找回密码
 新注册用户
搜索
楼主: Youth

[项目新闻] [BOINC] [数学类] yoyo@home

[复制链接]
发表于 2007-8-28 21:33:26 | 显示全部楼层
算了幾天,突然發現YOYO一個比較“嚴重”的問題:
只要一重啟,CPU TIME立即變為0!
任務是不是從頭做不清楚,反正從頭到尾都是0%……

而且得分完全跟最后上傳時的CPU TIME有關(就是說,萬一你在做完前幾秒重啟。。。可能照我估計,你的分數就會少得可憐了……)至少我遇到了。。。有一個沒重啟的包,做了三萬秒,得了幾十分。。其它的都經常關機,所以最少的居然才十幾秒。。得分更是少。。但我似乎就沒做過十幾秒的包。。

(現在還上傳不了圖,過幾天發過來。)

可以發了:

YOYO項目得分情況

YOYO項目得分情況


[ 本帖最后由 cnchina 于 2007-8-28 21:53 编辑 ]
回复

使用道具 举报

发表于 2007-9-1 13:23:24 | 显示全部楼层
昨天又算了幾個包,發現有百分點的儲存了……但是包的大小是忽大忽小。。。
回复

使用道具 举报

发表于 2007-9-2 12:00:08 | 显示全部楼层
回复

使用道具 举报

 楼主| 发表于 2007-9-5 09:23:04 | 显示全部楼层
嗯,有进度显示了,随便下了一个包就是43小时...

原帖由 cnchina 于 2007-9-1 13:23 发表
昨天又算了幾個包,發現有百分點的儲存了……但是包的大小是忽大忽小。。。
回复

使用道具 举报

发表于 2007-9-11 06:28:41 | 显示全部楼层
原帖由 Julian_Yuen 于 2007-9-2 12:00 发表
http://www3.clustrmaps.com/count ... echenkraft.net/yoyo




靠近台湾的那个点是我 ,我在北回归线入大陆处……
回复

使用道具 举报

发表于 2007-9-11 18:49:00 | 显示全部楼层
这个项目就是一个Distributed.net的子项目OGR-25的BOINC外挂~~~本来OGR的包就已经是长度非常不确定了~~~所以计算时间的变化也很正常~~~
OGR-25的介绍参见:
http://www.equn.com/forum/viewthread.php?tid=4312
回复

使用道具 举报

发表于 2007-9-27 19:58:49 | 显示全部楼层

项目新闻

27.09.2007 OGR project status
OGR has now completed more than 70% of its search space. Some predictions predict the end for the mid of 2008.

OGR项目状态
OGR已经完成了超过70%。预计在2008年中期将会结束这一阶段。

评分

参与人数 1基本分 +10 维基拼图 +5 收起 理由
霊烏路 空 + 10 + 5

查看全部评分

回复

使用道具 举报

发表于 2007-10-3 16:17:01 | 显示全部楼层

近期项目新闻

02 October 2007 64 bit Linux version update
Some volunteers reported problems on 64 bit Ubuntu. I hope version 1.08 now fixes this issue. This version is now a native 64 bit version.

30 September 2007 Free BSD version
Added a Free BSD version for Intel.

29 September 2007 Additional clients
Thanks to Dotsch clients for PS3 and Sparc Solaris could be added. Please test and report success and problems in our forum. A Boinc 5.11.0 client for Solaris can be downloaded here.

大意是有PS3和SOLARIS还有FREE BSD以及64位的UBUNTU的版本了

评分

参与人数 1基本分 +10 维基拼图 +5 收起 理由
霊烏路 空 + 10 + 5

查看全部评分

回复

使用道具 举报

发表于 2007-12-23 06:00:06 | 显示全部楼层
好久没做这项目了,这两天偶尔做了做。。吃了我一惊。。

24708217377116 Dec 2007 15:13:30 UTC22 Dec 2007 5:55:49 UTCOverSuccessDone70,420.52138.84458.01

这相当于我做SETI得分速度的三倍呀。。。大概是大包的奖励吧?

