CVS time sorted logs

Time Author Lines Revision Filename Comments
2008/04/23 00:15:16 jim +2 -59 1.17 src/ComputeSelfTuples.hCleaning up bond code, removed UniqueList for tuples.
2008/04/23 00:15:16 jim +1 -10 1.1019 src/ComputeImpropers.hCleaning up bond code, removed UniqueList for tuples.
2008/04/23 00:15:16 jim +8 -7 1.1049 src/ComputeHomeTuples.hCleaning up bond code, removed UniqueList for tuples.
2008/04/23 00:15:16 jim +1 -11 1.1006 src/ComputeDihedrals.inlCleaning up bond code, removed UniqueList for tuples.
2008/04/23 00:15:16 jim +1 -18 1.3 src/ComputeCrossterms.hCleaning up bond code, removed UniqueList for tuples.
2008/04/23 00:15:16 jim +1 -7 1.1005 src/ComputeBonds.inlCleaning up bond code, removed UniqueList for tuples.
2008/04/23 00:15:16 jim +1 -9 1.1008 src/ComputeAngles.inlCleaning up bond code, removed UniqueList for tuples.
2008/04/14 23:14:47 dbwells2 +1 -1 1.4 src/GridForceGrid.CRemoving "DEBUGM" definition to suppress debugging output.
2008/04/09 19:16:49 jim +2 -1 1.1030 src/ProxyMgr.hAIX compile fixes.
2008/04/09 19:16:49 jim +1 -1 1.1062 src/ProxyMgr.CAIX compile fixes.
2008/04/09 19:16:49 jim +1 -1 1.2 src/GridForceGrid.hAIX compile fixes.
2008/04/09 19:16:48 jim +3 -3 1.114 src/ComputePme.CAIX compile fixes.
2008/04/09 19:16:45 jim +1 -1 1.3 arch/Linux-POWER-xlc.archAIX compile fixes.
2008/04/09 19:16:45 jim +3 -3 1.9 arch/AIX-POWER-xlC.archAIX compile fixes.
2008/04/08 16:23:40 jim +1 -1 1.1232 MakefileRe-enable fixed extracopy parallel optimization.
2008/04/08 06:29:31 chaomei2 +3 -2 1.1061 src/ProxyMgr.CFixed a bug when removing the extra copy of ProxyResultMsg. The bug is caused by incorrect indexing into force arrays.
2008/04/08 02:30:21 jim +1 -1 1.1231 MakefileDisable buggy extracopy parallel optimization that was giving bad forces.
2008/04/07 21:22:24 dbwells2 +41 -35 1.3 src/GridForceGrid.CGridforce update: voltage offsets between periodic images of grid fixed.
2008/04/01 03:22:20 petefred +148 -112 1.2 ug/figures/dual_top.pdfFix figures from chipot's additions to the alchemy docs
2008/04/01 03:22:19 petefred +614 -449 1.2 ug/figures/dual_top.epsFix figures from chipot's additions to the alchemy docs
2008/03/30 19:36:21 petefred +1 -1 1.13 ug/ug_alchemy.texFix Jerome's name
2008/03/20 19:17:22 jim +2 -0 1.1057 src/NamdState.CFix crash when using constraints with amber coordinate file.
2008/03/20 19:12:33 petefred None 1.1 ug/figures/overlap_dens3.pdfAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:33 petefred None 1.1 ug/figures/overlap_dens2.pdfAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:32 petefred None 1.1 ug/figures/overlap_dens1.pdfAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:32 petefred None 1.1 ug/figures/overlap3.psAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:32 petefred None 1.1 ug/figures/overlap3.pdfAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:32 petefred None 1.1 ug/figures/overlap2.psAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:32 petefred None 1.1 ug/figures/overlap2.pdfAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:32 petefred None 1.1 ug/figures/overlap1.psAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:32 petefred None 1.1 ug/figures/overlap1.pdfAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:30 petefred +425 -213 1.12 ug/ug_alchemy.texAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:29 petefred +1 -1 1.11 ug/ug.texAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:29 petefred None 1.1 ug/fep.bibAdd in Chipot's changes to the fep documentation
2008/03/20 19:12:29 petefred +1 -0 1.40 ug/MakefileAdd in Chipot's changes to the fep documentation
2008/03/07 01:15:56 chaomei2 +8 -3 1.1027 src/Patch.hChanges due to keeping the extra copy overhead for ProxyDataMsg becauseif we still want the 32-byte alignment on CompAtom list regardless theunderlying Charm layer.
