Message boards :
Number crunching :
Too much credit ?
Message board moderation
Author | Message |
---|---|
Send message Joined: 16 May 12 Posts: 7 Credit: 135,265,500 RAC: 23,246 |
Hello, I have been calculating Work Units for years for the NumberFields project. I've been using the same computer since the beginning with the same CPU. Until last week, my credit was about 20,000 points per day. Then, with the same hardware, the same CPU, my credit went up to 350,000 points per day last week ! Is this a mistake ? Otherwise, I look forward to testing my two RTX 2080 Ti that are in the same computer, but my OS is Windows ! ;-) Thank you. |
Send message Joined: 11 Oct 18 Posts: 12 Credit: 1,013,277 RAC: 0 |
Strange. I checked my account and started seeing WUs with 8000 points. And still over 20 WUs pending. But no new badge for scoring paver 200k points in total. Seems to be a serious bug. |
Send message Joined: 13 Mar 19 Posts: 10 Credit: 34,136,941 RAC: 10,612 |
The credit system was changed because the gpus got to much credits. Now we get the same credits for every wu. In my opinion this is fair. |
Send message Joined: 16 May 12 Posts: 7 Credit: 135,265,500 RAC: 23,246 |
The credit system was changed because the gpus got to much credits. Now we get the same credits for every wu. In my opinion this is fair. In my opinion, these new credits cannot be fair ! My CPU has a theoretical power of 9.5 billion ops/sec. This must represent about 20,000 points per day. Now, I get 350,000 points a day, almost 18 times too many ! The remuneration of WU CPUs therefore no longer makes any sense. This remuneration should realistically indicate, as far as possible, the number of elementary operations performed by the CPU in a given time. I don't exactly know what's wrong, but I really think there's a beug ! |
Send message Joined: 13 Mar 19 Posts: 10 Credit: 34,136,941 RAC: 10,612 |
Before the change my GPU (1080) got >6.000.000 Credits a day which was a little but to much, There was a discussion in this Thread: https://numberfields.asu.edu/NumberFields/forum_thread.php?id=362&postid=2269#2269 |
Send message Joined: 16 May 12 Posts: 7 Credit: 135,265,500 RAC: 23,246 |
OK, sorry, I didn't see that discussion. I still think that this remuneration for CPU is far too high and does not reflect the reality of the work provided by the CPU. I don't know if it would be possible to process WU CPUs independently of WU GPUs ? It's just an observation ! I would also like to thank and congratulate the people behind the NumberFields project for all their work. This is my #1 project in order of preference and I look forward to making my GPUs available to NumberFields as soon as possible for me (my OS is Windows) ! I think that this project has a very large scope for the deep knowledge of integers and that this research is fundamental and may even influence physics ! |
Send message Joined: 8 Jul 11 Posts: 1341 Credit: 509,724,636 RAC: 554,429 |
I agree, the credit is too high. As a temporary bandaid, I have reduced it. And I can further reduce it if need be. Credits are currently set to a fixed value per WU (based on average runtime). It looks like this was set too high. I am not into the credit thing, so I have not been paying attention to it and the message boards were quiet about it (until now). It seems like a fixed credit per WU is fair. If a GPU is 20x faster it will get 20x the credit per hour. From what people have said, it seems like other projects pay a disproportionately higher number of credits for GPUs. Is there a good reason for this or is it just to attract more users? |
Send message Joined: 28 Oct 11 Posts: 180 Credit: 246,434,109 RAC: 161,068 |
... and the message boards were quiet about it (until now).It's a generally observed bit of psychology. If credit is too low, everyone complains. If credit is too high, there's an unspoken (and instinctive) conspiracy of silence. The definition of credit is 'work done' - the number of floating point operations computed during the course of the work. So, equal credit per task for the same WU is right - until somebody thinks up a more efficient algorithm and computes the task with less work, at which point there's no fair answer. Incidentally, notice those weasel words 'floating point operations'. Would I be right in thinking that this project mostly utilises integer arithmetic? And would GPUs be especially efficient - i.e. 'fast' - when processing integers? BOINC makes no effort to assess the real speed of GPUs - unlike CPUs, which are benchmarked (badly) in both integer and floating point mode. Eric, would it be possible for you to make notes on your credit experiences during this transition? I really think that BOINC should - belatedly - address the real effectiveness of these various credit schemes (especially at times of transition), and your cool-headed observations would be most helpful. |
