CAS@home summary for 2012
log in

Advanced search

Message boards : News : CAS@home summary for 2012

Author Message
wenjing wu
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 13 Sep 10
Posts: 161
Credit: 751,216
RAC: 0
Message 975 - Posted: 18 Feb 2013, 6:09:22 UTC
Last modified: 18 Feb 2013, 10:12:26 UTC

大家好,
2012年下半年度,CAS@home整个项目取得很大的进展,我们将两个应用程序(清华大学的纳米技术研究软件LAMMPS和中科院计算所的蛋白质结构预测软件TreeThreader)移植到CAS@home计算平台。在移植过程中,我们遇到种种问题,因此投入了大量的人力在改善应用程序的稳定性和容错性上。

--------------------------------------------------------------------------------

LAMMPS:
LAMMPS应用程序2012年5月正式上线,从2012.9到目前为止,我们一共更新了13个版本(当前的稳定版本是1.29)。在这个版本里,我们已经将错误主机率(错误主机是指没有成功运行过任何作业的主机)从31%(版本1.16)降低到5%(版本1.29), 这个是国际上志愿计算应用所能接受的主机错误率范围。
从上线开始,LAMMPS共获得 62万的有效CPU小时,完成了20个对动态分子进行模拟的任务。目前,这些数据需要被送往清华大学,等待科学家的后期处理。


--------------------------------------------------------------------------------

TreeThreader
TreeThreader应用程序于2012.9月正式上线,到目前为止,我们一共更新了18个版本(当前的稳定版本Windows为1.17, Linux为1.18)。错误主机率由34%(版本1.00)降低到27%(1.17)。但是由于应用程序本身在某些主机上还存在问题。计算所的开发人员还在对此进行调试,以实现一个较低的错误主机率。
从上线开始,TreeThreader共获得了9.5万有效CPU小时,完成了约1700条蛋白质序列的结构预测,目前还有大约6000条蛋白质序列需要进行预测。志愿者们可以按照如下的方式,查看自己所参与计算的蛋白质序列的被预测的结构:

我们目前正在编写一些程序,定期统计各个志愿者所参与的蛋白质序列预测,希望通过邮件形式,定期给志愿者一个反馈。

--------------------------------------------------------------------------------

请参加如下页面:
CAS@home应用程序版本发布历史

CAS@home应用程序错误主机率统计

CAS@home的项目统计

对于2012的工作,CAS@home团队还有很多方面需要改进:
由于项目的人员更替、人员缺乏、在BOINC平台移植跨平台程序的经验缺乏以及平台和项目运行方面的经验不足,我们的程序在运行过程中,出现了大大小小的问题,造成了对志愿者计算时间的浪费,对此我们深表歉意。在这个过程中,我们的志愿者给我们提供了积极的、有用的反馈信息,使得我们能够及时进行改正。这些将成为CAS@home在以后运行过程中的宝贵经验。目前CAS@home的团队还很年轻,在项目管理方面还在向国外其他成熟的项目进行学习。基于2012年运行的经验与教训,我们在平台管理方面进行了规范化。比如:
1) 建立较大规模的测试平台,对移植的应用程序进行更全方位的测试。
2) 对于每个新发布的版本,在测试平台上经过严格测试,才发布到CAS@home平台
3) 科学计算用户的作业需要在测试平台上进行测试,确保作业本身没有问题,才能被提交到CAS@home平台
4) 科学计算用户提交一批新作业前,必须与平台运行人员沟通,共同检查测试运行结果,以保证作业不出现异常情况

CAS@home是由中科院高能所计算中心志愿提供的一项服务。到目前为止,CAS@home没有任何正式经费的支持,我们的团队人员大多是利用自己的业务时间,在从事这项工作。跟每一位参与到CAS@home的志愿者一样,我们也希望能够为中国的志愿计算的发展贡献自己的一份力量。

我们相信,在大家共同的努力下,CAS@home这个项目会长远发展下去。CAS@home项目的宗旨不是为了发文章,捞科研资本,而是要搭建这个分布式计算的平台,成为广大志愿者和科学应用之间的一个桥梁,将广大志愿者所贡献的计算资源提供给有迫切需求的科学计算工作者们使用!

