Gromacs Problems/Lost/Restarting WU's

[Kvothe]

Over the last few weeks here alot of people have been talking about WU's that have been lost and deleted, that have restarted, or have just locked up their PC's.

I myself have been dealing with hard freezes off and on over the last few weeks with a stable rig (all stability testing ran sucessfully, Prime, Memtest, etc.) all while folding Gromacs WU's with FAHCore_78.exe.

After having done a little bit of digging at www.folding-community.org, it appears that there might be some various issues with the Gromacs WU's that are currently going out, and that the same issue's we've been posting about have been mentioned there.

There's no official word from Stanford/Vijay yet, but here's to hoping they get this taken care of!

 

[Duster]

Good find p[H]ant0m.
VV WAHOO I JUST HIT 250k JUST NOTICED

 

[unhappy_mage]

Right on the nose, too! What are the odds of that?

 

[Kvothe]

Well it appears that right now, the servers that had been assigning WU's that require Core 78 have no stopped assigning Gromacs for the time being.

 

[Kvothe]

Well as a final aside to this issue, Stanford has come through and produced a new Core 78 for use. Details on how to update this core can be found in the previously linked thread.

 

[KodiakStar]

We reproduced the error and have put up a new FahCore_78.exe v.1.86 that appears to fix the issue . The server that was forcing download of 1.84 is now forcing upgrade to the new core. If you wish to immediately upgrade, you can quit the client, delete the existing core, and restart the client.

For the lazy..

 

[clean_me]

So just to clarify, for all those 'borgs' that aren't readily available and need the new core they can fix themselves?

 

[OSUguy98]

I'm 99% sure F@H will check the next time it down loads a new GROMAC, and it'll grab the newest core... That's the way it's always seemed to happen for me in the past, at least....


Keep on Folding!! For the [H]orde!!

 

[Kvothe]

It appears that way. They should complete the WU they're currently working on and then the update is being pushed out from Stanford.

 

[Heliocentrical]

Crap! I was hoping I'd make it.

Launched as a service.
Entered C:\Program Files\F@H to do work.

[12:30:11] - Ask before connecting: No
[12:30:11] - Use IE connection settings: Yes
[12:30:11] - User name: Aratech (Team 33)
[12:30:11] - User ID: 524C4A722BF52843
[12:30:11] - Machine ID: 1
[12:30:11]
[12:30:11] Loaded queue successfully.
[12:30:11] + Benchmarking ...
[12:30:15]
[12:30:15] + Processing work unit
[12:30:15] Core required: FahCore_78.exe
[12:30:15] Core found.
[12:30:15] Working on Unit 07 [August 30 12:30:15]
[12:30:15] + Working ...
[12:30:21]
[12:30:21] *------------------------------*
[12:30:21] Folding@Home Gromacs Core
[12:30:21] Version 1.81 (April 6, 2005)
[12:30:21]
[12:30:21] Preparing to commence simulation
[12:30:21] - Looking at optimizations...
[12:30:21] - Files status OK
[12:30:28] - Expanded 197151 -> 968749 (decompressed 491.3 percent)
[12:30:29]
[12:30:29] Project: 886 (Run 2, Clone 259, Gen 9)
[12:30:29]
[12:30:29] Assembly optimizations on if available.
[12:30:29] Entering M.D.
[12:30:49] (Starting from checkpoint)
[12:30:50] Protein: p886_p53peptide886
[12:30:50]
[12:30:50] Writing local files
[12:30:50] Completed 1728815 out of 5000000 steps (35)
[12:30:51] Extra SSE boost OK.
[13:06:24] CoreStatus = 1 (1)
[13:06:24] Client-core communications error: ERROR 0x1
[13:06:24] Deleting current work unit & continuing...

 

[Kvothe]

Crap! I was hoping I'd make it.

That's strange, the errors the new core are supposed to fix are 0x79 and LINCS errors.

Is there any other evidence that you've had them fail like that?

 

[Heliocentrical]

That's strange, the errors the new core are supposed to fix are 0x79 and LINCS errors.

Is there any other evidence that you've had them fail like that?

What do mean by evidence? You mean other than the log?

 

[Kvothe]

Just more crashes like that in your logs.

If you've had multiple issues with a specific error, or WU's bombing out, stability testing is the order of the day.

Although with the issues presented by Stanford, I'd replace the core with the new one using the instructions K* posted and go from there.

 

[Heliocentrical]

Just more crashes like that in your logs.

If you've had multiple issues with a specific error, or WU's bombing out, stability testing is the order of the day.

Although with the issues presented by Stanford, I'd replace the core with the new one using the instructions K* posted and go from there.

So about the core fix...

Does it still apply if you download a new copy of the console? Or is it only applicable to existing installs?

 

[Kvothe]

Well the hope is that any time you get a new Gromacs WU, the new core will be pushed to you. So you could wait and see if you get it, it'll show up in the logs as Gromacs v1.86. If you want it now, stop folding, delete FAHcore_78.exe and it'll download itself the new one.

 

[Heliocentrical]

Well the hope is that any time you get a new Gromacs WU, the new core will be pushed to you. So you could wait and see if you get it, it'll show up in the logs as Gromacs v1.86. If you want it now, stop folding, delete FAHcore_78.exe and it'll download itself the new one.

I went to delete FAHcore_78 but I dont have one. I show FAHcore_79. Same?

 

[Kvothe]

I went to delete FAHcore_78 but I dont have one. I show FAHcore_79. Same?

Nope, not the same.

I can't see how you don't have 78 unless it has already been removed to prepare for updating.

Launched as a service.
Entered C:\Program Files\F@H to do work.

[12:30:11] - Ask before connecting: No
[12:30:11] - Use IE connection settings: Yes
[12:30:11] - User name: Aratech (Team 33)
[12:30:11] - User ID: 524C4A722BF52843
[12:30:11] - Machine ID: 1
[12:30:11]
[12:30:11] Loaded queue successfully.
[12:30:11] + Benchmarking ...
[12:30:15]
[12:30:15] + Processing work unit
[12:30:15] Core required: FahCore_78.exe
[12:30:15] Core found.

According to this snippet of the logfile you posted, it was there at that time.

 

[Heliocentrical]

Nope, not the same.

I can't see how you don't have 78 unless it has already been removed to prepare for updating.



According to this snippet of the logfile you posted, it was there at that time.

It was there. I deleted everything and reinstalled the console. That's when it changed to 79.