[ 本帖最后由 cnchina 于 2007-12-23 06:01 编辑 ]
回复

使用道具 举报

发表于 2007-12-23 22:01:21 | 显示全部楼层
刚才查了下,yoyo 的rac在distributed.net 里排第一,(没看懂是不是一个子项目)
总排名是28,看来boinc的号召力的确很大……

yoyo 的得分已经站boinc中daily week的1.1%了


ps: tsp 是不是对boinc版本有要求?以前做了几个包,都报错(没有输出路径数据)

[ 本帖最后由 老冬腌菜 于 2007-12-23 22:10 编辑 ]
回复

使用道具 举报

 楼主| 发表于 2007-12-27 23:09:50 | 显示全部楼层
现在项目把evolution@home也含进来了。

yoyo@home brings existing distributed computing projects to the Boinc world using the Boinc Wrapper technology.

利用BOINC的Wrapper技术把现有的非BOINC分布式项目引入到BOINC中来。

yoyo@home is currently running the following sub-projects:

目前运行的子项目有:

evolution@home (beta) simulates evolution of a simplified model of biological reality.

the distributed.net client and runs OGR work units. This project searches for the shortest Optimal Golumb Ruler of the length 25.
On distributed.net stats server this project is registered with a own account. All distributed.net stats will be credited on this.
yoyo@home at distributed.net as participiant and team

You can choose the projects you would like to run by going to the project preferences page.

可以在项目参数设置页面中选择想要运行的项目。

评分

参与人数 1基本分 +16 维基拼图 +8 收起 理由
霊烏路 空 + 16 + 8

查看全部评分

回复

使用道具 举报

发表于 2007-12-29 13:22:05 | 显示全部楼层
evolution@home的包和OGR的包哪个大,有没有比较过?
回复

使用道具 举报

发表于 2008-1-13 21:16:25 | 显示全部楼层
OGR的包大,大得多。EVO的包半天时间3~4个小时基本能搞定,OGR的预测时间在20几小时,实际运算时间大概10几个小时。
最头痛的是OGR的包运算时会卡住,苦等它3,4个小时都不见进度动一下。重新启动后它的运算时间又回到N个小时前没有卡住的时候去了……也不知道这么多时间有没有积分。
而EVO的包的计算似乎必需要整块的时间,原来一个算到80%+的包重新启动后(就为了测试那该死的OGR重启后有没有效果……)进度又回到0%……痛苦死了……
回复

使用道具 举报

发表于 2008-1-19 21:41:02 | 显示全部楼层

项目新闻

自boinc yoyo@home 讨论区的最后一条新闻(02 October 2007)开始更新,至今日(09 January 2008)

29 September 2007 Additional clients
Thanks to Dotsch clients for PS3 and Sparc Solaris could be added. Please test and report success and problems in our forum. A Boinc 5.11.0 client for Solaris can be downloaded here.
2007年 09月 29日
有了 PS3 和 Solaris linux 的计算程序,可以在论坛上贴问题,boinc 的 Solaris 专版可以下载。

30 September 2007 Free BSD version
Added a Free BSD version for Intel.
2007年 09月 30日
添加了 FreeBSD 下的计算程序。

02 October 2007 64 bit Linux version update
Some volunteers reported problems on 64 bit Ubuntu. I hope version 1.08 now fixes this issue. This version is now a native 64 bit version.
2007年 10月 02日
由于有人报告了问题,故升级了 linux 下的64位计算程序,这是在64位的 linux 环境下完成的。

06 October 2007
After two month providing OGR back to distributed.net the yoyo@home Boinc wrapper team is on the overall rank 157 of all teams in distributed.net statistic.
OGR has now completed more than 72% of its search space, predicted end is mid of 2008.
2007年 10月 06日
回传结果使 yoyo@home Boinc wrapper team (yoyo 在 distributed.net 的相应用户+团队)在 distributed.net 的团队排名上升到第157位。OGR子项目已完成了搜索空间的  72%,预计在 2008 年中结束。