2008/03/07 01:15:56 chaomei2 +16 -3 1.1047 src/Patch.CChanges due to keeping the extra copy overhead for ProxyDataMsg becauseif we still want the 32-byte alignment on CompAtom list regardless theunderlying Charm layer.
2008/03/07 00:59:34 chaomei2 +9 -5 1.1230 MakefileAdd flags for removing the extra copy overhead related to proxy msgs.
2008/03/07 00:57:41 chaomei2 +35 -14 1.1043 src/ProxyPatch.C1. Add comppile-time flags to allow users to choose whether to removethe extra copy overhead of proxy msgs (ProxyDataMsg and ProxyResultMsg).2. Add the optimization to remove the extra copy involved in the proxyresult msg.2. Add padding bytes to the varsize proxy msgs, so that the beginning ofCompAtom list in the ProxyDataMsg is 32-byte aligned if the underlyingCharm layer has ensured the beginning of a msg is 32-byte aligned;the beginning of Force list in the ProxyResultVarsizeMsg is 8-bytealigned. This is done to accord to the alignment requirement before thisoptimization since the alignment will give better performance,especially on the BG/L machine.
2008/03/07 00:57:41 chaomei2 +41 -0 1.1029 src/ProxyMgr.h1. Add comppile-time flags to allow users to choose whether to removethe extra copy overhead of proxy msgs (ProxyDataMsg and ProxyResultMsg).2. Add the optimization to remove the extra copy involved in the proxyresult msg.2. Add padding bytes to the varsize proxy msgs, so that the beginning ofCompAtom list in the ProxyDataMsg is 32-byte aligned if the underlyingCharm layer has ensured the beginning of a msg is 32-byte aligned;the beginning of Force list in the ProxyResultVarsizeMsg is 8-bytealigned. This is done to accord to the alignment requirement before thisoptimization since the alignment will give better performance,especially on the BG/L machine.
2008/03/07 00:57:40 chaomei2 +10 -0 1.1012 src/ProxyMgr.ci1. Add comppile-time flags to allow users to choose whether to removethe extra copy overhead of proxy msgs (ProxyDataMsg and ProxyResultMsg).2. Add the optimization to remove the extra copy involved in the proxyresult msg.2. Add padding bytes to the varsize proxy msgs, so that the beginning ofCompAtom list in the ProxyDataMsg is 32-byte aligned if the underlyingCharm layer has ensured the beginning of a msg is 32-byte aligned;the beginning of Force list in the ProxyResultVarsizeMsg is 8-bytealigned. This is done to accord to the alignment requirement before thisoptimization since the alignment will give better performance,especially on the BG/L machine.
2008/03/07 00:57:40 chaomei2 +63 -4 1.1060 src/ProxyMgr.C1. Add comppile-time flags to allow users to choose whether to removethe extra copy overhead of proxy msgs (ProxyDataMsg and ProxyResultMsg).2. Add the optimization to remove the extra copy involved in the proxyresult msg.2. Add padding bytes to the varsize proxy msgs, so that the beginning ofCompAtom list in the ProxyDataMsg is 32-byte aligned if the underlyingCharm layer has ensured the beginning of a msg is 32-byte aligned;the beginning of Force list in the ProxyResultVarsizeMsg is 8-bytealigned. This is done to accord to the alignment requirement before thisoptimization since the alignment will give better performance,especially on the BG/L machine.
