Message boards :
News :
NWChem long no longer in beta soon
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
Send message Joined: 23 Jul 19 Posts: 289 Credit: 464,119,561 RAC: 0 |
Hello, you're jumping too quickly to the conclusion that suits you. Our Windows WU's require acceleration for obvious performance reasons, but it is possible to install a Linux VM without acceleration and get linux WU's . In this case, the efficiency is likely to be quite bad. In any case, it's not a diagnosis of a problem with our Windows WU, it's expected. So, it's not the WUs that have a problem, but the configuration of your machine (hardware and software versions) that doesn't work properly with the tools provided by BOINC developpers. And yes, you're not the only one with this problem, and I'm sorry for that. I specify again that I did not modify the virtualbox wrapper provided by Berkeley. Probably, that the most stable projects have invested time to modify the official version, but I'm alone and I don't have that time. Our application is developed for Linux. The Windows and Mac applications we provide are a bonus. If that doesn't work for you, I'm sorry. |
Send message Joined: 12 Jan 20 Posts: 12 Credit: 220,914 RAC: 0 |
Please see your BIOS setting "Execute Disable Bit (XD)", try switch it to "disabled" or see more here |
Send message Joined: 16 Dec 19 Posts: 25 Credit: 11,938,843 RAC: 0 |
NWChem long queue is empty, no units to download. Hmm.. loved me those long work units...hahaha... Great to see those running on my computer. Some using 8 cores per work unit. Was nice to see that. Look forward to any new ones when they come out. Meanwhile, back to the short work units....Crunch crunch crunch........ Z |
Send message Joined: 12 Jan 20 Posts: 12 Credit: 220,914 RAC: 0 |
Virtualization is enabled but not available. Intel has included this limitation for the K-series processors to motivate people buy Xeons.
On the virtualized guest machine you have virtualized BIOS, not the same as the host machine's BIOS. So, see the BIOS setting i posted before. P.S. your computers are hidden |
Send message Joined: 13 Oct 19 Posts: 87 Credit: 6,026,455 RAC: 0 |
My little Windows machine happily crunches the short units without any issues we haven't discussed many times before. I was hoping the long ones might boost my RAC a little but they don't so I am happy taking whatever I get. |
Send message Joined: 13 Nov 19 Posts: 21 Credit: 2,596,565 RAC: 0 |
Hello Damotbe, OK . Boinc is also wrong . It is what you wrote ! Nice !!! I suggest you to contact Pr ANDERSON David. (See on SETI, the founder of BOINC, past century) All virtualisation work perfect onh all othe Boinc project.? But Berkeley is false, I am false, everyone is false. Only you are right. Then if Windows WU are "bonus", ok so far, It is your full right. Wanless also runs only under Linux. I think , you have never look my returned WU. Too easy for you to say " is not me , but your config". The returned WU by host 1764 (full manually real virtual machine) on host 956, have all capacities. So my config hardware is full correct. I repeat once again. It is the same PHYSICAL computer. Running basically Win7-Ultimate-x64 see description. Then, what you write about Linux "able" without acceleration, once again ??? Look real host and the sub host(virtual host) . Look the 13 WU (the latest on host 956, all returned on 03/03/20 on 20:10UTC ) Win7 Look WU 1077208, 1621082, and the others returned at same time. I repeat once again same host, hardware, as 1764. I7-2600K And look WU 1712747, 12748 (validated) and 1712792, 1712970 (waiting) I8t is the same physical computer. same processor, same Bios, same RAM, same HDD, same... Under Win7 your WU are not suitable. Not my host. Your WU. Instead of pblishing that your PRJ will not more long will be in beta phase, correct the problems. Or remove all your Windows. I have an other host (957) , also a i7-2600K. Perfect the same configuration as host (956) . Only GPU is not the same. The host 957 worked fine till 25 feb ! And also host 1174 (i7-3820K), same config Win7 as all others Look the WU. Take time, please. All report "virtualisation not capable". Nice, But it works on all other projects. So suddenly, all the windows config became wrong ??? All around the world !!! It is why i take the conclusion the problem is on your WU, not Boinc, not all the Windows World" I have a other host with a Ryzen7-3900X with 32Gb (Win10), later I will test your WU. I hope your wu will not report the same error "no virtualisation" By the way you not answer about the question "how to enable" virtualisation on Win7. Not forget, once again it works on all other PRJ. But you say it is the fault ofr BOINC and the users who are unablme to configure correctly computer ! We are all unable to handle and manage Windows. I am not the most capable, very far. But on Boinc from the first hours (past century) Look my ID on Boinc, look "the team" I am founder. 111th since creation ofr BOINC.I think I have enough knownledge to configure hosts . My best was to install à 6Gb HDD on a computer x86 under Vista. Now it turn (same host AMD 6300+ (dualcore) , 2Gb DDR2 (yes DDR twoo), with à RAID5 of four times two trra SSD !!! Impossible under 32 bits system !? OK, I know it is possible. This, to show Damotbe, that the problem is on your Windows WU. Not at our side. And not on BOINC Un peu d humlilité, n a jamais tué aucun Français !!! Hello, |
