Message boards :
News :
Running Modfit on MilkyWay@home
Message board moderation
Author | Message |
---|---|
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
Hello Crunchers, Now that MilkyWay@home Separation Modfit has been running for a year as a stable application, we are going to be releasing it as the official MilkyWay@home application under our main project. This means we will slowly (over the next month) be phasing out the dedicated Modfit project as it will be redundant. Additionally, this means some extremely old systems may no longer be supported on the MilkyWay@home project. These include Windows 32-bit systems, and very old (CAL) GPUs. If you are noticing compatibility issues with your system after we begin the change next week please post them here. Thank you all for your continued support. Jake [edit] Modfit is now up and running on the main application. If you are seeing any issues please let me know here. Thanks. For instructions on compiling 32-bit MilkyWay@home, please visit http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=3957. [/edit] |
Send message Joined: 13 Feb 09 Posts: 51 Credit: 72,959,707 RAC: 4,338 |
Additionally, this means some extremely old systems may no longer be supported on the MilkyWay@home project. These include Windows 32-bit systems I guess that means that I can kiss this project goodbye! :-( |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
w1hue, We are sorry for the loss. Maintaining support for 32-bit windows was becoming more and more difficult. (Especially since we don't currently have any Windows XP machines around to use for testing and testing GPU applications on virtual machines is almost impossible.) If you do still want to crunch for the project in the future, we do have 32-bit linux support. I understand asking for you to change you operating system may be a stretch, but it is the best solution we can offer at this time. Thank you for seven years of support. Jake |
Send message Joined: 9 Jul 08 Posts: 85 Credit: 44,842,651 RAC: 0 |
Host 691866 is a 64 bit Ubuntu 15.10 linux client running boinc 7.6.6 and every single new task is terminating with an error. It was previously running thousands just fine. Edit: The gpu tasks (opencl_amd_ati) are generating the errors. Not sure about the cpu tasks yet. Second edit: Looked back in the logs and yes, CPU tasks of the same type are all terminating with errors as well. |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
Hey Thunder, Looks like it didn't recognize my deprecation of the old applications. I gave everything a quick reboot. That usually fixes things. I will keep an eye on everything for a little while though. Jake |
Send message Joined: 9 Jul 08 Posts: 85 Credit: 44,842,651 RAC: 0 |
First thing it did on the next update was download a new application. :-) This machine is doing a few projects, but MW@H should come up in the rotation shortly. I'll let you know if all is well. |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
Thank you for the feed back. Jake |
Send message Joined: 9 Jul 08 Posts: 85 Credit: 44,842,651 RAC: 0 |
Yep, that fixed it. Thank you, Jake. Now, considering doing 2 tasks at once (with another project also running on the GPU at the same time) means each is about 14 seconds of work, I still wish I could get more of them. :-) |
Send message Joined: 11 Oct 09 Posts: 19 Credit: 202,475,569 RAC: 0 |
I only selected the MW app but now I also get modfit units... |
Send message Joined: 9 May 13 Posts: 10 Credit: 95,834,833 RAC: 0 |
Hello, i think something bad is happening on this project, I am able to do over 1.4 Mil /day and seeing how will be the evolution i prefer to move my power away. All the wus in all the hosts this evening suddendly gone bad with errors; some hosts with 2 280x doesn't receive any more any wuw admins, if something is working good and well, why would you like to modify it in such a terrible way? if your target /purpose is to let us to run away, just say it, express yourself cleraly, tnx between all, i have an host win 7 x86 with 2 280x crunching as a monster reading this "Additionally, this means some extremely old systems may no longer be supported on the MilkyWay@home project. These include Windows 32-bit systems" from this evening i think x86 systems are excluded from this prj; so bad news... now i understand why can't get any wus.... please let x86 system crunch some more time... |
Send message Joined: 13 Feb 09 Posts: 51 Credit: 72,959,707 RAC: 4,338 |
My XP machine has kissed the project goodbye, but I have an AMD64 machine running Win10 with a GT-730 still getting tasks. But -- I have it setup to run muntiple WUs (2, actually) at a time and I need to get the new executable as all milkey@home WUs are ending in error now. Modified fit tasks running OK -- can I use the same executable that is running those? |
Send message Joined: 8 Apr 09 Posts: 70 Credit: 11,027,414,782 RAC: 14,934 |
Just to clarify, what is the name of the new task, in the Boinc Projects directory? I am getting computation errors since yesterday, but i think it has to do with the PrimeGrid apps i am running since then. |
Send message Joined: 2 Apr 11 Posts: 14 Credit: 4,527,461 RAC: 0 |
My PPC Mac is now being refused tasks and the PPC app's no longer mentioned on the site's Applications page. My Intel Mac is receiving the new tasks even though my MilkyWay@Home preferences are set not to be sent Modified Fit or N-body tasks under any circumstances. The three results which have been reported so far have received inconclusive validations. My Tasks page shows that my PPC Mac was also sent one of the new tasks yesterday but errored out on it immediately. A Linux machine had previously errored out on the same task and another received a validate error. NG |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
