Great. I'll make sure that gets fixed in the next release.
Peter
Search found 2541 matches
- Fri Jul 19, 2013 9:52 am
- Forum: OpenMM
- Topic: Build error OpenMM5.1
- Replies: 3
- Views: 709
- Thu Jul 18, 2013 10:36 am
- Forum: OpenMM
- Topic: Build error OpenMM5.1
- Replies: 3
- Views: 709
Re: Build error OpenMM5.1
It looks like a new version of the APP SDK is incompatible with the OpenCL C++ bindings we're using. :( Try editing cl.hpp (found in platforms/opencl/src), and comment out lines 2039 through 2055. That's the section defining the setPrintfCallback() function. Does that make the error go away? Peter
- Tue Jul 16, 2013 2:32 pm
- Forum: OpenMM
- Topic: adding functions to CustomCompoundBondForce
- Replies: 19
- Views: 1794
Re: adding functions to CustomCompoundBondForce
I did some profiling to see why it's taking so long. It turns out parsing the expression isn't the problem - that takes a fraction of a second. The expensive part is differentiating it with respect to all the different variables, and optimizing all those derivative expressions.
Peter
Peter
- Tue Jul 16, 2013 1:04 pm
- Forum: OpenMM
- Topic: adding functions to CustomCompoundBondForce
- Replies: 19
- Views: 1794
Re: adding functions to CustomCompoundBondForce
Hi Aron, The code I quoted was taken from customCompoundBond.cl. The easiest way to do an if/else in an expression is by using a step() function. Compute the angle using both methods and store them in angle1 and angle2. Then you can specify something like "angle = select*angle1 + (1-select)*angle2; ...
- Mon Jul 15, 2013 3:35 pm
- Forum: OpenMM
- Topic: adding functions to CustomCompoundBondForce
- Replies: 19
- Views: 1794
Re: adding functions to CustomCompoundBondForce
Hi Aron, Computing angles in cases like this is a bit tricky. For example, here's the code that CustomCompoundBondForce uses to evaluate a dihedral term: /** * Compute the angle between two vectors. The w component of each vector should contain the squared magnitude. */ real ccb_computeAngle(real4 v...
- Fri Jul 12, 2013 10:30 am
- Forum: OpenMM
- Topic: Performance
- Replies: 3
- Views: 714
Re: Performance
Hi Matteo, Two hours for one ns means 12 ns/day, which seems a little slow. For comparison, we get 60 ns/day simulating DHFR (almost exactly half the size of your system) on a GTX 680 (slightly faster than your 580, but not enormously). I assume you're using OpenMM 5.1? It's a lot faster than earlie...
- Thu Jul 11, 2013 10:41 am
- Forum: OpenMM
- Topic: Simulations with >100k particles.
- Replies: 27
- Views: 2607
Re: Simulations with >100k particles.
Hi Spela, I don't think numerical roundoff is the issue. The spacing used in the original version is actually an accurate value for liquid water, and would be entirely reasonable if it were starting from an equilibrated configuration. The problem is that it just arranges the water molecules in a uni...
- Wed Jul 10, 2013 1:04 pm
- Forum: OpenMM
- Topic: Simulations with >100k particles.
- Replies: 27
- Views: 2607
Re: Simulations with >100k particles.
Hi Spela, It looks like HelloWaterBox needs to be cleaned up. The real problem is that it's creating water molecules that are very tightly packed together in a very high energy configuration, and the simulation is blowing up. Line 244 reads const double WaterSizeInNm = 0.3107; // edge of cube contai...
- Fri Jul 05, 2013 10:32 am
- Forum: OpenMM
- Topic: OpenCL platform and multiple devices
- Replies: 16
- Views: 2225
Re: OpenCL platform and multiple devices
Hi Siddharth, This problem was fixed a long time ago. Upgrade to a newer version of OpenMM and it should be fine. The fix was to specify only the particular device being used when creating a cl::Context. Here's the relevant code from the current version of OpenCLContext.cpp: vector<cl::Device> conte...
- Mon Jul 01, 2013 11:03 am
- Forum: OpenMM
- Topic: Minor Bug in CUDA code for trig functions
- Replies: 1
- Views: 607
Re: Minor Bug in CUDA code for trig functions
Thanks! I've just fixed it.
Peter
Peter