Commits · 1ebe88ba70e77c08dec11b9e81972acd978653aa · tsoc / openmm

09 Jul, 2015 1 commit
- OpenCL allows different force groups to have different cutoffs · f065df01
  Peter Eastman authored Jul 09, 2015
  
  f065df01
01 May, 2015 1 commit
- Optimization for AMD · 460f016e
  peastman authored May 01, 2015
  
  460f016e
10 Mar, 2015 1 commit
- More bug fixes to multi-GPU · f7ef2dd0
  peastman authored Mar 10, 2015
  
  f7ef2dd0
06 Mar, 2015 2 commits
- For multi-GPU, exceeding the size of the neighbor list causes the forces to be marked as invalid · bd666c27
  peastman authored Mar 06, 2015
  
  bd666c27
- Load balancing between GPUs forces the neighbor list to be rebuilt · 62f7c44c
  peastman authored Mar 06, 2015
  
  62f7c44c
07 Jan, 2015 1 commit
- Began implementing OpenCL support for triclinic boxes · 195bcecf
  Peter Eastman authored Jan 07, 2015
  
  195bcecf
04 Nov, 2014 1 commit
- Improvements to multi-GPU performance · f32c804b
  peastman authored Nov 04, 2014
  
  f32c804b
09 Oct, 2014 1 commit
- Further optimizations to NonbondedForce · 3d1b2186
  peastman authored Oct 09, 2014
  
  3d1b2186
09 Sep, 2014 1 commit
- Ported the new version of findBlocksWithInteractions() to OpenCL (not yet debugged) · 70bd61a6
  peastman authored Sep 09, 2014
  
  70bd61a6
26 Nov, 2013 1 commit
- Bug fix to selecting workgroup size for findInteractingBlocks() · a3fd6c23
  peastman authored Nov 26, 2013
  
  a3fd6c23
23 Nov, 2013 1 commit
- Workarounds for devices whose maximum workgroup size varies between kernels · 35b9d787
  peastman authored Nov 22, 2013
  
  35b9d787
16 Jul, 2013 1 commit
- Work correctly on devices with SIMD width less than 32 · d0800362
  peastman authored Jul 16, 2013
  
  d0800362
04 Jun, 2013 1 commit

Converted the array containing atom block indices for the neighbor list from... · 2ff0b0ae

peastman authored Jun 04, 2013

Converted the array containing atom block indices for the neighbor list from ushort2 to int.  This removes the hard limit of 2 million atoms.

2ff0b0ae

03 May, 2013 1 commit
- Very minor optimizations and cleanup to building neighborlist · 78f8d078
  Peter Eastman authored May 03, 2013
  
  78f8d078
23 Apr, 2013 1 commit
- Fixed some incorrect argument indices · 4e9167d0
  Peter Eastman authored Apr 23, 2013
  
  4e9167d0
19 Apr, 2013 1 commit

Fixes a hard to catch bug when a boundingBoxSize increases in size in between... · 282c6f2c

Yutong Zhao authored Apr 19, 2013

Fixes a hard to catch bug when a boundingBoxSize increases in size in between a timestep, but does not trigger a rebuild of the neighbourlist. Affects the usage of singlePeriodicCopy in nonbonded force.

282c6f2c

11 Apr, 2013 1 commit
- Changed workgroup size to work around NVIDIA bug. This also improves performance slightly. · 01afee8b
  Peter Eastman authored Apr 11, 2013
  
  01afee8b
22 Mar, 2013 1 commit
- Merged 5.1Optimizations branch back to trunk · 93c467b2
  Peter Eastman authored Mar 22, 2013
  
  93c467b2
26 Oct, 2012 1 commit
- Bug fixes · 61bb8786
  Peter Eastman authored Oct 26, 2012
  
  61bb8786
19 Oct, 2012 1 commit
- The test for identical exceptions between nonbonded forces is order insensitive · ff7f1035
  Peter Eastman authored Oct 19, 2012
  
  ff7f1035
17 Oct, 2012 2 commits
- Finished implementing double precision for OpenCL · 59f8cf87
  Peter Eastman authored Oct 17, 2012
  
