Gromacs update

[Bruce]

If you're running Gromacs under windows, there was a bug fix released yesterday. You'll get an automatic update when you get your next Gromacs WU, but if you'd like to get it sooner, you can stop the client, delete core_78, and restart the client.

I haven't heard a word about the Linux core or the Mac X core, so they probably didn't have the bug.

It was a "handle leak" which can lead to crashes (particularly with Win9x) and some performance degradation.

[22:09:55] Folding@home Gromacs Core
[22:09:55] Version 1.46 (April 21, 2003)

[Miles]

Version 1.46 O.K

I'm still running 1.45 on the dually. I'll look tomorrow to see if it updated.

Thanks.

[veggyhed]

Had no clue there was a problem. Thanks for the info

[veggyhed]

Wait a minute. My client just grabbed a GROMACS WU and I am still running 1.45. Is this a problem if all are running well.


[05:38:34] Loaded queue successfully.
[05:38:39] + Closed connections
[05:38:39]
[05:38:39] + Processing work unit
[05:38:39] Core required: FahCore_78.exe
[05:38:39] Core found.
[05:38:39] Working on Unit 07 [May 4 05:38:39]
[05:38:39] + Working ...
[05:38:39]
[05:38:39] *------------------------------*
[05:38:39] Folding@home Gromacs Core
[05:38:39] Version 1.45 (April 21, 2003)
[05:38:39]
[05:38:39] Preparing to commence simulation
[05:38:39] - Looking at optimizations...
[05:38:39] - Created dyn
[05:38:39] - Files status OK
[05:38:40] - Expanded 351453 -> 1760993 (decompressed 501.0 percent)
[05:38:40] - Starting from initial work packet
[05:38:40]
[05:38:40] Project: 542 (Run 77, Clone 86, Gen 15)
[05:38:40]
[05:38:40] Assembly optimizations on if available.
[05:38:40] Entering M.D.
[05:38:46] Protein: p542_BBA5_N in water
[05:38:46]
[05:38:46] Writing local files
[05:38:48] Writing local files
[05:38:48] Completed 0 out of 500000 steps (0)
[05:46:25] Writing local files
[05:46:25] Completed 5000 out of 500000 steps (1)

[Medic193]

As long as the update doesn't show up on the computer I'm fine. Last thing I need is my borged machines being found out. :)

[Nofinger]

Please help me for a moment, "Gromac" is this the Gah/Fah client or a add on for the client ?

[Medic193]

When you put in the -forceasm -advmethods flags it will use a file with 78 in it. This is the gromacs core. So technically it's both.

[zhotfire]

The FAH3 client uses 3 cores:

FahCore_65.exe - tinker(folding)
FahCore_78.exe - gromacs(folding)
FahCore_ca.exe - genome

Each core crunches it's own specific woonit(depending on availability and project requirements). The gromacs core has proved to be the most efficient.

Not sure if the gromacs core has the ability to crunch new genes like GAH.... Bruce?

[X-Calibur]

how do I get this core then? I have the _65 core and using both switches... something is wrong Lucy!

[Medic193]

Well if you are running the text client. Delete the work folder from your folding directory. If you are using EM3, type in the parameters, -advmethods -forceasm. Execute EM3 and it will start the client with the gromacs core and d/l 78. You can also start the client manually with the parameters after the target in the shortcut. Hope that makes sense. Basically, just delete the work folder and start the client again with the new parameters.

[zhotfire]

X: It should kick in when you download your next woonit... if it doesn't, something's amiss...

[veggyhed]

It depends on the server. Every week I seem to do all 3 WU`s. I would not worry about it. If u look at the project summary page and take a look at the ip address u have been assigned u will see what that server is dishing out and how many WU`s it has .

[Bruce]

Most of the questions have already been answered, but I'll add a bit of a summary.

Genome was originally run with software called "SPA" (which is an acronym, but I don't recall for what). In the transition to GAH2 it became known as "SPG" which included some improvements to the internals, as well as interfacing with the FAH3 client as a front-end.

Folding was originally run with software called "Tinker" and it's still doing a pretty good job. The new kid on the block for folding is Gromacs - - which contains almost everything that was in Tinker . . . and it runs several times faster. It also adds a much wider choice of analysis tools that the biochemists seem to really like. In the not-too-distant future, I expect Tinker to be discontinued, but it won't cause any particular uproar, because as far as the casual user is concerned, nothing changes.

As far as the team is concerned, however, the extra speed means most people will generate points at a significantly faster rate. For anybody who is on a folding team (like our team 131) and who has a relatively modern CPU (which supports SSE or 3DNow+) you really should be running the Gromacs core rather than the Tinker core. There still are a few bugs (such as the one that started this thread), but for all except the unattended machines, the risk is acceptable.

[Medic193]

OK so if tiner will eventually be discontinued, what will that do to the people who use the google bar?

[Bruce]

One of the key members of the Pande Group left Stanford and went to work for Google. They still have some backroom communications that keep both clients heading in somewhat the same direction.

The Google Toolbar must be based on extremely stable software (since it pretty much has to run, unmonitored). Gromacs has come a long, long way, but it's still not as stable as Tinker. The transition at Stanford will be slow, but will precede changes at Google, but I have no doubt that Google will support Gromacs when the right time comes.