2008/03/07 00:57:40 chaomei2 +5 -2 1.1055 src/HomePatch.h1. Add comppile-time flags to allow users to choose whether to removethe extra copy overhead of proxy msgs (ProxyDataMsg and ProxyResultMsg).2. Add the optimization to remove the extra copy involved in the proxyresult msg.2. Add padding bytes to the varsize proxy msgs, so that the beginning ofCompAtom list in the ProxyDataMsg is 32-byte aligned if the underlyingCharm layer has ensured the beginning of a msg is 32-byte aligned;the beginning of Force list in the ProxyResultVarsizeMsg is 8-bytealigned. This is done to accord to the alignment requirement before thisoptimization since the alignment will give better performance,especially on the BG/L machine.
2008/03/07 00:57:39 chaomei2 +29 -1 1.1152 src/HomePatch.C1. Add comppile-time flags to allow users to choose whether to removethe extra copy overhead of proxy msgs (ProxyDataMsg and ProxyResultMsg).2. Add the optimization to remove the extra copy involved in the proxyresult msg.2. Add padding bytes to the varsize proxy msgs, so that the beginning ofCompAtom list in the ProxyDataMsg is 32-byte aligned if the underlyingCharm layer has ensured the beginning of a msg is 32-byte aligned;the beginning of Force list in the ProxyResultVarsizeMsg is 8-bytealigned. This is done to accord to the alignment requirement before thisoptimization since the alignment will give better performance,especially on the BG/L machine.
2008/03/03 21:12:46 chaomei2 +14 -0 1.1250 src/SimParameters.CAdded a warning for memory optimized version that the load balancing period and the first step to do load balancing should be a multiple of stepspercycle in order to avoid some msgoverheads during runs.
2008/03/03 21:11:28 chaomei2 +9 -0 1.1027 src/ComputePatchPair.CA little change for the memory optimized version
2008/03/03 21:11:28 chaomei2 +3 -3 1.1040 src/ComputeNonbondedPair.CA little change for the memory optimized version
2008/03/03 21:10:29 chaomei2 +4 -0 1.1013 src/PositionOwnerBox.hAdded functions to return the owner of the position box.
2008/03/03 21:10:29 chaomei2 +4 -0 1.1007 src/PositionBox.hAdded functions to return the owner of the position box.
2008/03/03 21:09:31 chaomei2 +16 -8 1.1032 src/NamdTypes.hCompAtomExt class is now needed even for non-memory optimized version because of the change in ProxyPatch related msgs
2008/03/03 21:05:52 chaomei2 +13 -4 1.1008 src/ProxyPatch.h1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +111 -34 1.1042 src/ProxyPatch.C1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +34 -27 1.1028 src/ProxyMgr.h1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +9 -4 1.1011 src/ProxyMgr.ci1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +8 -90 1.1059 src/ProxyMgr.C1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +13 -2 1.1026 src/Patch.h1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +9 -8 1.1046 src/Patch.C1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +11 -0 1.1054 src/HomePatch.h1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/03/03 21:05:52 chaomei2 +61 -38 1.1151 src/HomePatch.C1. Declare the ProxyDataMsg and ProxyAllMsg as the same type which makes codes cleaner. The only overhead per msg added is just 4 bytes.2. Change the packed type ProxyDataMsg (including ProxyAllMsg) into varsize msg, and the position array inside the msg will be reused in the computation so that the overhead of extra copyis avoided.3. Fixed a tricky bug that is related to load balancing for the memory optimized version.
2008/02/25 23:58:12 petefred +1 -0 1.1015 src/structures.hAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:12 petefred +4 -0 1.1049 src/common.hAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:12 petefred +10 -6 1.1177 src/WorkDistrib.CAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:12 petefred +2 -0 1.1137 src/SimParameters.hAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:12 petefred +17 -0 1.1249 src/SimParameters.CAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:12 petefred +2 -3 1.1046 src/Sequencer.hAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:12 petefred +99 -94 1.1161 src/Sequencer.CAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:11 petefred +2 -0 1.1072 src/Molecule.hAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:11 petefred +74 -10 1.1139 src/Molecule.CAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:11 petefred +6 -1 1.1053 src/HomePatch.hAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:11 petefred +254 -21 1.1150 src/HomePatch.CAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/25 23:58:11 petefred +7 -3 1.1218 src/Controller.CAdd initial implementation of TIP4P water.TIP4P is requested with the watModel tip4p simulation parameter. I'll be curating and distributing compatible top/par files; a couple of namd-l denizens seem interested in early access, but all of the physical properties check out ok, so I don't expect that many further changes will be needed until we move to wanting more general lone pairs.