Send message Joined: 23 Sep 11 Posts: 1 Credit: 1,250,059 RAC: 513 |
Another point of view. I have been doing this for almost 20 years now and crunched on nearly 100 projects. After all this time I find the only way, in the long term, to reflect my effort is to collect credits..as I often said "I could give a ratzz azz about the science I'm in it for the credit..." not because I discount the scientific contribution but because experience has shown most projects have a short life and why bother with a low paying project?... There has certainly been an inflation of the credit scheme over the years and after the introduction of gridcoin it means pretty much nothing. I'll have 20 years of crunching in October and am seriously considering walking away as maybe the only benchmark that means anything anymore is years... |
Send message Joined: 8 Jul 11 Posts: 1341 Credit: 509,724,636 RAC: 554,429 |
Sure, I can send you my observations via PM if that works for you. Or would the BOINC projects mailing list be better? |
Send message Joined: 3 May 18 Posts: 18 Credit: 45,233,128 RAC: 1 |
Yeah I'm not a fan of Gridcoin either. I'd rather see them excluded from the general stats as an account manager rather than a team but I've already voiced my disdain about hat over the years. The stats have certainly changed over time. For example, I don't bother looking at my combined stats anymore because I choose not to crunch Collatz. Therefore I'll never have as high a rank as someone who does. I look at my stats based on the level playing filed of the project. Right now, if Eric keeps changing the credits awarded, those who have been able to crunch now while the big credits are offered will be off to a flying start, while the rest of us will be eating their dust. It's not something I'm losing sleep over but I'd like to see a decision made and stuck with to create an even playing field for all. Low paying projects definitely slip down my list of one's I choose to do. It's taken me a month of crunching to hit 1Mil in Cosmology @ Home but at least everyone is in the same boat. YAFU is another painful project but slightly worse because it's a multithreaded app. (so a 4t task chews up all 8 threads) I find Milkyway & Einstein pretty fair with credits but the main thing has been consistency. |
Send message Joined: 16 May 12 Posts: 7 Credit: 135,265,500 RAC: 23,246 |
800 points per WU, this is much more reasonable than 8000 and remains acceptable ! In my humble opinion, 500 would be much fairer... I'm not sure the number of GPU points will be very high with 800 or 500, but it will be more accurate. The same case occurs for most projects with GPU remuneration below 1,000,000 points per day, even with a 100% load on the "GPU Load": Einstein, AP27 or GFN (PrimeGrid), Milkyway... And for example Einstein remains very popular, despite this low remuneration compared to Collatz ! I remember very well : with a R9 290 X AMD GPU, during several years on Einstein, I made 60,000 points a day. Then it went up to 200,000 for a few years and then to 400,000. The reason is that the programmes have been improved, that is what I understood. Collatz remunerates almost twice too much ! But from what I understand, it is very convenient to program 3*x+1 on GPUs, which explains the very high efficiency of GPUs compared to CPUs. For other projects, this seems much more difficult. I am active within the Alliance Francophone and I know that for most of us, it is not only the number of points that motivates us, but also the subject of study of the project itself. For example, I have never crunched for the Bitcoin Utopia project (no scientific content !), but I crunched for Collatz, because I find it very interesting to observe the behavior of these digital sequences whose question nobody knows today how to approach in a theoretical way : we only have the computer to attack this question, so, in my opinion, it is worth it ! But some malicious people started cheating on this project because of GridCoin which I disapprove of ! For NumberFields, obviously, the remuneration will remain low and it is true that it will attract fewer people, that is for sure. I may be wrong, but the other way to attract people (people who are sensitive to the scientific interest of the projects) is to show how the project can advance many areas of science, even if only in the distant future. This is what I tried to do on a page in the Alliance Francophone forum where I rank NumberFields in the "Class 4" projects: https://forum.boinc-af.org/index.php/topic,7991.msg482370.html#msg482370 The problem with the NumberFields project is that the notions are extremely abstract for amateurs like most of us, including myself ! I had to spend several hours trying to understand what this project was