Send message Joined: 23 Jul 19 Posts: 289 Credit: 464,119,561 RAC: 0 |
Sorry, my interlocutor does not seem to understand the details of what I am explaining in English. I will answer in French. Cher Marsinph, Je suis enseignant-chercheur et actuellement la seule personne à m'occuper du serveur Quchempedia. L'Université française et particulièrement les petites universités ont de très grosses charges de travail. Personnellement, je suis entre 150% et 180% de charge de travail juste pour la partie enseignement. Un très bon stagiaire a développé la première version de Quchempedia@home et j'ai du faire face à de nombreux problèmes après son départ, notamment pour stabiliser la version Linux qui est aussi utilisée dans les VM. Avec le peu de temps qui me reste, j'essaye de mener des recherches, ce qui ne consiste pas à faire tourner et administrer un serveur BOINC qui n'est qu'une très faible partie de mon activité. Un peu d humlilité, n a jamais tué aucun Français !!! La méthode scientifique ne s'encombre pas de ses considérations ! Il y a des faits et des hypothèses compatibles ou incompatibles avec ces faits, point. Ton hypothèse est incompatible avec les faits. L'humilité, mon égo, ton égo n'a rien à voir la dedans. C'est un problème de ta méthode d'investigation : https://fr.wikipedia.org/wiki/Biais_de_confirmation . Les chercheurs ont l'habitude que leur travail soit critiqué, c'est même un point important de la méthode. Donc voici l'ensemble de faits dont j'ai la connaissance afin de réfuter ou non des hypothèses : - Il y a de nombreux hôtes pour lesquels les WU Windows fonctionnement très bien. - j'ai déjà observé tes WU ce qui m'a permis de voir que :
- ta machine virtuelle installée à la main et qui ne recquiert pas l'accélération de la virtualisation, accepte les WU Linux - j'ai depuis longtemps observé les WU d'autres utilisateurs avec des configurations similaires (Intel K-series) et tu n'es pas le seul avec ce problème
2/ Ta configuration (notamment ton modèle de CPU) pose problème au wrapper officiel vbox BOINC et au client de calcul. L'un croit que tu as l'accélération de virtualisation (client BOINC) et l'autre n'arrive pas à activer l'accélération (Vbox), ce qui provoque la récupération de WU alors que ta machine ne devrait pas en avoir. Problème qui disparaît en installant à la main une VM Linux qui n'utilisera pas le wrapper vbox BOINC.
|
Send message Joined: 3 Oct 19 Posts: 153 Credit: 32,412,973 RAC: 0 |
Sorry, my interlocutor does not seem to understand the details of what I am explaining in English. I will answer in French. By the way, your English is very, very good. You have studied or worked over here somewhere. |
Send message Joined: 23 Jul 19 Posts: 289 Credit: 464,119,561 RAC: 0 |
By the way, your English is very, very good. You have studied or worked over here somewhere. Thank you! However, I have a complex about my English. I don't consider myself comfortable enough with it, especially for oral communication... We write scientific articles in English and we work with people all around the world. English is just the common language. Staying in an English-speaking country would have been a great cultural enrichment for sure. |
Send message Joined: 13 Sep 19 Posts: 69 Credit: 399,347 RAC: 0 |
But on Boinc from the first hours (past century) Look my ID on Boinc, look "the team" I am founder. 111th since creation So long time on boinc and you are not able to quote a post correctly. |
Send message Joined: 13 Nov 19 Posts: 21 Credit: 2,596,565 RAC: 0 |
