[AMBER] KShake problem with GPU

From: Lorenzo Gontrani <lorenzo.gontrani.gmail.com>
Date: Wed, 9 Apr 2014 16:05:33 +0200

Dear Amber experts, we have recently installed a GPU version of Amber12,
and we are experiencing problems in running a simulation on a system made
up of about 15000 atoms, after the initial minimization.
The MD engine do not even start after reading the restart structure and
velocities, giving the following error: "Error: invalid configuration
argument launching kernel kShake".

>From a rapid inspection of the error routines, it seems that this
message regards
kshake.cu, that deals with the initial set-up of the process. No problem in
fact, is found when MPI is used, nor with a version of Amber11 we have on
another machine.
The non-working pmemd is installed on a machine having the following
graphic cards:

+------------------------------------------------------+

| NVIDIA-SMI 4.304.54 Driver Version: 304.54
|
|-------------------------------+----------------------+----------------------+
| GPU Name | Bus-Id Disp. | Volatile Uncorr.
ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute
M. |
|===============================+======================+======================|
| 0 Tesla S2050 | 0000:0F:00.0 Off |
0 |
| N/A 58C P1 N/A / N/A | 0% 6MB / 2687MB | 0%
Default |
+-------------------------------+----------------------+----------------------+
| 1 Tesla S2050 | 0000:10:00.0 Off |
0 |
| N/A 58C P1 N/A / N/A | 0% 6MB / 2687MB | 0%
Default |
+-------------------------------+----------------------+----------------------+


while the machine bearing the "good" program has


nvidia-smi
Wed Apr 9 16:00:56 2014
+------------------------------------------------------+

| NVIDIA-SMI 5.319.37 Driver Version: 319.37
|
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr.
ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute
M. |
|===============================+======================+======================|
| 0 Tesla C2075 Off | 0000:02:00.0 Off |
0 |
| 30% 77C P0 83W / N/A | 10MB / 5375MB | 0%
Default |
+-------------------------------+----------------------+----------------------+
| 1 GeForce 210 Off | 0000:83:00.0 N/A |
N/A |
| N/A 33C N/A N/A / N/A | 62MB / 1023MB | N/A
Default |
+-------------------------------+----------------------+----------------------+

Can anybody give us a clue on how to cope with this problem?

Thanks a lot, kind regards

Lorenzo Gontrani

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Apr 09 2014 - 07:30:02 PDT
Custom Search