Try setting the cmake variable FOUND_CUDART to the location of the the 64 bit cudart library. If that doesn't help, set CUDA_HAVE_GPU to TRUE, which should at least get the cuda tests built.
Search found 32 matches
- Wed Jul 14, 2010 2:19 pm
- Forum: OpenMM
- Topic: Compiling OpenMM on Ubuntu Linux
- Replies: 19
- Views: 5132
- Wed Jun 16, 2010 8:31 am
- Forum: OpenMM
- Topic: cannot find -lOpenMMCuda
- Replies: 3
- Views: 796
RE: cannot find -lOpenMMCuda
What was the solution? Describing it might help others with similar issues.
- Fri Mar 19, 2010 7:29 pm
- Forum: OpenMM
- Topic: OpenMM Source Build Linux 32-bit tests failed
- Replies: 9
- Views: 1889
RE: OpenMM Source Build Linux 32-bit tests failed
Hmm. It looks like all of the Cuda tests are failing. Do the cuda test programs such as deviceQuery, which should be in the cuda sdk, work correctly?
- Tue Mar 16, 2010 10:21 am
- Forum: OpenMM
- Topic: Why OpenMM1.1examples have not Makefile?
- Replies: 2
- Views: 685
RE: Why OpenMM1.1examples have not Makefile?
<r>Thanks for reporting this.<br/> <br/> I have replaced the OpenMM1.1examples.zip file on the openmm downloads page (<URL url="https://simtk.org/project/xml/downloads.xml?group_id=161"><LINK_TEXT text="https://simtk.org/project/xml/downloads ... oup_id=161">https://simtk.org/project/xml/downloads.x...
- Tue Mar 16, 2010 7:43 am
- Forum: OpenMM
- Topic: Gromacs-OpenMM on Windows
- Replies: 16
- Views: 3243
RE: Gromacs-OpenMM on Windows
<t>When you run Zephyr binaries from the command line, you need to set certain environment variables correctly. This should be described in chapter 5 of the Zephyr manual. Failure to find params.agb suggests that the GMXLIB variable needs to be set to a particular directory within the Zephyr install...
- Mon Mar 15, 2010 5:19 pm
- Forum: OpenMM
- Topic: Gromacs-OpenMM on Windows
- Replies: 16
- Views: 3243
RE: Gromacs-OpenMM on Windows
It's not obvious to me what the trouble might be. We don't have much experience building on 64-bit Windows, nor on Windows 7.
Are you by any chance running the cygwin cmake executable rather than native windows cmake?
Does Zephyr run correctly on your system?
Are you by any chance running the cygwin cmake executable rather than native windows cmake?
Does Zephyr run correctly on your system?
- Thu Mar 11, 2010 8:24 pm
- Forum: OpenMM Zephyr
- Topic: Zephyr and NVidia GPU
- Replies: 7
- Views: 2453
RE: Zephyr and NVidia GPU
<t>Hmmm. That's definitely odd. Do the cuda sdk test programs such as deviceQuery still report a cuda-capable GPU?<br/> <br/> What kind of computer is this on?<br/> <br/> Sometimes a GPU can get kind of grumpy and stop responding. Shut down the computer, unplug it, replug it, and restart. See if tha...
- Tue Mar 09, 2010 11:28 am
- Forum: OpenMM Zephyr
- Topic: Selecting which GPU to run a simulation
- Replies: 4
- Views: 1611
RE: Selecting which GPU to run a simulation
Good news! Thanks for the report. I wasn't certain it would work. I am glad to hear that it does.
Could you please mention Luis what sort of computer you are using? (Windows, Mac, Linux?)
Could you please mention Luis what sort of computer you are using? (Windows, Mac, Linux?)
- Wed Mar 03, 2010 4:26 pm
- Forum: OpenMM Zephyr
- Topic: Selecting which GPU to run a simulation
- Replies: 4
- Views: 1611
RE: Selecting which GPU to run a simulation
<t>Depending on what operating system you are using, it might be possible to run two Zephyr simulations on two different GPUs.<br/> <br/> Try launching two instances of Zephyr from two different shells. Before launching Zephyr, set the environment variable OPENMM_PLATFORM differently in each shell. ...
- Thu Feb 25, 2010 2:37 pm
- Forum: SimTK Core Toolset
- Topic: Pin the RNA to the ground
- Replies: 1
- Views: 652
RE: Pin the RNA to the ground
<r>The modelCompounds() method automatically creates a multibody system based on the BondMobilities of the molecule. But modelCompounds() always creates molecules that are free to move with respect to Ground. There is no easy way for you to get around that behavior.<br/> <br/> But you might be able ...