
ID: 18735
Credit: 70,267,107
RAC: 17,227
             
|
|
I would love a real manual challenge, you can even do both sieve and prpnet at the same time, and at the end you can count the PSA-credits everyone would get for that work.
The fact that there everyone will only know their result at the end of the challenge can make it interesting.
____________
 |
|
|
|
|
|
We need more GPU app...
Save Energy...
____________
|
|
|
|
|
|
I think we need multiple PPS events to return the project to significance. One gazillion day-long event would get dull quickly and my short attention span would drift to something shinier. |
|
|
Ross*Send message
Joined: 22 Apr 10
Posts: 66
ID: 59181
Credit: 1,199,022,485
RAC: 859,269
                  
|
We need more GPU app...
Save Energy...
Hi
I think we need more GPU challenges
From the comments about the last challenge re how much work got done, more effort neededs to convert more apps to GPU.
i.e on a 980x I can do 24 WUs per day PPS re CPU
with 1 GTX 580 I can do 240 WUs per day re GPU
i.e that is 1 GTX 580 does the same as 10 980xs
Cheers
Ross*
____________
|
|
|
Ken_g6 Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 969
ID: 3110
Credit: 366,832,466
RAC: 14,226
                      
|
|
Let's see...there are 16 projects on PrimeGrid.
- 3 are over. (AP26, TPS, and PSP Sieve)
- PSA encompasses a lot. Some of those projects are GPU-enabled.
- PPS Sieve is GPU-enabled.
- CW Sieve soon will be. :)
That leaves:
- 321 Sieve and TRP Sieve. Both use sr2sieve, and would require a port of that to the GPU. The math is more complicated than the other sieves, but somebody could probably do it. In the past I've suggested using Pollard Rho instead of the Baby Steps/Giant Steps algorithm.
- Eight existing LLR applications. That is the goal of PrimeGrid after all, to find primes. There's been lots of talk about LLR on GPUs on MersenneForum; but so far it's all vaporware.
____________
|
|
|
|
|
- CW Sieve soon will be. :)
Excellent!
I'm not exactly an expert on coding (cough) but I did think that the sr2sieve apps could be ported to GPU with little trouble...that's something to look forward to.
So I guess 2011 will see a challenge on the GCW sieve?
____________
|
|
|
Michael Goetz Volunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14688
ID: 53948
Credit: 1,042,672,583
RAC: 11,648
                             
|
|
I'm going to give a rather out of the box suggestion:
There should be no challenges for sub-projects which can run on a GPU.
The reason I'm suggesting that is because I'm assuming that at least part of the reason for having the challenges is to get people to focus on a single project. I suspect this helps some of the less popular projects get a significant amount of work done.
I think the GPU projects probably don't need much help in that area. :)
Not only are the GPUs able to get so much more work done, but since there's only one GPU project (for now), most of the available hardware is already focused on the one (or few) projects that are available.
____________
My lucky number is 75898524288+1 |
|
|
mfbabb2 Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 21,825,457
RAC: 0
|
|
I agree -- no GPU Challenges!
____________
Murphy (AtP)
|
|
|
|
|
|
I disagree. The challenge series should not focus on pushing single subprojects. It should consist of a wide range of PG subprojects to represent every aspect of PGs work (Sieve/LLR, CPU/GPU, etc.).
I suggest the admins could declare a sub project of the week/month/quarter/year to push a single sub project.
____________
|
|
|
Ken_g6 Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 969
ID: 3110
Credit: 366,832,466
RAC: 14,226
|
I'm not exactly an expert on coding (cough) but I did think that the sr2sieve apps could be ported to GPU with little trouble...that's something to look forward to.
I don't know what gave you that idea. Aside from the fact that the sr2sieve algorithm is intrinsically more complicated than the PPSieve algorithm, there's also a memory issue. sr2sieve's current algorithm, as I understand it, needs a significant (read:tens of kilobytes to megabytes) amount of random-access memory for a hash table. GPUs simply don't have that much quickly-accessible memory. Which is why I suggested Pollard Rho instead. Pollard Rho isn't the easiest algorithm either, and would tend to do worse on projects with more K's; but it doesn't need much memory at all.
____________
|
|
|
|
|
|
I disagree, there should be gpu challenge(s). If you look at primegrid's credit history here you can see that on the days of the proth sieve challenge primgrid was awarding up to 100,000,000 more credits per day. I assume that during the challenge some hosts were switched to primegrid rather than their usual projects. |
|
|
Vato Volunteer tester
Send message
Joined: 2 Feb 08
Posts: 989
ID: 18447
Credit: 2,380,616,839
RAC: 494,591
                        
|
|
There should be a good mix of all subprojects that are available, regardless of CPU vs GPU.
The Challenges should reflect the complete set of work that PG does.
(This is despite me losing ground massively due to not having much GPU power at all).
To put focus on an individual subproject outside of this can be achieved by a call to arms with a priorities posting and/or fiddling with the server settings.
____________
 |
|
|
John Honorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
There should be a good mix of all subprojects that are available, regardless of CPU vs GPU.
The Challenges should reflect the complete set of work that PG does.
Vato is on to something. :) Within the confines of a "balanced" Challenge series, all projects are considered.
The initial intent of the Challenges was to feature a project so users could learn more about that specific prime search. It was an attempt to expose these searches to the general BOINC community. This remains the goal.
Some projects lend themselves better to be included in the Challenge Series...but all projects are considered.
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1292
ID: 12001
Credit: 18,565,548
RAC: 0
|
Which is why I suggested Pollard Rho instead. Pollard Rho isn't the easiest algorithm either, and would tend to do worse on projects with more K's; but it doesn't need much memory at all.
Are you suggesting using Pollard Rho instead of BSGS for the discrete log? |
|
|
|
|
Vato is on to something. :) Within the confines of a "balanced" Challenge series, all projects are considered.
How about this for a 'balanced' idea...
Create a "Tutti-Frutti" Challenge in which all sub-projects must be crunched, with each user's final score being the score of the sub-project that achieved the lowest amount of credit.