相信2013会更好!
谢谢各位长久以来的支持!

-CAS@home 团队

Greetings, everyone!
During the year of 2012, we have made great progress on CAS@home with all the kind help, including yours! We ported two applications on the platform, including dynamical molecular simuation software LAMMPS (used for Nanotech research by scientsts from Tsinghua University) and protein structure prediction software TreeThreader (used by scientists from ICT, CAS). And we devoted a lot of manpower to improve the stability and fault tolerance of the 2 applications in the second half year of 2012 as we officially launced them and got sufficient results back from volunteer hosts.

LAMMPS:
LAMMPS was offically launced in May 2012, and we have released 13 new version after the official launch which significally reduced the faulty hosts rate (a faulty host is a host which never successfully finishes any job) from 31% (version 1.16) to 5% (version 1.29).
LAMMPS has gained 620K effective CPU hours worth proccess rate ever since its launch, and finished 20 batches of simulation. The results will be transfered back to Tsinghua and be post processed by the scientists as we finish setting up the data transfer channel between IHEP and Tsinghua.

TreeThreader:
TreeThreader was officially launched in September 2012. We have released 18 versions for both Windows and Linux platform ever since then to fix various bugs we saw along the running process or reported by volunteers. The faulty host rate is reduced from 34% (version 1.00) to 27%(1.18). However, the TreeThreader application still fails on 27% of the windows machines, developers from ICT are still working on investigating and fixing the problem.
TreeThreader has gained about 95K effective CPU hours worth process rate ever since its launch, and successfully finished the struction prediction of about 1700 protein sequences so far, and there are still 6000 more sequences needs to be processed in the following period.
Volunteers can follow these instructions to check the protein structure your computer contributed to compute.
We are also writting scripts to generate summary of each volunteer's contribution over periods, and would send emails to volunteers about the summary.

A few pages for your references:
CAS@home Application version release history

CAS@home Applicationfaulty hosts rate

CAS@home project wide statistics

In the year of 2012, we have run into a lot of problems on CAS@home due to the lack of manpower, lack of experience on porting cross platform applications on BOINC and project management experience. However, our volunteers have been giving us very positive and useful feedbacks on various problems then can detect from their computings which greatly helped us to debug and eventually find solutions to fix these problems. CAS@home is growing up with all your help, and we greatly appreciate your patience and kind help in the process.

We learnt valuable lessons through the past experiences, so we have put more regulations into our project management. For example:
1) We have deployed a more comprehensive test platform to test the applications
2) Each release of applicaiton must be strictly tested on the test platform and pass the standards before it can be launched on CAS@home
3) New jobs need to be tested thoroughly and results need to be verified by scientists on the test platform before they can be submitted to CAS@home.
4) Before submitting a new batch a job, scientists users need to communicate with platform managers and analyze the results together to avoid exceptions happening.


CAS@home is a project managed by volunteers from Computer Center, Institute of High Energy Physics. We formed a loose team to manage this project, and so far there is no official funding to support all the related work, so most work of this project is done by our team mates in their spare time. Like all the volunteers to CAS@home project, we would like also to contribute our effort to volunteer computing, and help with the scientific computing. The only purpose of CAS@home is to provide this distrbuted platform to bridge volunteers and scientific computing projects which are in bad needs of a large amount of computing power.

As the first Chinese volunteer computing project, CAS@home is still young and in the process of growing up. We are fighting to release more stable application versions to volunteer hosts, and also learning from the experience of other well managed volunteer computing projects.

We believe things can get better in 2013!

Many thanks for your great effort!

-CAS@home team
____________
加油!CAS@home!我们帮助科学家跟时间赛跑!
Go CAS@home! We help scientists to race against time!


Post to thread

Message boards : News : CAS@home summary for 2012