20 November 2007
I implemented a database backup and some cleanup of all the generated log and statistics.
The server runs smoothly without any manual intervention. More than 100 days OGR are crunched for distributed.net. In total individual and team stats we are in the top 100. On daily basis we are in the top 5.
I started now phase III, wrapping the next project.
2007年 11月 20日
备份了数据库,清理了日志和统计文件。服务器在没有人工干预的情况下运行得十分顺利。OGR 子项目已经运行了 100 多天。在用户和团队总分排名上我们已经在前100了,在单日排名上已经在前5了。
我开始了第三个步骤,包装下一个项目。

20 November 2007
To get rid of the sometimes appearing endless loop at 87% of the wu I created new versions for Intel and PPC Mac and for 32- and 64 bit Linux. This problem happened for round about 10% of the work units on Mac and Linux.
2007年 11月 20日
为了避免时常在 87% 附近反复循环的问题(约有10%在Mac和Linux平台的任务包会出现这种问题) ,我重写了 Intel 、 PPC Mac 上的32和64位 Linux 平台的计算程序。

22 November 2007
The first beta of evolution@home is running and it runs better than expected. The results contains already real usefull simulations which will be send back to the author.
Thanks to rebirther for the logo update.
BTW: If somebody wants to develop a nice screensaver send me a mail.
2007年 11月 22日
evolution@home 的第一个 beta 测试已经开始,并且比预期的更好。计算结果包含了将被发回作者的有用数据。
感谢rebirth重新设计了logo。
如果有人想设计个屏保请发个email与作者联系。

23 November 2007
Version 1.02 runs additional to Windows XP also on Windows 2000 and Vista. It has also a progress indicator. But it is hard to estimate the time to complete a workunit and also the progress. So it may happen that the workunit finishes already at 50%. But anyway better than no progress indicator.
2007年 11月 23日
1.02版能在 winxp win2000 和 vista 上运行。并添加了进程显示,但是估测完成一个任务的时间是很困难的,所以会有在50%时完成的情况,当然总比没有好。

25 November 2007
I added information on the server status page to see for which application work units are available.
2007年 11月 25日
更新了 server status 页面,以便查看任务类别情况。

01 December 2007
yoyo@home reached the top 50 in the overall distributed.net team stats.
2007年 12月 01日
yoyo@home 在 distributed.net 总分排名已经是前50了!

02 December 2007
I released a beta version of evolution@home for Mac. Give feedback in the forum.
2007年 12月 02日
发布 evolution@home Mac 专版,请在论坛给出回应。

14 December 2007
Since some days a race between the #1 and #2 team is ongoing. Therefore the OGR output povided to distributed.net is now 7 times more than before. I keep a very close eye to the server to ensure everytime enough work units.
2007年 12月 14日
因为团队的第一和第二的竞争很激烈,我放出了比以前多7倍的wu,以保证不会断粮…… (坐山观虎斗——译者)

21 December 2007
Short update information for evolution@home: In the last days I automated the sending of results to Laurence, the author of evolutio@home. I sent round about 100MB of compressed data. He was on a conference and busy with preparations for it. But now he reviews and imports the results. As next step I will work on fixed credits to aoid double computing. I clarified the concept and have now an idea how it could be calculated. If both things are working I will switch from beta to productive.
2007年 12月 21日
evolution@home 有短暂的更新:我把结果回传给 Laurence(evolution@home 的作者),他在开会而且很忙。不过现在他有时间了……
接下来我将在固定积分和双核计算方面下功夫,现在在固定积分方面有些想法,如果都完成了就将会将项目转换为正式了。

23 December 2007
The race between L'Alliance Francophone and SETI.Germany pushed yoyo@home in the top 30 of the overall distributed.net team stats. Thanks fellow crunchers for this christmas gift.
2007年 12月 21日
L'Alliance Francophone 和 SETI.Germany 之间的竞争使 yoyo@home 成为了 distributed.net 总分的前30,谢谢所有的参与者!