Hello, Thank you for answer. English was also OK ( you could send in FR by private message !!!). It prevent escalation and ping-pong effect. You always says in most of replies there are no problem with your WU. But in first part of this reply (in french), you write everyone have problem. Then you mix scientific method and BIAS, with WU !!! After , you suggest a problem about virualisation (and acceleration.. I agree if my host are not hardware able..... But I let remember that all was perfect working (exept sometimes, WU who become stuck, it can happen. No problem, it is in beta phase). But I still wait explanation why there are problems. For the same host, no any changes in config, why all worked perfect latest year, and since two mont, no more ! I appreciate your effort. it takes time. I know, I am to help., as admin, you have full access to my host. I have check once again all, to be sure, i am not doing wrong. It can happen. Then I doscovered again a other problem. I have only host 1764 (virtual Linux running on physical host 956. I have 2 only WU (110945 and 1109765). This situation on wed04mar 20:00UTC (WU received on 04mar 19:55:34 and 19:55:23) But on my account page it seem I have also three more WU : 1109635, 1109640, 1109486) received 04mar 19:29...) Info collected from a other host who not run QuChem). Where are those WU ??? Not at me. Then about credits, on same host, usually, on host 1764) run in about 1hr-1hr30. All for 200 CR. But the new , latest received WU are estimated to tun in about 12 hours. (I know it is not accurate) I let it run But It are your WU. Those two following seem to need 12 hours. We are very far away from what you predicted : about 200CR per hour od9_athome_b3lyp-321gd,batch99,000993556,nwchem,1581785149_1 od9_athome_b3lyp-321gd,batch99,000998915,nwchem,1581785138_1 Then compare with od9_athome_b3lyp-321gd,batch52,000521854,nwchem,1580756944_2 : running : three hours or od9_athome_b3lyp-321gd,batch60,000606303,nwchem,1580807016_3 : running 54 min All for a CR of 200 !!! Only to show that your CR system is strange. : running time 54 minutes or 3 hours : all the same CR !!! I repeat, same host. Not forget all my host are in full basic config, Only OS and Boinc installed. Not used. So I can compare. To conclude, I let finish the two very strange WU (12 hours estimated) Then I return to all the other PRJ who needs your virtualization/accelaration. It works everywhere. Only here not . So, you have enough information. If you want additional test, I will help you. But the by private message, so I can send my private mail. More easy to send screenshot. I NEVER, I repeat NEVER have make remarks about (against) scientist. I do not know from where you take this. I agree sometimes, I can answer not "sweet". But I do after I receive answer where it show a not full reading Then I make difference between "reply to post" and "new post in discusuion" |
Send message Joined: 13 Nov 19 Posts: 21 Credit: 2,596,565 RAC: 0 |
Mister, I suggest you read all very carefully the first line of each post !!! I understand what you mean. I make difference between "new message" and "reply or quote" . This message is sent by "quote" You will see your serlf where it appears ! Like a new message Best regards But on Boinc from the first hours (past century) Look my ID on Boinc, look "the team" I am founder. 111th since creation |
Send message Joined: 27 Mar 20 Posts: 2 Credit: 2,200 RAC: 0 |
Hello, I have a raspberry pi (proc A-53, arm v8 instructions) + sys raspbian (debian base) i'd like to use to participate to this very interesting project. Not a powerfull computer, but could be in use 24h/24h. A way to recyle a non use raspberry... maybe others could do the same ! too bad, the project doesn't seem to be compatible with arm AArch64 instructions. is any adaptation to this type of material planned in time ? Maybe it's a realy big work for little benefits, but asking anyway. thanks ! (and sorry for my english :-)) j'en profite pour passer un bonjour à la fac des sciences de l'université d'Angers, j'y suis passé comme élève il y a une dizaine d'année en filière biochimie : très bon souvenirs (cours, profs...). bonne continuation à vous tous ! |
Send message Joined: 23 Jul 19 Posts: 289 Credit: 464,119,561 RAC: 0 |
Hello Aurélien. Currently, the code is not ARM compatible and to be honest, when we started the project we chose not to go down this path. As you say, the returns would be small for a lot of time invested. On the other hand, if someone manages to compile NWChem for ARM, the investment should be less substantial. C'est très chouette d'avoir d'anciens étudiants qui calculent pour notre projet ! Merci à toi. |
Send message Joined: 27 Mar 20 Posts: 2 Credit: 2,200 RAC: 0 |
hello damotbe, Thanks for your answer. J'aurai bien aimé participer au projet, c'est une belle initiative. Mais mon PC de bureau en X86 souffle comme un sèche cheveu quand il tourne sous boinc, même en réduisant le temps d’utilisation processeur, d'ou la question sur la compatibilité ARM. Pour ce qui est de compiler sur architecture ARM pour aider, pareil, je n'y connais rien, c'est pas du tout ma partie... dommage, sinon j'aurai aidé avec plaisir. Mais bon, un de perdu, 10 de retrouvés comme on dit. Bon courage à vous, et bonne continuation ! |
©2024 Benoit DA MOTA - LERIA, University of Angers, France