2008/02/11 19:20:16 bhatele +38 -38 1.21 src/ComputePmeMgr.cifixed indentation for better readability
2008/02/11 02:56:19 jim +4 -0 1.1248 src/SimParameters.CCatch error where energy is printed without updated full electrostatics.Didn't affect dynamics but tended to confuse the users.
2008/02/11 02:14:44 jim +1 -0 1.2 src/GlobalMasterTMD.hFix bug that broke targetted MD for parallel runs. Forces were calculatedcorrectly but applied to the wrong targetted atoms in random permutation.
2008/02/11 02:14:44 jim +13 -2 1.3 src/GlobalMasterTMD.CFix bug that broke targetted MD for parallel runs. Forces were calculatedcorrectly but applied to the wrong targetted atoms in random permutation.
2008/02/10 20:29:43 jim +8 -0 1.6 src/GlobalMasterServer.CCatch bug. Due to a design error, GlobalMasterServer does not supportindividual atom requests from multiple global force clients on parallel runs.
2008/02/09 21:45:40 jim +1 -0 1.6 src/GlobalMasterFreeEnergy.CFix crash when truncating very long config for printing.
2008/02/03 05:21:50 bhatele +1 -1 1.1136 src/SimParameters.hTo use new load balancers use "ldbStrategy asb8" in the config file
2008/02/03 05:21:50 bhatele +4 -4 1.1247 src/SimParameters.CTo use new load balancers use "ldbStrategy asb8" in the config file
2008/02/03 05:21:50 bhatele +3 -3 1.84 src/NamdCentLB.CTo use new load balancers use "ldbStrategy asb8" in the config file
2008/02/03 05:08:37 bhatele +0 -3 1.27 src/Sync.CPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:37 bhatele +4 -4 1.80 src/Rebalancer.CPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:37 bhatele +1 -1 1.1007 src/ProxyPatch.hPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:37 bhatele +2 -0 1.1041 src/ProxyPatch.CPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:37 bhatele +1 -1 1.1027 src/ProxyMgr.hPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:37 bhatele +15 -14 1.1058 src/ProxyMgr.CPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:36 bhatele +0 -3 1.1048 src/common.hPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:36 bhatele +1 -1 1.1052 src/HomePatch.hPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 05:08:36 bhatele +19 -16 1.1149 src/HomePatch.CPROXY_SPAN_DIM is no longer a #defineit is now defined in ProxyMgr.C as proxySpanDimEventually it'll become a SimParameter
2008/02/03 02:26:10 bhatele +2 -4 1.11 src/TorusLB.Csome extra statements removed
2008/02/03 02:26:10 bhatele +19 -28 1.12 src/RefineTorusLB.Csome extra statements removed
2008/02/03 02:26:10 bhatele +2 -6 1.31 src/Rebalancer.hsome extra statements removed
2008/02/03 02:26:10 bhatele +7 -3 1.79 src/Rebalancer.Csome extra statements removed
2008/01/23 07:19:31 bhatele +3 -1 1.113 src/ComputePme.Cadded commonly wrappers around some work
2008/01/23 07:19:31 bhatele +11 -17 1.1048 src/ComputeHomeTuples.hadded commonly wrappers around some work
2008/01/20 21:07:53 bhatele +4 -4 1.11 src/RefineTorusLB.C*** empty log message ***
2008/01/17 23:30:10 bhatele +5 -4 1.30 src/Rebalancer.hcorrected some changes which shouldn't be turned on in the CVS version
2008/01/17 23:30:10 bhatele +2 -2 1.84 src/LdbCoordinator.Ccorrected some changes which shouldn't be turned on in the CVS version
2008/01/17 23:30:09 bhatele +9 -3 1.78 src/Rebalancer.Ccorrected some changes which shouldn't be turned on in the CVS version
2008/01/17 23:21:58 bhatele +5 -6 1.1135 src/SimParameters.hnew ldbs can be used from the config file now
2008/01/17 23:21:58 bhatele +4 -8 1.1246 src/SimParameters.Cnew ldbs can be used from the config file now
2008/01/17 23:21:58 bhatele +10 -52 1.83 src/NamdCentLB.Cnew ldbs can be used from the config file now
2008/01/17 23:21:57 bhatele +4 -6 1.83 src/LdbCoordinator.Cnew ldbs can be used from the config file now
2008/01/14 19:30:41 bhatele +4 -2 1.10 src/RefineTorusLB.Cchanges to the PROXY_CORRECTION code
2008/01/14 19:30:41 bhatele +22 -24 1.29 src/Rebalancer.hchanges to the PROXY_CORRECTION code
2008/01/14 19:30:41 bhatele +39 -100 1.77 src/Rebalancer.Cchanges to the PROXY_CORRECTION code
2008/01/07 22:15:20 bhatele +2 -3 1.10 src/TorusLB.Cfixed compilation error
2007/12/28 23:28:29 bhatele +1 -0 1.2 src/TorusLB.hadded badP