This would require each contestant to plan carefully how to best distribute tasks among the resources he has available, and give a fair-shake to those with older&slower CPUs in their farm - with no unfair advantage for the GPU-bunnies!...
Anyone up for this?
____________
 |
|
|
Vato Volunteer tester
Send message
Joined: 2 Feb 08
Posts: 989
ID: 18447
Credit: 2,380,616,839
RAC: 494,591
|
|
I think that would be interesting!
Given the time that things like SoB takes, this would need to be at least a 2 week contest.
It might take a while to complete validation though...
____________
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
             
|
give a fair-shake to those with older&slower CPUs in their farm - with no unfair advantage for the GPU-bunnies!
Maybe the Project should just Ban the GPU's from the Project since they put the CPU's at such an Unfair Advantage ?
____________
 |
|
|
Vato Volunteer tester
Send message
Joined: 2 Feb 08
Posts: 989
ID: 18447
Credit: 2,380,616,839
RAC: 494,591
|
|
No - the GPUs do fantastic job of the work they're good at, but they can't do everything. Whilst the GPU folks will dominate the credit rankings because of this, they can't, as yet, find primes, or even sieve everything we're interested in. But It's all complementary, and all should be included in the Challenges, including a tutti-frutti should it happen.
____________
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
Whilst the GPU folks will dominate the credit rankings because of this
The Project had to know that would happen if they opened it up to the GPU "Bunnies" if they didn't want that then they should have kept them out. Personally I find it Insulting & Offensive to be even called a "Bunnie" ...
____________
|
|
|
Vato Volunteer tester
Send message
Joined: 2 Feb 08
Posts: 989
ID: 18447
Credit: 2,380,616,839
RAC: 494,591
|
|
I'm not sure the project necessarily knew it would draw as much GPU resource as it has, but that can only be a good thing, as not only PG will benefit, but also NPLB, RPS and a number of individual prime searchers. I sincerely doubt the project particularly cares about BOINC credit per se - more about finding primes - and massively-deep GPU-driven sieving is definitely helpful in that regard.
On a personal level, I'm ever so slightly sad that I've dropped down the credit rankings, but I can solve that by putting GPUs to work if I care enough to do so. That particular writing has been on the wall for some time, and isn't a reason to panic or whinge. The same situation happened to a lesser extent with 64bit versus 32bit sieving in the past. My plan is to embrace change at the speed my wallet allows.
BTW I won't call you (or anyone else) a "Bunny" (or anything else for that matter).
____________
|
|
|
rroonnaalldd Volunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
          
|
The Project had to know that would happen if they opened it up to the GPU "Bunnies" if they didn't want that then they should have kept them out. Personally I find it Insulting & Offensive to be even called a "Bunnie" ...
Yeahhhh I am a bunny 
...and now  (sports)...
____________
Best wishes. Knowledge is power. by jjwhalen
 |
|
|
mfbabb2 Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 21,825,457
RAC: 0
|
I think that would be interesting!
Given the time that things like SoB takes, this would need to be at least a 2 week contest.
It might take a while to complete validation though...
If each computer has to do all available sub-projects, it had better be a 2 month challenge! (At least for my old single core AMD!)
____________
Murphy (AtP)
|
|
|
|
|
|
I dislike the idea of a challenge being on every single subproject because it would take forever for those of us that just have one or very few computers to run things on.
I don't have a problem with one of the challenges being a GPU challenge, after all that is just about all that it can be right now. If people have the hardware to be able to help out, then let them help out in anyway that they can. The overall picture is to help complete the project by focusing as much resources on it as possible, the goal is not to ban GPUs from competing just so that you can continue to rock the charts with your CPUs.
____________
|
|
|
Ross*Send message
Joined: 22 Apr 10
Posts: 66
ID: 59181
Credit: 1,199,022,485
RAC: 859,269
|
I'm not sure the project necessarily knew it would draw as much GPU resource as it has, but that can only be a good thing, as not only PG will benefit, but also NPLB, RPS and a number of individual prime searchers. I sincerely doubt the project particularly cares about BOINC credit per se - more about finding primes - and massively-deep GPU-driven sieving is definitely helpful in that regard.
On a personal level, I'm ever so slightly sad that I've dropped down the credit rankings, but I can solve that by putting GPUs to work if I care enough to do so. That particular writing has been on the wall for some time, and isn't a reason to panic or whinge. The same situation happened to a lesser extent with 64bit versus 32bit sieving in the past. My plan is to embrace change at the speed my wallet allows.
BTW I won't call you (or anyone else) a "Bunny" (or anything else for that matter).
Hi
Well I did open a can of worms!!
The whole object is to find primes and if sieving helps great and if GPUs help just great too.
There will always be tension between those that happen to have slow older boxes and the "farms of boxes" and the GPUs.
I am sure who every controls the next year of challenges will look at what is the best mix for the best result.
I personally only have 4 main boxes and don't intend to get any more, and I want to do a verity of projects. I fine it rather intimating that some "Farms" can dominate the challeges.
That was until the last challenge, when GPUs called the shots.
So please see the GPUs as a opporunity to progress further faster than every.
Cheers
Ross*
____________
|
|
|
Ken_g6 Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 969
ID: 3110
Credit: 366,832,466
RAC: 14,226
|
Are you suggesting using Pollard Rho instead of BSGS for the discrete log?
Yes. Only on GPUs.
I just realized something else about this challenge series. In the past we've had a roughly equal balance between LLR and sieves. Right now we have nine challenge slots and four active sieves (not including PSA). 4/9 sieve challenges would be about the balance we had before.
However, 321 sieve is probably ending soon (unless someone wants to get creative with Pollard Rho on a GPU). That would leave 3/9 sieve challenges, a 1/2 ratio but still not bad.
Now, assuming we don't want more than one GPU challenge, that leaves either two sieve challenges (presumably CW and TRP) or just one (if someone gets creative with Pollard Rho on a GPU).