working on. I don't think many people are willing to spend hours on it ! And in fact, I'm not sure I understood exactly, you can see below ! Who among us knows what the Langlands program is (https://numberfields.asu.edu/NumberFields/ProjectDescription.html) ? Humbly, I think that on this page (https://numberfields.asu.edu/NumberFields/), should appear a few lines explaining very simply for example something like that (it'is just an idea !) : "NumberFields is a project that should help to unify different fields of mathematics apparently unrelated. This will make it possible in a more or less distant future to progress much faster in many areas of mathematics. However, progress in mathematics also means progress in physics, computer science and biology. The potential indirect applications of NumberFields are countless. It is a fundamental research project that could even contribute to the creation of a new physics, beyond Einstein's relativity and quantum mechanics. However, everyone knows what applications have resulted from these two physical theories born at the beginning of the 20th century : understanding black holes, cosmology, mastering the atom, inventing the computer and all the applications allowed by this new machine (genetics, biology, astronomy, mathematics...) In short: NumberFields is a project that aims no less than that: to better understand mathematics and the relationships between its different fields to create a new physics: that of the 21st century !". In my opinion, NumberFields is not an ordinary project and if everyone understands that, there should be more people crunching on it ! I hope I myself have understood the scope of NumberFields' research object, because I am only a non-professional scientist ! If I say anything stupid in my bold italic paragraph above, please let me know ! ;-) Humbly, I think it is essential that mathematicians do the same work as astronomers to explain in terms understandable to all audiences the subject of their researchs. Then people will be passionate about it, like for the Einstein project, which talks about "pulsars" and "black holes", fascinating physical objects ! Mathematical abstractions are just as fascinating, if not more so, but it is extremely difficult to explain to the public ! I just have one more question about WU CPUs and WU GPUs : are they exactly the same or are CPUs calculating WUs that are impossible to calculate on GPUs ? |
Send message Joined: 8 Jul 11 Posts: 1341 Credit: 509,724,636 RAC: 554,429 |
Jean-Luc: Thank you for the well thought out and detailed post. Your understanding of the project is quite good for an "amateur". I should hire you to handle our social media and marketing! But yes, I should do a better job of explaining the fundamentals behind this project, something which I find very hard to do. I will rework the home page and add a few words as you suggest. Maybe I could also start a thread on the message boards where people could ask questions. Anyways, this will take a few weeks to get to, as I am focusing right now on getting the rest of the gpu app versions done. But please remind me again in a few weeks, in case I forget to come back too it. To answer your other question, the WUs are exactly the same for both gpu and cpu. |
Send message Joined: 16 May 12 Posts: 7 Credit: 135,265,500 RAC: 23,246 |
Eric, Thanks for your answer. I'll be there in a few weeks to remind you that there are things to change on the first page ! But yes, it is more urgent to finalize the rest of the GPU app versions. I look forward to testing... |
Send message Joined: 28 Oct 11 Posts: 180 Credit: 246,434,109 RAC: 161,068 |
Sure, I can send you my observations via PM if that works for you. Or would the BOINC projects mailing list be better?Just seen that you've found the 'outlier' setting :-) CreditNew doesn't kick in until host_app_version.pfc_avg and app_version.pfc_avg have usable values, so you'll probably have to unleash the outliers for a few days to fill up the tables. But there are still genuine outliers in the data, so we'll have to - eventually - find a way of distinguishing them: not by an absolute time value, but something relative. You'd probably fell able to speak more freely in PM, but I think we need to get this on the record sometime, so I think the time has come for a formal report to the projects mailing list. In the meantime - being Windows only - my GPUs haven't joined the party, but my CPUs are plodding along exactly as normal. I request that the evidence in https://boincstats.com/en/stats/122/user/detail/1969/charts be taken into account. |
Send message Joined: 23 Nov 17 Posts: 1 Credit: 33,246,284 RAC: 0 |