2007/12/28 23:27:58 bhatele +28 -5 1.9 src/TorusLB.CThis case was not caught by the LDBs earlier (where we couldnt find a processorto place a compute). Now we place it on a overloaded processor - badP
2007/12/20 16:47:22 jim +1 -1 1.9 arch/Win32.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.6 arch/Tru64-Alpha.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.5 arch/T3E.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.5 arch/Solaris-i686.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.7 arch/Solaris-Sparc.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.8 arch/Origin2000.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.4 arch/MacOSX-i686.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.5 arch/MacOSX-PPC.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.4 arch/Linux-ia64.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.12 arch/Linux-i686.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.2 arch/Linux-i686-redhat9.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.2 arch/Linux-PPC.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.2 arch/Linux-POWER.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.5 arch/Linux-Alpha.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.5 arch/HPUX.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.3 arch/HPUX-ia64.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.4 arch/HPUX-PA2.0.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.4 arch/Cygwin-i686.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.4 arch/Cygwin-i686-MSVC.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.3 arch/CRAY-XT3.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.3 arch/CRAY-X1.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.2 arch/BlueGeneP.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:22 jim +1 -1 1.2 arch/BlueGeneL.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:47:21 jim +1 -1 1.2 arch/AIX-POWER.tclRemove USE_NON_CONST flags, since source now has USE_COMPAT_CONST.
2007/12/20 16:02:20 bhatele +3 -3 1.9 src/RefineTorusLB.Ccompilation error fixed
2007/12/13 04:57:02 bhatele +69 -63 1.8 src/TorusLB.Cmade the load balancers aware of multiple nodes per processor
2007/12/13 04:57:02 bhatele +83 -76 1.8 src/RefineTorusLB.Cmade the load balancers aware of multiple nodes per processor
2007/12/11 05:49:58 chaomei2 +14 -1 1.1071 src/Molecule.hFixed errors that multiple memory deletion operations are performed regard with some data structures in the Molecule object. This error has caused NAMD-smp version to crash. Only one Molecule object copy is kept on a node in the smp version.
2007/12/11 05:49:58 chaomei2 +2 -0 1.1138 src/Molecule.CFixed errors that multiple memory deletion operations are performed regard with some data structures in the Molecule object. This error has caused NAMD-smp version to crash. Only one Molecule object copy is kept on a node in the smp version.
2007/12/10 23:06:17 bhatele +36 -4 1.76 src/Rebalancer.Cadding a penalty to proxy creationthis helps on 511 procs on Abe (turned off by default)
2007/12/10 23:04:48 bhatele +3 -3 1.82 src/NamdCentLB.Ctopo placement of proxies is good with the new LDBs alsoonce the spanning tree is moved to the LDBs, we wouldnt need it anymore
2007/12/10 22:06:35 bhatele +44 -14 1.7 src/TorusLB.Cbug fixes
2007/12/10 21:59:57 bhatele +41 -7 1.7 src/RefineTorusLB.Cbug fixes
2007/12/05 22:08:37 jim +1 -0 1.9 src/TclCommands.hDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +1 -0 1.15 src/TclCommands.CDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +1 -0 1.28 src/ScriptTcl.hDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +6 -5 1.61 src/ScriptTcl.CDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +1 -0 1.2 src/Measure.hDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +1 -0 1.7 src/GlobalMasterTcl.hDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +5 -4 1.17 src/GlobalMasterTcl.CDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +1 -0 1.5 src/ComputeTclBC.hDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/12/05 22:08:37 jim +4 -3 1.5 src/ComputeTclBC.CDefine USE_COMPAT_CONST before every include of tcl.h to allow buildingwith either Tcl 8.3 or 8.4. Treat strings returned from Tcl as const.