I am going to try to answer all questions here in this post. Scan through until you see your name and answer. Nigel, Power PC Macs will also no longer be supported under the new program. This is a similar position with the Windows 32 bit application where we simply do not have the ability to develop and test for these systems anymore. Sebastian* The new work unit names will be something like: de_modfit_fast_14_3s_136_ModfitConstraints1_1 de_modfit_fast_14_3s_136_ModfitConstraints1_2 de_modfit_fast_14_3s_136_ModfitConstraints1_3 de_modfit_fast_14_3s_136_ModfitConstraints1_4 w1hue, You can always compile your own 32-bit Windows application. I actually forgot to mention this in the news post so I will add this as an edit there too. I will try to write up instructions on how to do this on the Number Crunching forums a little later. Also the Modfit and MilkyWay@home applications are now the same so yes you can use the same executable. BlindaTRON, You want to know:
Well first of all think about your question for a minute. Of course we don't want our volunteers running away from our project! We love you guys (well most of you, rarely one of you can be pretty mean) and without you all, this project could not function. Now to answer the question of why change the MilkyWay@home application that was working so well. The original MilkyWay@home application was at the limit of what it could do scientifically. Over the past 4 years since it was first released, the scientists behind the project have developed new mapping algorithms that are superior to those originally used, now known as Modfit. So to continue making scientific progress with the project, it was deemed time by myself and the other project scientists that we move forward with making Modfit the officially supported application. (For the record the other option was to simple let you guys continue crunching useless work units using the old application where there would be no scientific gain, but you would still be getting credit. In my mind this was a much worse option.) I will also point you to the fact that since we are an entirely open source project you are more than welcome to compile your own version of our application to run on your 32-bit systems. This is actually not a terribly difficult process if you have a 32-bit computer to do it on. I will post a list of instructions that can help you do this in the number crunching forums today or tomorrow. (Disclaimer: I do not currently have a 32 bit windows system to test these instructions on so that may not work the first time around but I'm optimistic with some of the users help and technical know how we will come up with a good set of instructions soon.) Herge, Yes, as of yesterday Modfit is now being run on the official MilkyWay@home application as it is now the officially supported application. Only expect to see Modfit work units going forward. Everyone, Thank you all for your patience and understanding as we migrate over to Modfit for the whole project. If you have any more questions keep posting them here and I will try my best to answer them. Jake |
Send message Joined: 2 Apr 11 Posts: 14 Credit: 4,527,461 RAC: 0 |
I wrote: My Intel Mac is receiving the new tasks even though my MilkyWay@Home preferences are set not to be sent Modified Fit or N-body tasks under any circumstances. The three results which have been reported so far have received inconclusive validations. One of the reasons I originally set my preferences against these tasks — apart from their higher-than-acceptable tendency at the time not to get validated — was the miserly credit awarded when they were validated. It's too early to say anything about current validation rates, but a couple of the above-mentioned tasks have now validated and the miserliness is even more extreme than I remembered! The old MW@H tasks took around 4,800 seconds of CPU time on my Intel Mac and earned 106.88 points apiece. These new ones are taking nearly twice as long and earning only a quarter of the credit. While I don't support BOINC projects for the credit, this does feel a bit like the proverbial slap round the face with a wet kipper. NG |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
Hey Nigel, Okay these work units should run much faster than the old work units because they are using far fewer integral steps than the previous version of MilkyWay@home. This should equate to a 4 times speed up. I'm not sure why your taking much longer to run work units from Modfit. The credit calculation should be about correct (I can tweak it a bit in the future as we see how long these runs are taking comparatively to the old runs). Jake |
Send message Joined: 27 Jun 09 Posts: 12 Credit: 148,038,330 RAC: 0 |
tasks are too short. ATI sleeps more like running. Please... longer tasks or greater job limit 1 WU ->6 sec |
Send message Joined: 25 Feb 13 Posts: 580 Credit: 94,200,158 RAC: 0 |
Peciak, The tasks can not be made any longer at this time without adding essentially useless calculations to the workunits. (They would run longer but not change the science.) To get around this problem we suggest configuring BOINC to allow you to run multiple GPU work units at a time. Also you are allowed to download something like 150 work units every minute. Not sure how you are running out of work to do. Maybe check your BOINC network settings to see if you can ask for work more often. Jake |
Send message Joined: 26 Apr 08 Posts: 87 Credit: 64,801,496 RAC: 0 |
I'm running into the same problem as Nigel Garvey - tasks take twice as long and the credit earned is a fourth of what it used to be. I don't know if it is some sort of artifact of running Macs or something else. I have core 2 duos, and i5 computers running OS X from Snow Leopard to El Capitan. I am curious as to what is going on. Plus SETI Classic = 21,082 WUs |
Send message Joined: 27 Jun 09 Posts: 12 Credit: 148,038,330 RAC: 0 |
Peciak, but 40 WU per host |
©2025 Astroinformatics Group