new cores for qmd & gromacs
csimon
Acadiana Icrontian
Today I noticed a new qmd core 1.04 and a new gromacs core 1.81. While I'm not sure what the changes to the gromacs core is I think I know that the qmd core is addressing previous stall issues.
Check out the thread here at the community!
http://forum.folding-community.org/viewtopic.php?p=96647#96647
"I found the hole in locking mechanism, and patched the core code. Guha is uploading the core right now, so you can get it soon. The new core will be downloaded automatically as usual from the next QMD WU, but you may want to force it by stopping, deleting old core, and then restarting the client. However, if you had this problem, you have to make sure that no core is running after stopping F@H. (by using task manager)
The reason was a left-over locking object from a previous run, and I could exactly reproduce what was reported in this thread. And I suspect that HT problem is related to this (I can't be sure 100% since I couldn't reproduce it here -- but in theory the reported symptom can be explained with this bug.) "
Check out the thread here at the community!
http://forum.folding-community.org/viewtopic.php?p=96647#96647
"I found the hole in locking mechanism, and patched the core code. Guha is uploading the core right now, so you can get it soon. The new core will be downloaded automatically as usual from the next QMD WU, but you may want to force it by stopping, deleting old core, and then restarting the client. However, if you had this problem, you have to make sure that no core is running after stopping F@H. (by using task manager)
The reason was a left-over locking object from a previous run, and I could exactly reproduce what was reported in this thread. And I suspect that HT problem is related to this (I can't be sure 100% since I couldn't reproduce it here -- but in theory the reported symptom can be explained with this bug.) "
0
Comments
Stall issues, schmall issues! I want my QMD!
I had a machine stalled at 97% on a 600-Point WU. Needless to say, I was not going to let those points get away if I could help it.
I stopped the client and deleted the old core, then heaved a sigh of relief when the WU finished normally.
The next WU on that machine was a 149-Point Gromac. It is running at a 191ppw pace, far below the 1000+ppw average that particular computer had been making. I deleted the core (again), but it didn't help.
Anyone out there having a similar experience?
either I'm just lucky, or the fixes actually worked