2007/11/30 15:34:10 kunzman +16 -2 1.1047 src/common.hAdded comments to atom sort defines. Modified merge sort to try andimprove performance.
2007/11/30 15:34:10 kunzman +145 -0 1.1138 src/ComputeNonbondedBase.hAdded comments to atom sort defines. Modified merge sort to try andimprove performance.
2007/11/27 22:32:48 kunzman +2 -1 1.1046 src/common.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +16 -0 1.1040 src/ProxyPatch.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +13 -0 1.1026 src/ProxyMgr.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +43 -0 1.1057 src/ProxyMgr.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +14 -0 1.1025 src/Patch.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +5 -0 1.1045 src/Patch.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +13 -0 1.1031 src/NamdTypes.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +1 -6 1.1051 src/HomePatch.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +12 -0 1.1148 src/HomePatch.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +6 -0 1.1013 src/ComputePatchPair.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +8 -0 1.1026 src/ComputePatchPair.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +6 -0 1.1013 src/ComputePatch.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +5 -0 1.1025 src/ComputePatch.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +21 -0 1.1058 src/ComputeNonbondedUtil.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:48 kunzman +8 -0 1.5 src/ComputeNonbondedStd.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:47 kunzman +7 -0 1.1033 src/ComputeNonbondedSelf.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:47 kunzman +7 -0 1.1039 src/ComputeNonbondedPair.CModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/27 22:32:47 kunzman +647 -40 1.1137 src/ComputeNonbondedBase.hModified atom separation code to pass the number of water atoms on tothe compute code in ComputeNonbondedBase.h.Modified pairlist generation code to sort atoms along the line betweenthe center of masses for a pair of patches. The sort information isthen used to skip as many distance caclculations as possible to reducepairlist generation time if possible.
2007/11/20 05:04:39 bhatele +14 -2 1.81 src/NamdCentLB.Ccorrect the way hops are calculated
2007/11/12 19:01:41 bhatele +349 -60 1.115 Make.dependsincludes the TorusLB.o and RefineTorusLB.o targets
2007/11/12 17:31:29 dbwells2 +1 -0 1.1134 src/SimParameters.hAdded new transverse force feature to SMD. The user may now optionallyspecify a second spring constant (keyword SMDk2) which will describe aforce perpendicular to the pulling direction. This second springconstant defaults to zero, which reproduces the original SMDbehavior. This feature is particularly useful for restraining thecenter of mass of atoms in all three directions simultaneously(i.e. SMDVel = 0.)
2007/11/12 17:31:29 dbwells2 +6 -0 1.1245 src/SimParameters.CAdded new transverse force feature to SMD. The user may now optionallyspecify a second spring constant (keyword SMDk2) which will describe aforce perpendicular to the pulling direction. This second springconstant defaults to zero, which reproduces the original SMDbehavior. This feature is particularly useful for restraining thecenter of mass of atoms in all three directions simultaneously(i.e. SMDVel = 0.)
2007/11/12 17:31:29 dbwells2 +4 -1 1.3 src/GlobalMasterSMD.hAdded new transverse force feature to SMD. The user may now optionallyspecify a second spring constant (keyword SMDk2) which will describe aforce perpendicular to the pulling direction. This second springconstant defaults to zero, which reproduces the original SMDbehavior. This feature is particularly useful for restraining thecenter of mass of atoms in all three directions simultaneously(i.e. SMDVel = 0.)