I guess a move to LLR is a natural progression for PrimeGrid. But how fast do we want to move there? Is it worth moving faster on this just to minimize GPUs in the challenges?
Oh, and remember, if/when someone comes up with an LLR client for GPUs, this GPU argument sort of goes out the window.
____________
|
|
|
|
|
Personally I find it Insulting & Offensive to be even called a "Bunnie" ...
Apologies if my image caused offense... that was certainly not the intention!
You may note that ruby badge - a clear mark of one of the 'bunnies', and I wouldn't be insulting myself now, would I?
The idea was to convey a sense of 'hopping ahead of the rest of the crowd' - as in 'tortoise-and-the-hare'...
..but if that (somewhat Euro-centric) image grates.., and since Cats (of all sizes) are far better tolerated on the Net in general.., I'm perfectly happy to dump it in exchange for a rather swifter, home-grown (Out-of-Africa) metaphor..
Henceforth we shall be known as the GPU-Cheetahs!
Better?
____________
|
|
|
|
|
|
My GPU power is very limited in comparison to my CPU power. But GPUs have fantastic output and so they speed up the process of finding primes quite a lot. Nobody wants to ban them from the project. It would be stupid to not accept the tremendous help the're offering. A good number of people will overtake me in the stats quite easily but that's fine. They do a lot of work, so they have to get the reward. It's time to move on and do things efficiently.
So why not have two challenges on the same GPU project? We had PPS(LLR) twice last year... (ok, it was PPS and PPSE...)
And I still like the idea of a SoB "background" challenge running for 3 months or so. These WUs are so long that people don't like to concentrate on them. There are short races on other projects etc. But having a long stretch of time with every WU done within counting sounds like fun. It should start rather early so we have the rest of the year for validation ;-)
____________
There are only 10 kinds of people - those who understand binary and those who don't
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
My GPU power is very limited in comparison to my CPU power. But GPUs have fantastic output and so they speed up the process of finding primes quite a lot. Nobody wants to ban them from the project
I never said anybody did want to Ban them from the project I just said why don't you Ban them from the Project. If you going to start running Challenges that take away their Unfair Advantage you may as well Ban them then.
I've never seen anybody complain about the Advantage the Usual Dominating Corporate/Collage CPU Pharms had in the Challenges but as soon as the GPU's are thrown into the mix right away there's talk of taking away their Unfair Advantage Advantage.
Be my Guest & do it, I've left Teams over just the same talk so it would be nothing for me to leave a Project (Not that anybody would give a rip) for a more Receptive GPU Project than here if that's what the mood is going to be here ...
____________
|
|
|
|
|
I never said anybody did want to Ban them from the project I just said why don't you Ban them from the Project. If you going to start running Challenges that take away their Unfair Advantage you may as well Ban them then.
Sorry, it seems I got you wrong there. My apologies.
I hope I could make it clear that I love the amount of power GPUs are adding to the project and so I think they should be rewarded by getting due credits AND challenges. This is not my native language so I have difficulties understanding / expressing the finer details sometimes.
I don't think anybody wants you to leave the project. I surely don't.
____________
There are only 10 kinds of people - those who understand binary and those who don't
|
|
|
Michael Goetz Volunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14688
ID: 53948
Credit: 1,042,672,583
RAC: 11,648
|
I dislike the idea of a challenge being on every single subproject because it would take forever for those of us that just have one or very few computers to run things on.
If you are including all sub projects, including SoB and the Cullen and Woodall LLRs, the 2011 tutti-frutti challenge probably should be for... 2011. All of it. The whole year.
It's not just that slow computers will take a wihle. Trying to run that many projects on even a fast multicore computer will be challenging because of the way BOINC refuses to schedule things the way you want it to. Getting what you want to run can be a micromanagment nightmare, and not all of us have the luxury of sitting in front of our computers 24/7.
So if you want to do something like this, it probably should be a parallel challenge that runs for the whole year.
This brings up two potential challenges for the admins:
1) They need to already have started collecting data for this, assuming this challenge would have already started on January 1st. I don't know if this data can be collected from logs or if they have to have special software running in adcance.
2) Credit between projects needs to be carefully balanced. If there's a project getting abnormally low credit, people will flock to that project because they'll want to raise their lowest score. That may not be in PrimeGrid's best interest. Also, I think LLRs may intrinsically get less credit than sieves (to encourage people to sieve rather than find primes). Finally, as 'n' increases, LLR projects often hit their credit cap, which also throws the scoring off. If that's true, then this challenge could cause a shift towards LLR away from sieving, which, again, might not be in PrimeGrid's best interests. (On that note... somebody should take a look at the cap on Cullen LLR -- most of my WUs are maxed out on credit.)
The Tutti-Frutti challenge is an intriguing idea.
EDIT: Personally, I'd call it "Potpourri", but that's probably from watching too much "Jeopardy" as a kid.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
This has certainly made for some interesting and entertaining reading :)
Isn't the challenge supposed to be fun and educational? I fully understand the competitive side since of course we all want to do well. The GPU factor is new correct and now we all know what it can do. It almost sounds to me that the main argument here now revolves around "fairness".
Unfortunately in life, fairness is almost never a reality (paybacks are :) ). So, you can go buy more resources or as I did, join a team that is continuing to grow and as a team, make a splash in the challenges.
It will always be the big and haves vs the have nots. So for what makes up the challenges, let the admins create it and then YOU choose to participate or not.
After all, the main goal is to find primes even if the side benefits are credits and badges.
____________
@AggieThePew
 |
|
|
|
|
|
@Ken: Sorry my coding ignorance showed through a bit. I failed Algebra II, if that gives you any benchmark - I might know names of algorithms and coding approaches, but have no real idea of their significance. (hence, running but not developing for BOINC projects)
I was always under the impression, though, that LLR is not (as it stands now) suitable for application on a GPU because it's not an inherently parallel algorithm (that is, you're testing one huge number against a lot of other numbers). Someone, at some point, told me that porting an LLR to GPU might increase completion times by a few hours at most.
