Posts by Tern
|
1)
Message boards :
Number crunching :
GPU version requirements for the search up to 10^20
(Message 616)
Posted 30 Aug 2017 by Tern Post: I know this is a low priority - I just got my (modified) MacPro up on Amic, and CPU tasks are running now. It appears there is no Mac version of the GPU applications... :-/ (Yeah, I know, I may be the ONLY one on this project with a Mac with an AMD card in it...) |
|
2)
Message boards :
Number crunching :
MacOS 64-bit version
(Message 404)
Posted 3 Apr 2017 by Tern Post: Laptop (host 1055), Mac OS X 10.11.6, first task completed and validated. iMac (host 2650), Mac OS X 10.8.5, first task about 1/2 way through, no problems expected. Looking good! :-) Intel MacPro equiv tied up on other projects, but will run a bunch of Amic through it Real Soon Now. PPC MacPro, etc, eh, not expecting an application for them. :-/ Latest-OS Macs, sorry, can't help; have too many apps that break after El Capitan, waiting on updates to those before I upgrade OSes. |
|
3)
Message boards :
Number crunching :
MacOS 64-bit version
(Message 403)
Posted 3 Apr 2017 by Tern Post: First one started on older OS host (2650). No error in first minute or two, will report back! |
|
4)
Message boards :
Number crunching :
MacOS 64-bit version
(Message 397)
Posted 2 Apr 2017 by Tern Post: "Error while computing" on site, "Computation error" on host. Only difference on two hosts was speed - 15 seconds vs. 20 seconds to failure. Hosts 1055 and 2650. |
|
5)
Message boards :
Number crunching :
MacOS 64-bit version
(Message 396)
Posted 2 Apr 2017 by Tern Post: Woo hoo! Put two slower hosts on, we'll see how it works! :-) |
|
6)
Message boards :
News :
TEAM CHALLENGE
(Message 271)
Posted 1 Mar 2017 by Tern Post: Since I think my naive question started this... I've been crunching since before BOINC, active (very at times) on the boards off and on ever since. Supported a lot of projects, or what seems like a lot to me (25ish). However, other than a couple of PrimeGrid challenges, this is my first "Team Challenge" and I'd never heard of "bunkering" before. So maybe I can see both sides of the argument without having a preconceived notion. At first glance it seemed "unfair" that someone could do work "before the start time" and then get the credits after the start. Being a programmer, my first thought was "how do you stop that". My second thought, after some consideration, was "oh, $^!#! You DON'T stop that!" Unless you go the whole PrimeGrid routine with counting only tasks _issued_ after the start (talk about server load!) and then _returned_ (even though not validated... which you have to count 'later' - head slap) before the end. Maybe PrimeGrid can handle that load. I can't imagine that any smaller projects can. (I wouldn't even want to write the SQL necessary, even discounting the server load...) So then my thought turned to "what is fair?" - and the answer of course is that it is fair if everyone has to play by the same rules. So on at least some challenges, "bunkering" is allowed, everyone has the same opportunity to do work and hold it until the start time. It's sort of (vaguely) like running Bitcoin Utopia. You can get a ton of credits, but some call it "unfair". But they COULD do the same thing, so... (Of course, my personal choice is to limit my BU credits to 50% of my total credits. I have 3 ASICs - I could have a few billion BU credits if I wanted them, without impacting any other project. My choice not to, but also my choice to "allow" myself to get up to 50% in BU.) Fairness as always is in the eye of the beholder. But given my knowledge of server programming, etc., I just see no practical way to do anything other than what we are seeing here, so even if some call it "unfair", it's just the way it has to be. Shrug. |
|
7)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 251)
Posted 28 Feb 2017 by Tern Post: I am trying, so far I have just 20, so still some more work to do to get to 100. They are not consistent with credits - it's the "luck of the draw", I suppose. After 357,000 credits I have found 377 pairs... |
|
8)
Message boards :
News :
TEAM CHALLENGE
(Message 235)
Posted 26 Feb 2017 by Tern Post: Note that challenge begins at 00:00 UTC on 3/1 and ends at 00:00 UTC on 3/6. Still trying to get Team USA signed up... |
|
9)
Message boards :
Number crunching :
Auto-ban faulty hosts
(Message 230)
Posted 26 Feb 2017 by Tern Post: Will a notice be sent to banned hosts? Otherwise I can see a lot of hours spent scratching heads going "why am I not getting work?!?!?"... I assume this is in addition to the 'normal' BOINC rollback mechanism; a host returning garbage should be limited to 1 task/day pretty quickly even without the ban. lhcathome-dev had problems because of VirtualBox and their redundant work scheme, BOINC didn't know if a task was good or not, so tasks kept being sent and notices not sent out. They had to resort to manually sending private messages. And then emails when nobody read those. But they are a special (self-inflicted) case. |
|
10)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 219)