I think the current adjustment in credit is fair, and within the range of other projects I support. Not the best, and not the worst, and to me, still worth the resources I contribute. On that point, I’ve read other discussions about the purity of science and the view that somehow GridCoin, or other reward mechanisms, sully the aims of science. Balderdash! I’ve been a supporter of Seti since 1999, well before the advent of crypto rewards. Further, I think it’s fair to say that science, and financial, or other material backing, has a long (if less than pure) history. Supporting projects and furthering science is one of my aims. If I can be rewarded something of value in return, then what’s the harm of that? That value comes in the form of satisfaction for contributing to research, as well as a crypto that could very well end up being worth nothing. Either way, I am spending cold hard cash as it were, on electricity bills, and have been, going back some 20 years now. I suppose I could have just opted to buy crypto outright, and with my electricity bills (in the range of $200-$300 a month), owing in part to a small contingent of computers (9 at present), I’d have considerably more currency than I do now. But then, it’s my choice. And more research is done. Again, where’s the harm in that? Suffice to say that without the financial incentive of crypto, I’d be contributing less compute power to projects than I do now. Some, but not at the level I do now. It might not even be rational to contribute as much as I do instead of investing or saving that money, and yet I do it anyway. I trust that doesn’t make me a horrible person, nor GridCoin and its ilk, into some sort of pact with the devil. It’s a choice, and I’ll not cast aspersions on those who do it for strictly altruistic, or love of the science, reasons, or on those who hope there is a financial reward, or something in between. I do agree that a greater exposition on the science of the project would be beneficial, so that potential contributors, and non-mathematicians like myself, gain greater insight and understanding. Mind you, not having that won’t stop me from contributing at this point, since I also value the passion and effort that went into this project, and others like it, and that is worth supporting, as, and while I am able. Take care, be well, and happy computing. |
Send message Joined: 8 Jul 11 Posts: 1341 Credit: 509,724,636 RAC: 554,429 |
How "formal" are we talking here? I hope you don't want a powerpoint, I deal enough with that at work :-) Anyways, I'll throw something together in the next couple days. |
Send message Joined: 31 Oct 12 Posts: 14 Credit: 30,615,327 RAC: 0 |
Also posted here https://numberfields.asu.edu/NumberFields/forum_thread.php?id=362&postid=2351#2351 In my opinion, the change to 800 credits per task on a GPU is far too low. Here's a simple number Eric. In all the years of running numberfields on a CPU, I found 7 decic's. Since the GPU app became available, I am now up to 18 in about a week of running on only 2 cards. I'm sure this will be replicated by everyone else running GPU's here. I believe you are quoting that a GPU does 20 x the work of a CPU. You need to encourage GPU's not make the project not worthwhile crunching compared to other GPU projects. Apparently you had to drop the GPU credit because it was making the CPU credit too high. Why can't the award of points be based on whether the task was run on a CPU or a GPU ? Other project's manages that task. The increased numbers of decic's found by just myself should indicate how much more work is being achieved by people running GPU's and should encourage you to award and attract more crunchers to use that platform. On most projects, the introduction of GPU apps makes the CPU app almost obsolete. |
Send message Joined: 28 Oct 11 Posts: 180 Credit: 246,434,109 RAC: 161,068 |
I disagree. Payment should be for the work done in searching - the same number of credits for the search task, whatever the device used. GPUs will win out vastly in the number of credits awarded per unit of time - per second, per hour, per day, however you choose to measure it. Your own statement that you've found more candidate values since the GPU app was released confirms that you're conducting more searches. Good on you - that's your reward. You don't need to be compensated twice - once for doing more searches, and again for doing searches on a different device. |
Send message Joined: 31 Oct 12 Posts: 14 Credit: 30,615,327 RAC: 0 |
It's not compensation for finding candidates. Points are an incentive to conduct the searches in the first place, at least for me. Incentivise GPU crunching and you get more people finding more candidates far quicker than using CPU's. Set the points too low and I use my GPU's elsewhere. Mercenary as that may be, points awarded are my incentive to provide my equipment and burn my electricity for projects. Others may have different motives which I can respect. This is simply my outlook. |