Believe me, I'd love to help if I could. Must be a tad annoying to ya, having people say what they want without understanding it at all.
That said, I like the idea of focusing on a few sub-projects rather than all at once. I mentioned, for instance, PSP LLR - clear out the work buffer (please). I've had tasks in validation for going on two weeks now - a challenge on that might be nice. It's the most credit a CPU can get short of SoB.
Which might also be a good thing, except that SoB WUs tend to error out and I don't think many people would want to "waste" their resources.
How about a "finish 'em off" challenge, where we devote a month to bumping the high shelf for PSP LLR, reaching optimal depth for 321, solving the Riesel Problem....I don't know, you get it, whichever projects are kind of stagnating or could do with a push towards the finish line. ;)
____________
|
|
|
|
|
This has certainly made for some interesting and entertaining reading :)
Isn't the challenge supposed to be fun and educational? I fully understand the competitive side since of course we all want to do well. The GPU factor is new correct and now we all know what it can do. It almost sounds to me that the main argument here now revolves around "fairness".
Unfortunately in life, fairness is almost never a reality (paybacks are :) ). So, you can go buy more resources or as I did, join a team that is continuing to grow and as a team, make a splash in the challenges.
It will always be the big and haves vs the have nots. So for what makes up the challenges, let the admins create it and then YOU choose to participate or not.
After all, the main goal is to find primes even if the side benefits are credits and badges.
What Rick said.
Aggie The Pew doesn't have a party line and I hope never to see it have one. If it does ever get one this, I hope, would be close to it.
I (and Aggie The Pew as currently constituted) will do relatively badly on GPU projects and relatively well on LLR projects. Great stuff; that means more people get a little time in the sun with a nice score at different times of the year.
|
|
|
|
|
|
We need a Fermat divisor challenge through the Proth prime search. Lower k values would be helpful... The lower the k value, the higher the probability of a prime actually dividing a Fermat number. IMHO, k values above 1,000 should not be given as much attention as lower k values.
GIMPS searches for factors of small Fermat numbers. Perhaps there could be a fun rivalry between GIMPS and Primegrid in terms of which project can find the most Fermat divisors each year! With this idea in mind, a PPS challenge would serve a fun purpose. Additionally, k=111 should receive ten times as much computer attention as k=1111, due to the probability of primes dividing Fermat numbers. Currently, it seems that k=9999 is weighted the same as considerably smaller k values. Could each k value have a weight in proportion to its probability of dividing a Fermat number? |
|
|
|
|
I dislike the idea of a challenge being on every single subproject because it would take forever for those of us that just have one or very few computers to run things on.
If you are including all sub projects, including SoB and the Cullen and Woodall LLRs, the 2011 tutti-frutti challenge probably should be for... 2011. All of it. The whole year.
It's not just that slow computers will take a wihle. Trying to run that many projects on even a fast multicore computer will be challenging because of the way BOINC refuses to schedule things the way you want it to. Getting what you want to run can be a micromanagment nightmare, and not all of us have the luxury of sitting in front of our computers 24/7.
So if you want to do something like this, it probably should be a parallel challenge that runs for the whole year.
This brings up two potential challenges for the admins:
1) They need to already have started collecting data for this, assuming this challenge would have already started on January 1st. I don't know if this data can be collected from logs or if they have to have special software running in adcance.
2) Credit between projects needs to be carefully balanced. If there's a project getting abnormally low credit, people will flock to that project because they'll want to raise their lowest score. That may not be in PrimeGrid's best interest. Also, I think LLRs may intrinsically get less credit than sieves (to encourage people to sieve rather than find primes). Finally, as 'n' increases, LLR projects often hit their credit cap, which also throws the scoring off. If that's true, then this challenge could cause a shift towards LLR away from sieving, which, again, might not be in PrimeGrid's best interests. (On that note... somebody should take a look at the cap on Cullen LLR -- most of my WUs are maxed out on credit.)
The Tutti-Frutti challenge is an intriguing idea.
EDIT: Personally, I'd call it "Potpourri", but that's probably from watching too much "Jeopardy" as a kid.
Perhaps just a pentathlon, heptathlon or decathlon where some points system is used to balance across CPU vs GPU, Sieve vs LLR, allowing for how many factor or primes found, etc. For example, wikipedia has the woman's point system that combines running, jumping and throwing events into a single number.
Probably hard to do but you could also go for the shortest time to complete a set number of tasks or points across a set number of subprojects. That means forcing a more hands on approach than the current 'setting and forgetting' style because you have to balance your approach.
Realistically I do not think there should be a SOB challenge unless credit is given for failed WUs. I have seen too many SOB WUs crash without credit due to some unknown error.
|
|
|
|
|
|
I'd like to see a semi-open ended challenge. Have it run until X number of primes are found. Award significant bonus points for finding a prime. |
|
|
|
|
|
SoB WUs don't crash because of "some unknown error," they crash because the slightest hardware issue (memory, CPU usage, etc) can cause them to miss a checkpoint, or something like that - I forget exactly what I was told regarding SoB errors, but it seems like the application is prone to errors on older or not-so-optimized hardware.
I'd imagine that people who do lots of SoB tasks have overclocked CPUs or memory (or lots of components) which can make things unstable.
I originally liked the idea for a SoB challenge but then realized pretty much exactly what you did, ConflictingEmotions, and given the amount of time and/or effort it takes to award manual credits, it doesn't seem worth it. SoB has its niche and whoever does those WUs is likely to continue doing so. They clearly don't mind the time it takes.
I wonder if there's some kind of trick involved in getting a SoB WU to run to completion in one go without running into errors. I can imagine what it's like to have something run for days, reach 100%, and only then say "Computation error" and give no credit, essentially wasting CPU time and lengthening the overall time til the task is validated. Eh. Never having tried SoB (yet) I can't really offer a definite opinion.