Posted 25 Feb 2017 by Tern Post: I'm going to differ on the color idea... Standard color badges for CREDITS, yes; for pairs found, why? The number is already in the badge, I kind of like the idea of one that says "this is project-specific" - and I like the color you picked. (The maroon "100" level.) "Standard" colors are probably the ones PrimeGrid uses, although many projects have slight variations. I do like the idea of more levels. The more the better - keeps interest up. |
|
11)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 194)
Posted 24 Feb 2017 by Tern Post: You can see your pogs badges here as well as which galaxies you've helped process etc. The site requires registration but tinfoil hats are easy to make ;) Good grief. I have 46 badges there. Just NO!!! |
|
12)
Message boards :
Number crunching :
GPU version
(Message 174)
Posted 22 Feb 2017 by Tern Post: Just looked to make sure I had "run beta apps" checked, but it's not an option yet... |
|
13)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 173)
Posted 22 Feb 2017 by Tern Post: The badge a day projects you mention, besides POGs, what other project does that? Pogs is the one I had in mind; since I don't "do" but 20 or so projects, and I see others that have done many, many more, I assumed there were others. Odd thing about Pogs, (or maybe it's just that I get my sig from Free-DC?) I've got a half-mil or so Pogs credits, but no badges... I don't even see any for me on THEIR site though, at least not at a glance... I've never worried about it, because if I DID have a bunch of those ugly monstrosities, I'd be asking how to get RID of them! I've seen sigs that have dozens of them. Your sig only shows one - THAT, I could live with. What that one MEANS (yours is "Sc" at the moment, Scandium) I have no clue. Even Statseb doesn't say, and I'm too lazy to hunt it down on Pogs site. But yeah, they have so many, seeing them just says "oh, he/she runs Pogs" but doesn't instantly tell me "how much". That's why I've gone to putting my only 'badge' emphasis on WuProp. I will never have 20 stars (or want them) but I know at a glance just what it takes to get the three I've got, and I know what colors I want those three to be next! I just fought like crazy to get mine green, red, and gold for Christmas. Missed by two months. Sob! (The hard one was the gold! I don't run many projects.) To at least make a token 'touch' of the topic of the thread, I think Ami is heading the right direction. :-) |
|
14)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 168)
Posted 20 Feb 2017 by Tern Post: Technically it's possible, but it will bring a huge load to the server Thresholds it is - no point in loading down the server. The drug discovery badges are nice looking, but I don't care for their thresholds; 10k 100k 1000k 50000k 250000k 1000000k 10000000k 100'000'000k That just doesn't make sense... 10/100/1000/10,000/100,000/1,000,000/10,000,000 etc would be better than seemingly random "steps" there in the middle, but of course it all depends on the difficulty. If, say, reaching 1000 is "easy" but reaching 10,000 is "hard" (harder than the "10x time" would imply), then you stick in a 5,000 level. But then for consistency you should have a 500 and a 50,000 and a 500,000... I'm new here and haven't done the research to know this difficulty - I've now found 49 in just a couple of non-dedicated host/days, which seems to have been very "easy". If that holds linearly, then you don't need as many steps, but you do need higher and higher levels... I guess the simple thing there is just to keep looking at your top producer and try to keep up! :-) Badges are only annoying if you are "stuck" at a certain level for a very long time. Balancing that against the "new badge every day" projects is the trick. |
|
15)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 161)
Posted 19 Feb 2017 by Tern Post: Maybe silly question: Don't know how hard it would be to have the badge just be "31AN" (I've discovered 31 right now) in a circle? That seems more meaningful on THIS project than credits would be; and would be an instant comparison, unlike "100", which would be shared by everybody with 100-999 pairs. I just have no idea how difficult it would be to update the badge every time someone discovered a new pair; obviously the user page updates, since mine is showing 31... or how crazy the sig sites would go with the badges changing that often... If you had to create a different jpeg file for every possible number of pairs "up front", that obviously wouldn't work, it'd have to be done on-the-fly. If it's simpler to do the credit badge, so be it. I'm not picky, just wanting something "different" that isn't obnoxious. I REALLY like the simplicity of the new TN-Grid badges, for example. PrimeGrid's are boring, but make sense for them because they have so many applications. Sigh. I'll do it... my puny little badge collection: (TN-Grid's is the one on the far right.) |
|
16)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 160)
Posted 19 Feb 2017 by Tern Post: What do you mean, badge whores? :) Pbbbth! 