2007/11/12 17:31:29 dbwells2 +6 -2 1.7 src/GlobalMasterSMD.CAdded new transverse force feature to SMD. The user may now optionallyspecify a second spring constant (keyword SMDk2) which will describe aforce perpendicular to the pulling direction. This second springconstant defaults to zero, which reproduces the original SMDbehavior. This feature is particularly useful for restraining thecenter of mass of atoms in all three directions simultaneously(i.e. SMDVel = 0.)
2007/11/12 17:31:29 dbwells2 +2 -1 1.1073 src/ComputeMgr.CAdded new transverse force feature to SMD. The user may now optionallyspecify a second spring constant (keyword SMDk2) which will describe aforce perpendicular to the pulling direction. This second springconstant defaults to zero, which reproduces the original SMDbehavior. This feature is particularly useful for restraining thecenter of mass of atoms in all three directions simultaneously(i.e. SMDVel = 0.)
2007/11/12 07:06:25 bhatele +9 -10 1.1176 src/WorkDistrib.Cnow uses TopoManager for RecBisection, this change should hopefullymake topology awareness work on Cray
2007/11/12 07:06:25 bhatele +12 -11 1.23 src/RecBisection.Cnow uses TopoManager for RecBisection, this change should hopefullymake topology awareness work on Cray
2007/11/12 02:37:25 bhatele +5 -5 1.6 src/TorusLB.Ccorrected the brick expansion logic
2007/11/12 02:37:25 bhatele +7 -5 1.6 src/RefineTorusLB.Ccorrected the brick expansion logic
2007/11/10 19:45:58 bhatele +8 -6 1.80 src/NamdCentLB.Csome changes in the dump logs function
2007/11/08 20:55:07 bhatele +7 -7 1.5 src/TorusLB.Cincreasing the overload factor
2007/11/07 21:36:12 bhatele +2 -0 1.1229 Makefileadded targets for TorusLB and RefineTorusLB
2007/11/07 16:43:26 jordi +1 -1 1.11 ug/ug_alchemy.texDocs: Mention that fepElecLambdaStart does not affect PME
2007/11/07 16:42:50 jordi +2 -13 1.112 src/ComputePme.C*** empty log message ***
2007/11/07 16:42:50 jordi +7 -3 1.27 src/ComputeNonbondedBase2.h*** empty log message ***
2007/11/05 20:27:49 bhatele +3 -3 1.5 src/RefineTorusLB.Csmall error
2007/11/05 20:23:49 bhatele +40 -25 1.4 src/RefineTorusLB.Cmore bug fixes
2007/11/05 20:23:48 bhatele +14 -7 1.4 src/TorusLB.Cmore bug fixes
2007/11/05 19:49:16 kunzman +8 -0 1.1045 src/common.hModified HomePatch to separate hydrogen groups (waters vs.non-waters). To enable this separation, set NAMD_SeparateWaters (incommon.h) to a non-zero value.
2007/11/05 19:49:16 kunzman +22 -0 1.1050 src/HomePatch.hModified HomePatch to separate hydrogen groups (waters vs.non-waters). To enable this separation, set NAMD_SeparateWaters (incommon.h) to a non-zero value.
2007/11/05 19:49:16 kunzman +433 -18 1.1147 src/HomePatch.CModified HomePatch to separate hydrogen groups (waters vs.non-waters). To enable this separation, set NAMD_SeparateWaters (incommon.h) to a non-zero value.
2007/11/04 20:22:52 johns +52 -46 1.10 src/Settle.CConverted all of the SSE code to intrinsics, as they are availablein new versions of GNU gcc, and Microsoft's compilers, whereas theIntel SIMD vector classes are not. The intrinsics are also more portablethan inline assembly, since there are several dialects of inlineasm blocks which differ between Intel, GNU, and Microsoft.The new intrinsics-based code should generate identical assemblylanguage to the previous class-based code. There may be a coupleof places where we can eliminate some redundant conversion instructionssince the intrinsics expose a few more details than the classes didand map more directly to the actual machine instructions.