Perhaps Robert's idea of an "open-ended challenge" is best suited to PPS LLR. Since the sieve is running pretty much indefinitely at this point, I take it the LLR will be going for a while as well. Short WUs mean if everyone focuses on PPS LLR, there's a good chance we can advance the buffer a whole lot.
Perhaps a parallel challenge with the PPS LLR here and the PPS Extended LLR on PRPnet? From what I understand, if our sub-project here advances significantly, then the range we are testing will get closer and closer to what PRPnet is. I'm still not exactly an expert on how all the prime-searching parameters work, so do correct me if I'm wrong.
I guess what I'm trying to say is that we should have a challenge that, if participation is as high as it has been in the past, would really give a boost to whichever sub-project(s) on which it focuses.
Also, 321 sieve is ending soon, no? I don't remember which challenge is slated for February on PrimeGrid, and can't find a schedule anywhere...but if we want to end sub-projects, maybe a good ol' challenge is the way to do it. :)
____________
|
|
|
Michael Goetz Volunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14688
ID: 53948
Credit: 1,042,672,583
RAC: 11,648
|
I wonder if there's some kind of trick involved in getting a SoB WU to run to completion in one go without running into errors
You're only partially correct in your assessment that various hardware faults are the cause of the errors. That's certainly one possible cause. It's not the only likely cause, however.
Not everyone has the luxury of running BOINC on dedicated crunching servers; BOINC (and its predecessor, SETI@Home) were designed, after all, to run in the background on machines that were being used for other purposes.
That's really the key thing about a lot of 'computation errors'. If *something* running on a computer crashes Windows (or whatever OS is running), the side effects of that crash may corrupt the BOINC WUs that are in progress. Sometimes this won't hurt anything because the system crashes and then when it starts up, the WUs just restart from the latest checkpoint and everything is fine. But sometimes the crash isn't instantaneous, and the WU keeps running but is unable to use some system services such as accessing the file system or allocating memory. When this happens, the WU will error out and you've got your computation error.
So, sometimes it's hardware. Sometimes it's some other software (Windows and/or a driver or application) running on the computer. And sometimes it's the WU itself (although that does not appear to be the case here.)
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Ah, ok. That explains a lot. This means, hopefully, that running SoB on my Mac (which has 100% uptime) should be a pretty safe bet, as I keep the system very well-maintained. At least, I try my best.
Thanks! |
|
|
|
|
SoB WUs don't crash because of "some unknown error," they crash because the slightest hardware issue (memory, CPU usage, etc) can cause them to miss a checkpoint, or something like that
That, by definition, is a some unknown error. More informative error messages may help.
I forget exactly what I was told regarding SoB errors, but it seems like the application is prone to errors on older or not-so-optimized hardware. I'd imagine that people who do lots of SoB tasks have overclocked CPUs or memory (or lots of components) which can make things unstable.
I do not follow that line because it does crash with new systems and without overclocking.
Perhaps is it just the time required as the same app is about to handle other subprojects without the same crashes. |
|
|
|
|
SoB WUs don't crash because of "some unknown error," they crash because the slightest hardware issue (memory, CPU usage, etc) can cause them to miss a checkpoint, or something like that
That, by definition, is a some unknown error. More informative error messages may help.
I forget exactly what I was told regarding SoB errors, but it seems like the application is prone to errors on older or not-so-optimized hardware. I'd imagine that people who do lots of SoB tasks have overclocked CPUs or memory (or lots of components) which can make things unstable.
I do not follow that line because it does crash with new systems and without overclocking.
Perhaps is it just the time required as the same app is about to handle other subprojects without the same crashes.
Even new hardware may be marginal. The quality of consumer computer hardware is often more marginal than you might be aware of. There are a lot of stress test you can run to verify if you hardware is really stable.
A lot of consumer computer hardware is marginal because of everyone wanting to make (more) money. Marginal components are simply used to save money. In many cases this won't be noticeable until you really stress the hardware and very few programs do that.
Even if you buy components and assemble the computer yourself, you will still have to test it to be sure it's stable. The cheaper components you use, the more likely you are to get an unstable system.
/Snf
|
|
|
|
|
|
I'd kinda supposed that my computers fubared units once every x hours the computer was running and thus the probability of the unit failing to complete was a function and probably a fairly linear function of the processing time. So PPS units hardly ever crash because they are so short. SoB's are running so long there is more time for them to crash. It is just that it stings more when an SoB goes belly up.
Probably it is harder than that. |
|
|
|
|
|
I can state that SOBs complete without issue if you have stable machines.
I had 5 machines with all cores running SOB simultaneously and did not lose one of them.
2 Core 2 Duo (4 cores)
1 Core 2 Quad (4 cores)
2 I-7 8 core (one I only run 7 cores because I keep the other core for me) (15 cores)
It took between 12-18 days for the units. I have my SOB Gold now.
So, if you have good quality hardware, keep it clean and monitor it well, you can get good clean work out of them.
Now running SOB as a challenge may be a bit overboard as it does take quite some time to finish, I myself would love to see PPS keep in the rank of production top 5000 units so a long run of that would be nice to keep it on the map.
I know Sieves for 321 is getting close to ending, and CW is probably not far behind. I think we'll finish 321 without any challenge, but we could do some good cleanup of CW with a nice challenge.
____________
Lucky numbers 121*2^4553899-1 and 3756801695685*2^666669±1
My 12 min movie https://vimeo.com/manage/videos/502242 |
|
|
|
|
I can state that SOBs complete without issue if you have stable machines.
I had 5 machines with all cores running SOB simultaneously and did not lose one of them.
2 Core 2 Duo (4 cores)
1 Core 2 Quad (4 cores)
2 I-7 8 core (one I only run 7 cores because I keep the other core for me) (15 cores)
I guess that 119827519 was not one of them :-) Now at lucky number 17: 3 inconclusive results, 3 computing errors, 2 Abandoned, 3 time outs and 4 some other error. Yet some of these hosts are stable enough to complete other subproject WUs.