  59f8cf87
- Continuing to implement double precision in OpenCL · c8dac206
  Peter Eastman authored Oct 17, 2012
  
  c8dac206
10 Oct, 2012 1 commit

OpenCLArray is no longer templatized and doesn't provide a host buffer. This... · 1107aa83

Peter Eastman authored Oct 10, 2012

OpenCLArray is no longer templatized and doesn't provide a host buffer.  This is in preparation for adding mixed/double precision support.

1107aa83

23 Jun, 2012 1 commit
- Fixed errors under Intel OpenCL · f0454a4b
  Peter Eastman authored Jun 23, 2012
  
  f0454a4b
20 Mar, 2012 1 commit
- Tolerate round-off error when checking the box size against the cutoff distance · 56766760
  Peter Eastman authored Mar 20, 2012
  
  56766760
14 Mar, 2012 1 commit
- AMD support for 64 bit fixed point forces (currently disabled) · a5a7b902
  Peter Eastman authored Mar 14, 2012
  
  a5a7b902
02 Mar, 2012 1 commit
- Throw exception if the box size is reduced to less than twice the cutoff · 801c43ee
  Peter Eastman authored Mar 02, 2012
  
  801c43ee
14 Feb, 2012 1 commit
- OpenCL implementation of multiple time step integrators · 91247209
  Peter Eastman authored Feb 14, 2012
  
  91247209
27 Jan, 2012 1 commit
- Tony's optimizations to reduce local memory use · fed50628
  Peter Eastman authored Jan 27, 2012
  
  fed50628
17 Nov, 2011 1 commit
- Fixed bug in parallel nonbonded force · 1e2606ec
  Peter Eastman authored Nov 17, 2011
  
  1e2606ec
02 Nov, 2011 1 commit
- Fixed some incorrect argument indices · 63819881
  Peter Eastman authored Nov 02, 2011
  
  63819881
25 Oct, 2011 1 commit
- More changes from Tony to reduce local memory use · ce74c6ae
  Peter Eastman authored Oct 25, 2011
  
  ce74c6ae
15 Oct, 2011 1 commit
- Fixed some incorrect argument indices · b982b0e4
  Peter Eastman authored Oct 15, 2011
  
  b982b0e4
11 Oct, 2011 1 commit
- Tony Tye's optimization to reduce local memory use · 4405b165
  Peter Eastman authored Oct 11, 2011
  
  4405b165
07 Oct, 2011 1 commit
- Added "const" and "restrict" to lots of kernel arguments to let the compiler do more optimizations · 69e75377
  Peter Eastman authored Oct 07, 2011
  
  69e75377
23 Sep, 2011 1 commit
- Optimized number of work groups · 5ed1d277
  Peter Eastman authored Sep 23, 2011
  
  5ed1d277
02 Sep, 2011 1 commit
- On Fermi, store forces as 64 bit fixed point and update with atomic... · d9029f61
  Peter Eastman authored Sep 02, 2011
```
On Fermi, store forces as 64 bit fixed point and update with atomic operations.  Also various other minor optimizations.
```
  d9029f61
22 Aug, 2011 1 commit
- Fixed another case where the exclusions for CustomGBForce could be handled incorrectly · 84a14401
  Peter Eastman authored Aug 22, 2011
  
  84a14401
18 Aug, 2011 1 commit
- Fixed bug when CustomGBForce was told to use exclusions, but didn't have any... · e274d455
  Peter Eastman authored Aug 18, 2011
```
Fixed bug when CustomGBForce was told to use exclusions, but didn't have any exclusions.  (See bug 1519.)
```
  e274d455
18 Jul, 2011 1 commit
- Reduced number of force buffers on AMD · 952757a8
  Peter Eastman authored Jul 18, 2011
  
  952757a8