26 December 2007
I wish you a merry christmas and a happy new year. Thank you for supporting yoyo@home and running to rank #25 in the distributed.net team stats.
2007年 12月 26日
圣诞快乐+新年快乐!感谢支持!yoyo 即将是 distributed.net 的第25名

27 December 2007
Fixed credits for evolution@home workunits are now in place. Report problems or questions in our forum.
2007年 12月 27日
evolution@home 的固定积分任务已经就绪,请在论坛报告问题。

01 January 2008 Evolution@home now productive
The evolution@home application is now productive. By default you will get now workunits from both applications. You can change these settings in your preferences.
2008年 01月 01日
evolution@home 正式运行,默认是您将得到两种任务包,您将可以在参数设置页面修改。

06 January 2008 1.04 of evolution@home
Version 1.04 will no more crash if the work unit restarts. But it starts from the very beginning, it does not resume the work unit.
2008年 01月 06日
1.04版解决了wu重新开始时出现的程序错误,但是将重新开始算。(所以不要在计算过程中关机——译者)

06 January 2008 new evo wus
I will create new workunits for evolution@home. The new workunits runs longer than before. Because of missing checkpoints I will grant 30% more credits for all credits which are higher than 100.
2008年 01月 06日
我创建了新的evolution@home的wu,它们将会花更长的时间。由于没有存盘点,高于100分的wu将得到30%的奖励。

13 January 2008 new PS3 Cruncher
A memory corruption problem was recently detected in the PS3 version of the cruncher. Read more about it in our forum and report problems or even that there are no problems.
2008年 01月 13日
刚发现了PS3版程序的内存泄露问题,请在论坛提出问题——当然最好没有问题。

14 January 2008 new evolution@home wus
I created some new evolution@home workunits, which are now twice as long as before.
2008年 01月 14日
创建了用时为以前两倍的evolution@home的wu

评分

参与人数 1基本分 +60 维基拼图 +30 收起 理由
霊烏路 空 + 60 + 30

查看全部评分

回复

使用道具 举报

发表于 2008-2-8 00:17:51 | 显示全部楼层

有兴趣的人探讨一下 貌似是个参数设置(心中小小怀疑)

\BOINC\projects\www.rechenkraft.net_yoyo\S005_Preferences_1.02_Win32.txt 文件内容

READ THIS LINE FIRST: You may change preferences by editing this file. But if you enter nonsense or destroy its structure, this file is renamed as corrupt preferences. So: *Don't* change the pattern (1 textline then 1 numberline except for this line). *Don't* change the sequence of parameters. The computing work counters for this machine are for your information only. They are only valid if dont change them and preserve this file in the directory of the simulator. You may change your identity at any time or become anonymous again. The identity you enter determines what account your contributions are added to on the website.   
Public family or group name (Enter name with up to 55 characters or "anonymous". Will be published in highscores at http://www.evolutionary-research.net. If this name is not unique your results will be automatically added to the existing group with this name. ) :
Boinc Wrapper
Public personal or member name (Enter name with up to 55 characters or "anonymous". Will be published in highscores at http://www.evolutionary-research.net) :
Boinc Wrapper
Not-published GroupNameCheckPIN (Enter "anonymous" or a code with up to 20 numbers or letters, that is not similar to any passwords you use on important occasions. This code is neither published with the highscores, nor especially encrypted while stored in the preferences file on your harddisk or transmitted by email. Please store it also in some other place in case you delete the preferences file.This string must not contain spaces or tabs and is used for identification when calculating the highscores: If you submit your first result, the NameCheckPIN in there is remembered. If you then submit other results after that, their NameCheckPIN is compared to the first one to make shure that you submitted these results with the URL they promote.) Now enter the code (up to 20 characters without space or tabs) :

Not-published PersonalNameCheckPIN (see explanation above for GroupNameCheckPIN). This ID keeps other members of your group from adding to your highscores, as all results submitted under your group and personal name without this ID will be anonymized. Now enter the code (up to 20 characters without space or tabs) :