Anyhow, I have done 60 WUs mainly with Linux because I have had and seen too many failures with Windows (may be 64-bit related as I recall that most tend to be out of memory errors on vista/Win 7 hosts). So I do not buy into the argument that these errors are just about stability or overheating.
|
|
|
|
|
Anyhow, I have done 60 WUs mainly with Linux because I have had and seen too many failures with Windows (may be 64-bit related as I recall that most tend to be out of memory errors on vista/Win 7 hosts). So I do not buy into the argument that these errors are just about stability or overheating.
All the machine on mine are Win 7-64. The Core 2 each have 2G of memory, the Quad has 3G the I7s have a lot more.
So I ran 2 in 2G and 4 in 3G simultaneously and did not lost any. |
|
|
|
|
|
So, February 3 is coming up. Any news on what the first challenge will be or has that been posted and I missed it somewhere?
____________
@AggieThePew
|
|
|
pschoefer Volunteer developer
Volunteer tester
Send message
Joined: 20 Sep 05
Posts: 721
ID: 845
Credit: 4,005,928,549
RAC: 106,020
          
|
|
Yes, some more details about the first challenge would be nice, since it's only 13 days away. ;)
____________
 |
|
|
|
|
|
How about an IRON MAN contest sometime this year?
Need 3 projects-run 18hr each. consequetive order. Total 54 hours.
Beauty is rotating times across globe so each has to sacrifice sleep sometime
Any subvarinat in number of projects or individual time duration possible.
____________
 |
|
|
|
|
|
How about the 5 GLASS TRICK contest.
Need 5 projects with vastly different WU credits allotted. Length anywhere 3-7 days.
Run project times at user disretion.
Highest and lowest scores are tossed out (Only for contest, all credits earned go to
user's account. This prevnts "sandbagging" or skewing final outcome)
3 project scores left, which lowest of remaing 3 projects is your base score. (You have to effifciently manage these during contest.
Prevents runaway high value)
Highest Base score wins!
____________
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
|
You get to tricky with the Challenges and the Challenge will be over before most people even figure out what or how their supposed to be running it ... Especially people with fairly large Pharms, having to switch them 2-3-4 times during a Challenge will present problems for them ...
____________
|
|
|
|
|
|
Steve,
IRON MAN is basically 3 mini challenges in one. Not too much problem except rotating times. Makes it fair across ALL timezones.
5 GLASS TRCIK, I agree is more daunting but it makes the playng field
more level so to speak. This is supposed to a CHALLENGE that offers difficulty to everyone. Or should teams or individuals with the bigget farms win the challenge each time. Requires a little bit of strategy (read effort) mixed in rather than relying solely on the scope of our hardware. I don't see anywhere, where Challenges have to be easy.
Only my two cents for suggestion(s) asked for.
____________
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
I don't see anywhere, where Challenges have to be easy
Absolutely, make them as hard to run as possible to run, but I don't see anywhere where anybody has to run them either once they get beyond reasonable to run them.
____________
|
|
|
|
|
Steve,
IRON MAN is basically 3 mini challenges in one. Not too much problem except rotating times. Makes it fair across ALL timezones.
5 GLASS TRCIK, I agree is more daunting but it makes the playng field
more level so to speak. This is supposed to a CHALLENGE that offers difficulty to everyone. Or should teams or individuals with the bigget farms win the challenge each time. Requires a little bit of strategy (read effort) mixed in rather than relying solely on the scope of our hardware. I don't see anywhere, where Challenges have to be easy.
Only my two cents for suggestion(s) asked for.
Waiting for the validation of up to 5 LLR sub project challenges for a final score that will be rather random and totally intransparent during the competition? Great idea! That's a certain way to scare the majority of the participants away. In the end one of the farmers who runs the 5 subprojects on dozens or hundreds of boxes will win anyway before the people who run 5 subprojects on less boxes and winning such a challenge will need no more effort from him/her than to simply keep the boxes running. In the rear ranks things will be a matter of pure luck. This is no challenge: This is gambling.
Edit: Fixed a typo.
____________
|
|
|
mfbabb2 Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 21,825,457
RAC: 0
|
|
Weighting the score "per core" would go a long ways to eliminating the "unfair" advantage large pharms have ... of course then those of us with old(er) CPUs will still come in at the bottom on a per hour of calculation basis ... There is just no way to please everyone (or even make it fair for that matter)!
Suck it up and enjoy the Challenge (whatever it is)!
____________
Murphy (AtP)
|
|
|
|
|
|
Still think running a "Find a Prime or Bust" challenge would be fun.
Challenge runs until X number of primes are found. Considerable amount of bonus challenge points awarded to those that find a prime. |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1292
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
The Iron Man idea is interesting, but to be a true Iron Man the three projects would have to be very different. For example, one would have to be sieving and one would have to be PRP testing. I don't know about the third. Maybe the other would be PRP testing, but with different software, i.e. one BOINC and one PRPNet. |
|
|
|
|
|
And what about making two parallel challenge series (but not running at the same time)? For example one "classical challenge series" and another one "special challenge series". Many people prefer just a "brutal force challenge", where they can show the performance of their computers.
But there is another group for people. They like competitions, it is not just a challenge for their computers, but also for them ! They like risking and tactics, it´s a kind of game for them ;-).
It´s quite impossible to take some advantage in "the classic ones", because the winner is always known before the beginning of the challenge. It is not a challenge for the users, but just for their computers.
That´s why I´m talking about two parallel series - we can choose, which challenge will we participate in. The "classical", the "special" or both? It would be just the user´s choice.
It´s just my suggestion and my oppinion. Sorry for my bad english...
____________
 |
|
|
|
|
|
How about a coast to coast run? You would run each sub project in a specified order for a week before changing to the next one. And just for fun, you could do something like multiply total credit times the number of primes found for a challenge score. Of course this would depend on whether enough WUs could be provided for each sub project as needed. Then again, that could also be part of the challenge.
____________
 |
|
|
|
|
|
There's always the challenge that separates the big from the small. Let the Big Dogs haul ass with their big pharms in CUL LLR or WOO LLR and us little guys in SGS LLR.