|
|
17)
Message boards :
Number crunching :
GPU version
(Message 150)
Posted 17 Feb 2017 by Tern Post: Random thought on the GPU version; you probably do NOT want to have a "# of GPU tasks" choice like Einstein has, without doing a LOT of testing first. Much better to keep it to one task at a time and let people change it with config files if they want more. That's the opposite of the CPU case where you want them to be able to LIMIT your use easily. You probably need the whole GPU anyway for your app, since you need memory space. I have not looked at your code, but you say there's little/no floating point, so GPU should work great whether AMD or nVidia. Will you use OpenGL only or have a Cuda version? People wanting a lot of double-precision FP go for AMD, so lots of those cards on BOINC, even though (for most cases) the nVidia cards work better on most projects. Many people have problems with drivers and OpenGL, especially now that Win10 keeps "upgrading" perfectly good drivers with stuff that doesn't work. Up to you whether it's worth having two versions or not. Regardless, be sure to test well with plenty of different cards, on multiple OS versions... or use the "run beta apps" choice in prefs and let us test it for you. Speaking of which, where is the Mac version??? :-) We want it ALL! And NOW! We can wait till next week for the Sun/Sparc and Mac PPC releases (maybe) I guess! |
|
18)
Message boards :
Random stuff :
Credit system.
(Message 148)
Posted 17 Feb 2017 by Tern Post: One other suggestion - now that you are "live", make any changes gradually. If you do decide you need to give fewer (or more) credits, don't make big changes all at once. If, for example, you decided to go from 1800 to 1200, go to 1600 one day, wait a while (at least a few days), go to 1400... Just remember to tell us the "plan" here before you do it! |
|
19)
Message boards :
Bug tracker :
Forum issue
(Message 145)
Posted 17 Feb 2017 by Tern Post: I don't think I've ever had issues fixed quite this quickly on any other project before... Great work! :-) |
|
20)
Message boards :
Number crunching :
# cores limitation
(Message 143)
Posted 17 Feb 2017 by Tern Post: GREAT! I can get rid of that app_config file and not worry about it! I'd still put something on the signup or home page about being mt and that the user should set the number of CPUs in the preferences... |
|
21)
Message boards :
Number crunching :
# cores limitation
(Message 141)
Posted 17 Feb 2017 by Tern Post: Something to work on for later. Right now I'd worry about putting "This app uses ALL your CPU cores, it is multi-threaded. If you want to limit this app, you will need to create an app_config file. We are working on allowing this to be set in the project preferences." on the signup page! (Oh - once work fetch occurs, existing and new WUs use the 'new' plan class even though the existing WUs still SAY they have the old value...) And a trivial issue; I've got a bunch of WU's "in progress" that are "ghosts" - due to my messing with the app_config file and getting it wrong a couple of times. They aren't on my machine and will time out on deadline. Luckily your deadlines are reasonably short. |
|
22)
Message boards :
Bug tracker :
Forum issue
(Message 139)
Posted 17 Feb 2017 by Tern Post: "Bug tracker" page https://sech.me/boinc/Amicable/forum_forum.php?id=2 at top path goes 'back' to "questions and answers". "Random stuff" page and "News" page https://sech.me/boinc/Amicable/forum_forum.php?id=3 (and id=1) at top path goes 'back' (correctly) to "message boards". |
|
23)
Message boards :
Number crunching :
# cores limitation
(Message 137)
Posted 17 Feb 2017 by Tern Post: That SHOULD work - but only after the next work fetch. It won't affect tasks already downloaded to the client. Depending on how you set plan class in your server code... |
|
24)
Message boards :
Random stuff :
Are Badges On the Way?