Public PromotedURL (Your email or homepage that shall be used as a link to allow people surfing the highscores, to be published at http://www.evolutionary-research.net to visit your website or email you. If your URL is not correct, people will not find you. The email account used to send your results to evolutionary-research will not be published whatsoever.) Use "anonymous" or meaningful URL with up to 90 characters :
http://www.rechenkraft.net/yoyo
Nation (enter name or "anonymous", may be published in highscores at http://www.evolutionary-research.net) :
Germany
ANSI CPU information (automatic, don't correct manually) :
Unknown
ANSI OS information (automatic, don't correct manually) :
Win32 on Windows NT 5.1
Systemdetails known (unknown or details like CPU type+speed, Cache size+speed, RAM size+speed, up to 70 characters) :
unknown
EEPerformance of this system (MegaIndividualsPerSecond in a 1165084 individuals (= 100 MB) population):
0.719848366654281
Total computing time (days) recorded in this preferences file :
0.0804564351851843
Total number of workunits (in GigaIndividuals) recorded in this preferences file :
4.723075831
Worldhistory runs recorded in this preferences file :
4
InterruptFast: Call after so many individuals :
8301
InterruptSlow: Check for break-file after so many individuals :
1660281
ScreenActivity: Write a short message if the Ratchet clicks or this number of Generations are over. This may lead to out of memory errors, if RAM is short for long simulations, as every message is recorded in memory! So use this feature with caution:
0
StandardOutputToFile (default = 0): 1 = Batch-run output will be written to log-file (uses diskspace) 0 = Batch-run output will NOT be written to log-file (saves diskspace). (You may get problems with a completely filled harddisk, if you request too much output) :   
0
StandardOutputToScreen (default = 1): 1 = Batch-run output will be written to screen (uses RAM) 0 = Batch-run output will NOT be written to screen (saves RAM, allows for simulation of larger populations). (Simulator may terminate, if you request too much output) :   
0
Maximum log-file size before simulation stops can be from 100KB to 2GB (default = 10 000 000 bytes). One standardsimulation with all reporting off needs ca. 10KB in log-file  (+ ca. 100KB in separate results-file). Think about the number of simulations you want to compute with one run-file. Then estimate the number of bytes needed in the log file and add space for extra reports.Check whether the harddisk with the simulators folder has enough free space.Here is the current maximal number of bytes a log-file is allowed to have:
2000000
Write-intervall for intermediate results to disk in minutes (60 = default; this saves intermediate results so that they can be added to the resultsfile after a systemcrash. The intermediate result file is deleted automatically after adding its content to the result file after starting the simulator the next time) :
10
GetTimeSeriesRequestsFromFile (Shall the simulator evaluate the file "TimeseriesRequests.txt" during every evolve-loop? yes = 1, no = 0) :
0
WriteProgressFileIntervallSec (Write one "eProgress.txt" file after how many seconds) :
25
WriteLastParametersListFileIntervallMin (write a new ParametersList print out to file after how many minutes):
10
AcceptKeyRAMminimumMB (Reject keys that need less MB than) :
0
AcceptKeyRAMmaximumMB (Reject keys that need more MB than specified here; Do not use virtual memory, as this will block your computer; Leave enough for your other work; as a rule of thumb, your upper limit should be less than half of your physical RAM) :
1000
AcceptKeyComputingTimeMinimumHours (Reject keys that need less hours computing time than) :
0
AcceptKeyComputingTimeMaximumDays (Reject keys that need more days computing time than) :
7
Last change : 2007Oct29_19h19m59
At startup 811942_-1465029363_-2040615522 was determined as the CodeSecurityCheckSum for this file. Compare it to the public CodeSecurityCheckSum at the download site, if there are problems. Differences might be due to transmission errors, read-write errors on disks, viruses or other unauthorized modifications. If you find an discrepancy, delete the corresponding binary simulator file immediately and get a new one.

评分

参与人数 1基本分 +10 收起 理由
霊烏路 空 + 10

查看全部评分

回复

使用道具 举报

您需要登录后才可以回帖 登录 | 新注册用户

本版积分规则

论坛官方淘宝店开业啦~

Archiver|手机版|小黑屋|中国分布式计算总站 ( 沪ICP备05042587号 )

GMT+8, 2024-3-29 21:37

Powered by Discuz! X3.5

© 2001-2024 Discuz! Team.

快速回复 返回顶部 返回列表