Possibly awarding two sets of points.
____________
 |
|
|
|
|
|
The problem is: Now matter how the challenges are organized (in terms of rules, stats, points, etc.) there will be always a bunch of people complaining*. Making things more complicated than neccessary will lead to more accusations, more complaints and more discussions and will completely take the fun out of the challenges.
*The "funny thing" is: A long as their teams are on top of the rankings you don't read a single complaint from them in their team forums. "Loudmouthed" is the best description for their attitude in this case. As soon as they loose their place in the sun they change their behaviour and the whining starts. Sometimes the most interesting thing in a challenge is to read their hilarious accusations and conspiration theories...
____________
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
There's always the challenge that separates the big from the small. Let the Big Dogs haul ass with their big pharms in CUL LLR or WOO LLR and us little guys in SGS LLR.
Possibly awarding two sets of points.
What makes you think ? Big Dog's are any better Equipped to run 70 Hr's Wu's any more than Little Dog's are. Having more Computers doesn't mean you can run any of the Wu's faster just that you can run more of them. Big Dog's in most cases have the same type Computers that the Little Dogs have, in some/lots of case's even Inferior Computers to the Little Dog's because their Collage or Corporate Computers or a Person just can't Upgrade them as often as somebody with less Computers may be able to.
Also just what is your Cut off Point for Little Dog to Big Dog in Amounts of Computers. I see you have 4 Showing here the last 30 Day's at PrimeGrid but a Possible 89 listed, it's amusing you even consider yourself a Little Dog. To somebody with 1 or 2 or even 10 Computers I'm sure they would have a different opinion of just what kind of Dog you were, sorry no Offense meant there, just a little play on words ... ;)
____________
|
|
|
Vato Volunteer tester
Send message
Joined: 2 Feb 08
Posts: 989
ID: 18447
Credit: 2,380,616,839
RAC: 494,591
|
|
We should not be looking at Big vs. Small Dogs - too many problems in setting fair criteria and policing them.
Any challenge should be fully open to all (subject to them having CPU and/or GPU capable of running the WUs) just the same as outside the challenge.
____________
|
|
|
Scott Brown Volunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2542
ID: 1178
Credit: 30,633,505,607
RAC: 10,718,300
                                                       
|
|
Again (I think this has been said in the "suggestions" thread for the last two years), setting up challenge classes similar to what is done frequently in American road racing would solve much of these complaints. In other words, in addition to the overall challenge points as they are now, add a few classes of competition with their own sub-class challenge totals (e.g., participants with only 1 machine, with 2 to 5 machines, with 5+ machines--cutoffs for such challenge classes could be off course different/more numerous than my suggestions).
____________
141941*2^4299438-1 is prime!
 |
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
Again (I think this has been said in the "suggestions" thread for the last two years), setting up challenge classes similar to what is done frequently in American road racing would solve much of these complaints. In other words, in addition to the overall challenge points as they are now, add a few classes of competition with their own sub-class challenge totals (e.g., participants with only 1 machine, with 2 to 5 machines, with 5+ machines--cutoffs for such challenge classes could be off course different/more numerous than my suggestions).
I've often thought of that as being a good Idea myself Scott, having different Classes for maybe 1-5 6-10 11-15 etc ... But the problem with that is there could possibly be to many ways to fool the system on how many Box's you ran during the Challenge. I won't go into them so as not to put idea's in anybody's heads if they haven't already thought of ways around the system ...
I won the last Individual Challenge on the strength on my GPU's, I know how some people must have felt having to go up against somebody with the amount of GPU's that I have. But in the CPU Challenges I feel the same way having to go up against somebody like a shanky123 or Mr Hankey, Mumps & others with way more CPU Cores than I have. I do the best I can & try not to let it bother me to much, it's just the nature of the beast & we all have to live with it ...
____________
|
|
|
John Honorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Yes, some more details about the first challenge would be nice, since it's only 13 days away. ;)
3 Days (3 Feb - 6 Feb) 18:00 UTC PPS (LLR)
expect changes in other projects to accommodate this: longer WU's in PPS (Sieve) and CW (Sieve); shorter purge times
NOTE: this coincides with the upcoming Tour de Primes as well. Therefore, 5th of Feb. will be the mountain stage.
____________
|
|
|
|
|
|
Ok then, How about a ONE DOG race.
Each individual (and/or team member) select ONLY ONE of their fastest/best
computers to enter challenge. Can have as much bells and whistles that is installed
but only one machine. Prime grid would know by computer id to prevent multiple
"sleds" (ie computers). By the way, start and finish with one computer,
if it dies midway thru challenge, "YOUR OUT", no switching to a backup computer. It's
like one man/one dog iditarod race.
____________
|
|
|
|
|
3 Days (3 Feb - 6 Feb) 18:00 UTC PPS (LLR)
NOTE: this coincides with the upcoming Tour de Primes as well. Therefore, 5th of Feb. will be the mountain stage.
Uhmm hmm!
Goin' to be some vermin kickin' some ***es I think. Could be the year of the rat again.
We have had some really fun Tour de Primes previously and I bet this one will be great too so thanks for setting it up again
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
|
What is the Tour de Primes ??? Haven't seen anything about that yet or I missed it ???
____________
|
|
|
|
|
What is the Tour de Primes ??? Haven't seen anything about that yet or I missed it ???
A small prime finding competition...
Here is last years thread:
http://www.primegrid.com/forum_thread.php?id=1623&nowrap=true#20426
BTW: You might know this cruncher ;)
Prime Counts: 2 - PoorBoy
Prime Score: 21.962 - PoorBoy
I guess you found the primes during last years "Year of the Tiger" challenge...
____________
|
|
|
|
|
Ok then, How about a ONE DOG race.
Each individual (and/or team member) select ONLY ONE of their fastest/best
computers to enter challenge. Can have as much bells and whistles that is installed
but only one machine. Prime grid would know by computer id to prevent multiple
"sleds" (ie computers). By the way, start and finish with one computer,
if it dies midway thru challenge, "YOUR OUT", no switching to a backup computer. It's
like one man/one dog iditarod race.
won't work - there already are several hosts which are used to upload results chrunched offline on other machines.
