fci-Folding@home Info Question

This forum contains information about 3rd party applications which may be of use to those who run the FAH client and one place where you might be able to get help when using one of those apps.

Moderator: Site Moderators

Post Reply
Shadowtester
Posts: 79
Joined: Tue Dec 04, 2007 4:18 am

fci-Folding@home Info Question

Post by Shadowtester »

I have been using fci for a while and love it one thing I was trying to figure out is how to get project info for the gpu2 client I have it monitoring the client fine and it shows the project and some info but it does not show the gpu2 projects in the Projects list. Is there some configuration problem that is preventing it from displaying the detailed gpu2 project info.
Shadowtester

Image
smoking2000
Posts: 471
Joined: Mon Dec 03, 2007 6:20 am
Location: Amsterdam
Contact:

Re: fci-Folding@home Info Question

Post by smoking2000 »

The GPU client doesn't generate a work/current.xyz file from which an image of the project can be generated. Only projects which have a current.xyz are processed into the known-projects.xml file which is primarily used by the Projects page.

If you have more than 50 known projects, paging will be enabled on the Projects page, which also includes a form to jump to a specific project number as you can see on my FCI server:
http://fci.fatalerrorgroup.com/?target=projects.plc

The project you jump to is looked up in the known-projects.xml file, and if it's not found there it's looked for in the client-projects.xml file and as a last resort it's looked for in the project-summary.xml. But what's the difference between these files you may ask.

known-projects.xml stores the project info for projects that had a current.xyz file uploaded, it stores the types of images generated from the xyz file and source files, and it also stores the data on Stanfords psummary for that project (that's why you need to run fci-update-stanford-files.pl). Example:

Code: Select all

<known_projects>
  <project>
    <number>1150</number>
    <date_created>2006-01-31 22:22:27</date_created>
    <date_updated>2006-04-17 23:38:10</date_updated>
    <files>
      <pdb>project-data/p1150.pdb</pdb>
      <xyz>project-data/p1150.xyz</xyz>
    </files>
    <image_list>
      <image>
        <type>rotate</type>
        <file>project-data/p1150.gif</file>
        <script>project-data/p1150.rotate.script</script>
        <type_name>Rotating Molecule</type_name>
      </image>
      <image>
        <type>still</type>
        <file>project-data/p1150.jpg</file>
        <script>project-data/p1150.still.script</script>
        <type_name>Still Image</type_name>
      </image>
      <image>
        <type>xyz180</type>
        <file>project-data/p1150.xyz180.gif</file>
        <script>project-data/p1150.xyz180.script</script>
      </image>
    </image_list>
    <jmol_url>http://jmol.sourceforge.net/fah/?project=1150</jmol_url>
    <project_summary>
      <atoms>84665</atoms>
      <code>GROMACS</code>
      <contact>csnow</contact>
      <credit>600.00</credit>
      <deadline>82.00</deadline>
      <description>http://fah-web.stanford.edu/cgi-bin/fahproject?p=1150</description>
      <frames>100</frames>
      <last_update>2006-04-03 09:52:47</last_update>
      <name>p1150_RIBO_semihelixfrom1141</name>
      <preferred>30.00</preferred>
      <server>171.65.103.156</server>
      <status>
        <active>false</active>
      </status>
    </project_summary>
    <workunit_name>p1150_RIBO_semihelixfrom1141</workunit_name>
  </project>
  ...
</known_projects>
client-projects.xml stores all the projects that are currently being worked on by clients monitored by FCI, it stores the data on Stanfords psummary for the project and a list of clients working on it, for each client it stores the current work unit information. This file is used by the Assigned Projects page. Example:

Code: Select all

<client_projects>
  <project>
    <number>1152</number>
    <atoms>393</atoms>
    <client_list>
      <client>
        <name>amon-ra</name>
        <current_work_unit>
          <code>TINKER</code>
          <credit>239.00</credit>
          <credit_per_hour>3.950</credit_per_hour>
          <expected>Sun Apr 30 09:48:01 2006</expected>
          <folding>
            <benchmark>722</benchmark>
            <clone>53</clone>
            <generation>15</generation>
            <misc>500, 200</misc>
            <run>148</run>
          </folding>
          <issued>Thu Apr 27 21:18:00 2006</issued>
          <name>p1152_L939_K12M_ext_from638</name>
          <progress>0.5%</progress>
          <progress_bar>[__________]</progress_bar>
          <progress_state>unknown</progress_state>
          <project_name>p1152_L939_K12M_ext_from638</project_name>
          <project_number>1152</project_number>
        </current_work_unit>
        <last_update>2006-04-27 21:42:26</last_update>
      </client>
    </client_list>
    <code>TINKER</code>
    <contact>csnow</contact>
    <credit>239.00</credit>
    <deadline>--</deadline>
    <description>http://fah-web.stanford.edu/cgi-bin/fahproject?p=1152</description>
    <frames>400</frames>
    <last_update>2006-04-03 09:52:47</last_update>
    <name>p1152_L939_K12M_ext_from638</name>
    <preferred>45.00</preferred>
    <server>171.64.122.112</server>
    <status>
      <active>false</active>
    </status>
  </project>
  ...
</client_projects>
project-summary.xml is the master list of projects, which becomes quite big over time, because every time fci-update-stanford-files.pl is run it updates all projects on the psummary page in the project-summary.xml. This file only stores the data of Stanfords psummary page. Example:

Code: Select all

<project_summary>
  <project>
    <number>1014</number>
    <atoms>132</atoms>
    <code>GROMACS</code>
    <contact>esorin</contact>
    <credit>39.00</credit>
    <deadline>13</deadline>
    <description>http://fah-web.stanford.edu/cgi-bin/fahproject?p=1014</description>
    <frames>100</frames>
    <last_update>2007-01-13 03:50:52</last_update>
    <name>p1014_a21nat_289_dd19</name>
    <preferred>9</preferred>
    <server>171.67.89.149</server>
    <status>
      <active>false</active>
    </status>
  </project>
  ...
</project_summary>
So depending on in which file the project is listed, some types of additional information may be available besides the project info provided by Stanford.

If an image is available for the project, it's listed in the known-projects.xml file and therefor the Files & Work Unit Image sections can be shown on the Projects page.

If no image is available for the project, but the project is currently being worked on by a client, it's listed in the client-projects.xml file which does not contain an image, so the Files & Work Unit Image sections are not shown on the projects page.

If the project is not listed in either the known-projects.xml nor client-projects.xml file, but is listed in the project-summary.xml file, only the Stanford data is available so this is the only thing shown (the Project Information section).
Shadowtester
Posts: 79
Joined: Tue Dec 04, 2007 4:18 am

Re: fci-Folding@home Info Question

Post by Shadowtester »

Ok thanks a lot I did not know that the gpu2 client did not produce the current.xyz file needed to generate the project info. Also I would like to say thank you for such a great program I may not have t set up exactly correctly I am using separate fci-client.py for each client even if there are more than one client on the same computer I was not sure if you could setup the fci-client.py to monitor multiple FAH clients on the same machine so I just set up a second fci-client.py with a different name and a second crontab job to take care of the second client. Is that the best way?

Also a request in the next version it would be nice to have the option on how fci-Folding@home inf calculates ppd for the clients like Fahmon with the option of either taking the total time from the download to completion like currently and also maybe an option for just the last 3 steps or just the last step as well if the client is not running 24/7 the expected times using the current method can be way off where a last 3 steps or last step would be more accurate.
Shadowtester

Image
smoking2000
Posts: 471
Joined: Mon Dec 03, 2007 6:20 am
Location: Amsterdam
Contact:

Re: fci-Folding@home Info Question

Post by smoking2000 »

Shadowtester wrote:I may not have t set up exactly correctly I am using separate fci-client.py for each client even if there are more than one client on the same computer I was not sure if you could setup the fci-client.py to monitor multiple FAH clients on the same machine so I just set up a second fci-client.py with a different name and a second crontab job to take care of the second client. Is that the best way?
Yeah, if you need to monitor more than one FAH client on a system, like the SMP & GPU clients on your Shuttle system, you need to add an additional crontab entry for each client specifying the FAH client directory with --dir and giving the client in FCI a unique name with --name (because the system hostname is the default if no name is explicitly set).

Example crontab:

Code: Select all

user@hostname:~$ crontab -l
5 * * * * /usr/local/bin/fci-client.pl --name hostname-smp --dir /home/user/fah-smp/ --url http://example.com/fci/index.pl --username fci-user --password fci-password
6 * * * * /usr/local/bin/fci-client.pl --name hostname-gpu --dir /home/user/fah-gpu/ --url http://example.com/fci/index.pl --username fci-user --password fci-password
Shadowtester wrote:Also a request in the next version it would be nice to have the option on how fci-Folding@home inf calculates ppd for the clients like Fahmon with the option of either taking the total time from the download to completion like currently and also maybe an option for just the last 3 steps or just the last step as well if the client is not running 24/7 the expected times using the current method can be way off where a last 3 steps or last step would be more accurate.
FCI doesn't calculate PPD, only the TPF. The PPD FCI shows is calculated by qd and if I'm not mistaken FahMon has adopted qds way of calculating PPD and expected time of completion.

There is a bug/feature in the new SMP & GPU clients on Windows where the FahCores are not updating the mtime (last modification timestamp) of the work/logfile_??.txt and/or work/wuinfo_??.dat. qd can calculate an insane expected date in this case, because qd uses the mtime of either of those files to know when the FAH client last had activity (it could be shutdown when qd is run), the progress qd reads from either of the files is used to calculate the expected time based on the time passed since the WU was begun and the FAH clients last activity. When the mtime of the file is not updated, qd thinks that a lot of progress was made in a short amount of time (the time between the WU start and creation/last mtime update of the file), and the projection in to the future is therefor screwed.
Post Reply