24小时热门版块排行榜

返回列表

当前只显示满足指定条件的回帖，点击这里查看本话题的所有回帖

shenghaisi

金虫 (正式写手)

应助: 0 (幼儿园)
贵宾: 0.2
金币: 1621.9
散金: 2200
红花: 2
帖子: 586
在线: 2066.6小时
虫号: 54610
注册: 2004-10-01
专业: 生物大分子结构与功能

[交流] Desmond多cluster并行已有1人参与

大家可以看看这个问题，不知道desmond多cluster 并行如何做到？

Hello, I am a college student attempting a summer project of setting
up a small computer cluster to run desmond. I have three machines
(1) 4 core xeon
(2) 6 core Phenom
(3) 2 core Athlon

So far, I have been able to setup a beowulf cluster with these
machines and have MPI running successfully across all 12 cores.
I have installed desmond all three machines. I have created a
schrodinger.host file that looks like this
####################################
#
name:       localhost
host:       athlon.helix.hpu.edu
schrodinger: /usr/local/schrodinger2010
tmpdir:    /usr/tmp
processors: 2
#
name:       xeon
host:       xeon.helix.hp.edu
schrodinger: /usr/local/schrodinger2010
tmpdir:    /usr/tmp
processors: 4
#
name:       phenom
host:       athlon.helix.hpu.edu
schrodinger: /usr/local/schrodinger2010
tmpdir:    /usr/tmp
processors: 6
#
############################################
I am using the athlon machine as the machine for job submission. I
can run desmond on individual machine but how do I specify desmond to
run multisim on all 12 CPUs? Does the schrodinger.host file look
correct? Sorry if this is a very basic question and I should be
knowing this.
Thanks for you help

Hi ,
That looks good to me.  However, when running Desmond across multiple
machine without a queueng system like Torque, SGE, etc, you may find
that the machines file for MPI is malformed in some cases.  Because of
this, I recommend that you change the names of your entries in the
schrodinger.hosts file, so that they are resolvalbe on the network:
####################################
#
name:       localhost
host:       athlon.helix.hpu.edu
schrodinger: /usr/local/schrodinger2010
tmpdir:    /usr/tmp
processors: 2
#
name:       xeon
host:       xeon.helix.hp.edu
schrodinger: /usr/local/schrodinger2010
tmpdir:    /usr/tmp
processors: 4
#
name:       athlon
host:       athlon.helix.hpu.edu
schrodinger: /usr/local/schrodinger2010
tmpdir:    /usr/tmp
processors: 6
#
############################################
Here, I just changed the "name" parameter of the athlon.helix.hpu.edu
machine so that it matched the FQDN (fully qualified domain name).
This should ensure that the machines file (which is auto-generated by
Desmond'd Job Control mechanism) is generated properly.  Then, you
launched the job using:
$SCHRODINGER/utilities/multisim -HOST "xeon:2 althlon:2" -cpu 4 -mode
umbrealla ...
For more details on launching multisim from the command line, see the
Desmond v2.4 User Manual.
Best,
James
On May 20, 5:08 pm, raymond ...@gmail.com> wrote:
- 隐藏被引用文字 -
- 显示引用的文字 -
> Hello, I am a college student attempting a summer project of setting
> up a small computer cluster to run desmond. I have three machines
> (1) 4 core xeon
> (2) 6 core Phenom
> (3) 2 core Athlon
> So far, I have been able to setup a beowulf cluster with these
> machines and have MPI running successfully across all 12 cores.
> I have installed desmond all three machines. I have created a
> schrodinger.host file that looks like this
> ####################################
> #
> name:       localhost
> host:       athlon.helix.hpu.edu
> schrodinger: /usr/local/schrodinger2010
> tmpdir:    /usr/tmp
> processors: 2
> #
> name:       xeon
> host:       xeon.helix.hp.edu
> schrodinger: /usr/local/schrodinger2010
> tmpdir:    /usr/tmp
> processors: 4
> #
> name:       phenom
> host:       athlon.helix.hpu.edu
> schrodinger: /usr/local/schrodinger2010
> tmpdir:    /usr/tmp
> processors: 6
> #
> ############################################
>  I am using the athlon machine as the machine for job submission. I
> can run desmond on individual machine but how do I specify desmond to
> run multisim on all 12 CPUs? Does the schrodinger.host file look
> correct? Sorry if this is a very basic question and I should be
> knowing this.
> Thanks for you help

回复此楼