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
Ok then, How about a ONE DOG race.
Each individual (and/or team member) select ONLY ONE of their fastest/best
computers to enter challenge. Can have as much bells and whistles that is installed
but only one machine. Prime grid would know by computer id to prevent multiple
"sleds" (ie computers). By the way, start and finish with one computer,
if it dies midway thru challenge, "YOUR OUT", no switching to a backup computer. It's
like one man/one dog iditarod race.
won't work - there already are several hosts which are used to upload results chrunched offline on other machines.
One of the reasons (several hosts which are used to upload results chrunched offline on other machines) I said to Scott there are ways to fool the Server, that and not all of us have a 16 or 24 or 32 or 48 Core Server running, most of us at best have a 6 Core i7 ...
____________
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
What is the Tour de Primes ??? Haven't seen anything about that yet or I missed it ???
A small prime finding competition...
Here is last years thread:
http://www.primegrid.com/forum_thread.php?id=1623&nowrap=true#20426
BTW: You might know this cruncher ;)
Prime Counts: 2 - PoorBoy
Prime Score: 21.962 - PoorBoy
I guess you found the primes during last years "Year of the Tiger" challenge...
Never heard of that Guy ... Lol'z
____________
|
|
|
Lumiukko Volunteer tester
Send message
Joined: 7 Jul 08
Posts: 169
ID: 25183
Credit: 1,315,276,948
RAC: 256,807
     
|
3 Days (3 Feb - 6 Feb) 18:00 UTC PPS (LLR)
The countdown clock is counting to 3 February 00:00 UTC?
--
Lumiukko |
|
|
|
|
|
No offense taken, Steve.
How about adding up the GHz of all CPU's used. Like I have 3 P4s at 2.6, 2.8 and 2.8. I also have a laptop with an M430 with 2 cores running at 2.27. This would all add up to 12.74 GHz.
Categories could be set like:
1-4.99GHz- Class 1
5- 9.99GHz- Class 2
10-14.99GHz- Class3
15- 19.99Ghz- Class 4
20- 29.99GHz- Class 5
30- 39.99GHz- Class 6
40- 49.99GHz- Class 7
50- 99.99GHz- Class 8
etc., etc., etc.
Complicated, yes.
____________
|
|
|
STE\/E Volunteer tester
Send message
Joined: 10 Aug 05
Posts: 578
ID: 103
Credit: 3,969,806,318
RAC: 1,225
|
|
You've done very well with what you have ... :)
____________
|
|
|
|
|
|
I used to have a 24 machine pharm, then I retired and couldn't afford a $500 a month electric bill.
____________
|
|
|
|
|
Ok then, How about a ONE DOG race.
Each individual (and/or team member) select ONLY ONE of their fastest/best
computers to enter challenge. Can have as much bells and whistles that is installed
but only one machine. Prime grid would know by computer id to prevent multiple
"sleds" (ie computers). By the way, start and finish with one computer,
if it dies midway thru challenge, "YOUR OUT", no switching to a backup computer. It's
like one man/one dog iditarod race.
won't work - there already are several hosts which are used to upload results chrunched offline on other machines.
So I'm not the only one being crazy enough to do this? ;-)
____________
There are only 10 kinds of people - those who understand binary and those who don't
|
|
|
|
|
|
When we can expect a full set of races of 2011?
____________
Polish National Team |
|
|
|
|
|
I hope we can see it soon...
I´m planning my holidays & events for this year just now, an I don´t want to miss many challenges (like the last one - I´ve been on skiing holiday).
____________
|
|
|
|
|
I hope we can see it soon...
I´m planning my holidays & events for this year just now, an I don´t want to miss many challenges (like the last one - I´ve been on skiing holiday).
Not to be to sarcastic but, missing a challenge to go skiing is to me a no brainer regardless of the challenge :) However if I had my wish, I'd be headed back to Hawaii and out of the cold.
Rick
____________
@AggieThePew
|
|
|
|
|
|
Ok, bad example, nonsense, forget it... :D
Just going around hot pot - I am just curious, when the challenges are going on, like many people there :-)
____________
|
|
|
|
|
|
Any news?
____________
|
|
|
|
|
|
It would be very helpfull to announce all races in 2011. ;)
____________
Polish National Team |
|
|
|
|
|
Just noticed - in the Boinc Stats page under the project challenges section there's a whole list of 2011 Primegrid upcoming challenges. Guess the update on the PG main page is not ready yet but the schedule is out there now.
Was checking on the current challenge when I noticed the rest of 2011.
http://boincstats.com/stats/challenges.php
Rick
____________
@AggieThePew
|
|
|
mfbabb2 Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 21,825,457
RAC: 0
|
Just noticed - in the Boinc Stats page under the project challenges section there's a whole list of 2011 Primegrid upcoming challenges. Guess the update on the PG main page is not ready yet but the schedule is out there now.
Was checking on the current challenge when I noticed the rest of 2011.
http://boincstats.com/stats/challenges.php
Rick
See also http://www.primegrid.com/challenge/challenge.php
____________
Murphy (AtP)
|
|
|
|
|
|
there is a mistake in http://www.primegrid.com/challenge/challenge.php
the next challenge will be during 1-3 April |
|
|
|
|
|
There's a bug in the script that generates the challenge stats html files i.e. http://www.primegrid.com/challenge/2011_2/top_teams.html
<a href="file://localhost/team_display.php">
and
<a href="file://localhost/show_user.php">
____________
Join BOINC United now! |
|
|
Kevin Volunteer tester
Send message
Joined: 4 Aug 09
Posts: 61
ID: 44488
Credit: 5,675,896
RAC: 0
 
|
|
Will an ATI build of CW Sieve be released before the June Challenge? It would be nice to know the plan for ATI projects in the near future...
____________
May the Force be with you always.
 |
|
|