(Message 135)
Posted 17 Feb 2017 by Tern Post: My opinion: Boinc badges: (the ribbons) - nobody cares. They come and go. Credit badges: Each user has one badge per application the project offers, starting at some minimum level. (In this case, right now, one app, so one badge.) The color of the badge changes with the number of credits. Be sure to leave room at the top for expansion, because some people will accrue more credits than you ever thought possible. For the "standard" colors used, see all the other projects, but I think PrimeGrid has the right idea. The colors should be different enough to be identifiable; on a couple of projects, I can't tell at a glance if that one is bronze or gold or some color they picked to give for ten bazillion credits. Note that PrimeGrid had to change the shape of the badge when they ran out of colors! Achievement badges: Nice idea, not sure about application. If someone finds 100 pairs today and gets a badge, is that badge replaced later when they find 1000, or do they get another badge? I personally would replace not add to - Asteroids "adds to" (on credits even!) and it is annoying. For a project like this though, an achievement badge really would mean something. I'd concentrate on this one first! (Although for the average user, the credit badge will be a lot more understandable.) There are some who just want a huge number of badges for a sig. I'd rather have meaningful badges. To me, my one single badge from WuProp (and obviously what colors it has) means more than getting a dozen from some other project just for showing up. |
|
25)
Message boards :
Random stuff :
Credit system.
(Message 134)
Posted 17 Feb 2017 by Tern Post: Credit is a never-ending discussion. Go back ten-twelve years and you'll see the same points raised as today. Cobblestones, "credit-new", all are an attempt to be 'fair'. Sergei has the "least objectionable" (there is no "best") method with fixed credit per WU. Whether the number chosen is "too high" or "too low" is not answerable. It just "is". Picking the number based on a reference machine at least gives it some validity, it's not just a made-up number. The trick is to make sure that the number per DAY the reference machine can generate is at least somewhat comparable to what that same machine can produce on other projects that are considered to be in the "reasonable" credit range. You don't want to be Sztaki or (as much as I love the project) Rosetta. You don't want to be BitCoin or Collatz. PrimeGrid is difficult to use because different subprojects there give vastly different credits. Personally, I'd use Einstein. They are quite a bit higher than Seti, but not so much so to be objectionable. Or Seti would work but then people would complain the credit is too low. Or maybe run both and set Amicable to equal the average of those two. Sigh. Bottom line is that it doesn't matter what you do, you cannot make everyone happy. I haven't been running here long enough to have an opinion yet. |
|
26)
Message boards :
Number crunching :
# cores limitation
(Message 131)
Posted 17 Feb 2017 by Tern Post:
Several things incorrect here... "set client to use" - not settable to a project, it uses most recently changed for computing preferences. "for all projects" - true for computing preferences, but not project preferences. "independently" - computing preferences are overridden (if desired) at each host. Only project preferences are relevant for the question at hand, though. lhcathome-dev does it. See here: <https://lhcathomedev.cern.ch/lhcathome-dev/prefs.php?subset=project> Einstein@home does it (for GPU utilization). See here: <https://einsteinathome.org/account/prefs/project> The number of CPUs used by mt tasks is rightfully part of the project preferences (plan class), not the BOINC preferences, so is project-specific. Multiple computers are handled through the venue (work/school/home/etc, an unlimited number of them if you do as PrimeGrid has). (And no, unfortunately, the hosts page would be very difficult to use for this, though it would be nice.) Again, the main thing is not 'how to deal with mt tasks after joining', but 'why did a new user not know they WERE mt tasks BEFORE joining' - that points out a lack of communication in the project description and signup process, and is the problem. Also, "Bug Tracker" is not a normal place to look for project information in the forum - 'Number Crunching' is the standard heading once a project is "production". Communication is something I harp on from projects, because it is key to success. You can have all the problems in the world and still get volunteers if you communicate clearly and frequently. (Einstein, Rosetta.) You can lose people very quickly even with no problems at all, if you never clearly communicate with them. (lhcathome-dev, Sztaki, too many others to remember.) At least you aren't using VirtualBox without telling new users about it... |
|
27)
Message boards :
Number crunching :
# cores limitation
(Message 126)
Posted 17 Feb 2017 by Tern Post: It would really be nice to be able to set this on the Preferences page - many people are NOT going to be able to edit app_config files and get it right... Also would be good to explain back on the "join" page that this project uses ALL your CPUs, because in spite of reading everything I could find about the project, and even some of the forum postings, I was surprised when "Use: 11C" showed up on 21 tasks! I can't dedicate the whole box to this project, sorry! Cut that to a more reasonable 5, one at a time... |
©2024 Sergei Chernykh