2007/11/04 20:22:52 johns +34 -34 1.1136 src/ComputeNonbondedBase.hConverted all of the SSE code to intrinsics, as they are availablein new versions of GNU gcc, and Microsoft's compilers, whereas theIntel SIMD vector classes are not. The intrinsics are also more portablethan inline assembly, since there are several dialects of inlineasm blocks which differ between Intel, GNU, and Microsoft.The new intrinsics-based code should generate identical assemblylanguage to the previous class-based code. There may be a coupleof places where we can eliminate some redundant conversion instructionssince the intrinsics expose a few more details than the classes didand map more directly to the actual machine instructions.
2007/11/02 22:26:38 jim +1 -1 1.1039 src/WorkDistrib.hEnsure that basednodes and PME nodes do not overlap if possible. Onlyworks with bit-reversal assignment (i.e., not with Blue Gene torus mapper).
2007/11/02 22:26:38 jim +1 -0 1.1029 src/PatchMap.hEnsure that basednodes and PME nodes do not overlap if possible. Onlyworks with bit-reversal assignment (i.e., not with Blue Gene torus mapper).
2007/11/02 22:26:38 jim +4 -0 1.1045 src/PatchMap.CEnsure that basednodes and PME nodes do not overlap if possible. Onlyworks with bit-reversal assignment (i.e., not with Blue Gene torus mapper).
2007/11/02 22:26:37 jim +12 -11 1.1175 src/WorkDistrib.CEnsure that basednodes and PME nodes do not overlap if possible. Onlyworks with bit-reversal assignment (i.e., not with Blue Gene torus mapper).
2007/11/02 22:24:17 jordi +2 -2 1.1244 src/SimParameters.CFEP: Responded to feedback suggesting that it's worth breaking with thepast and set the default parameters to something that will work, ratherthan trying to maintain compatibility with the old behavior (which didn'twork well anyways). fepElecLambdaStart and fepVdwLambdaEnd now bothdefault to 0.5. This is not always optimal, but at least it's neverterrible!
2007/11/02 22:24:14 jordi +2 -2 1.10 ug/ug_alchemy.texFEP: Responded to feedback suggesting that it's worth breaking with thepast and set the default parameters to something that will work, ratherthan trying to maintain compatibility with the old behavior (which didn'twork well anyways). fepElecLambdaStart and fepVdwLambdaEnd now bothdefault to 0.5. This is not always optimal, but at least it's neverterrible!
2007/11/02 18:36:32 jordi +13 -2 1.111 src/ComputePme.CFEP: FEP electrostatics clamping (through the fepElecLambdaStartoption) now supports PME.
2007/11/02 04:19:01 johns +4 -4 1.1135 src/ComputeNonbondedBase.hMade the SSE pairlist code use 16-byte-aligned stores, which areslightly faster.
2007/11/01 21:40:35 bhatele +53 -27 1.3 src/TorusLB.C1. more bug fixes2. additions so that the code can run on non-topo machines
2007/11/01 21:40:35 bhatele +70 -51 1.3 src/RefineTorusLB.C1. more bug fixes2. additions so that the code can run on non-topo machines
2007/11/01 21:39:41 bhatele +4 -4 1.79 src/NamdCentLB.Cwe dont need to use the topomap requiredProxiesOnGrid for the new ldbs
2007/11/01 21:37:53 bhatele +4 -2 1.28 src/Rebalancer.hadded a new function withinBrick
2007/11/01 21:37:53 bhatele +25 -3 1.75 src/Rebalancer.Cadded a new function withinBrick
2007/11/01 18:41:40 bhatele +32 -0 1.11 src/NamdCentLB.hcomments
2007/11/01 18:38:59 bhatele +15 -4 1.27 src/Rebalancer.h1. changed the way how LDB_DEBUG is used2. added a new flag USE_NEW_LDBS3. added comments and headers
2007/11/01 18:38:59 bhatele +14 -7 1.74 src/Rebalancer.C1. changed the way how LDB_DEBUG is used2. added a new flag USE_NEW_LDBS3. added comments and headers
2007/11/01 18:38:59 bhatele +43 -17 1.78 src/NamdCentLB.C1. changed the way how LDB_DEBUG is used2. added a new flag USE_NEW_LDBS3. added comments and headers
2007/11/01 17:42:35 bhatele +14 -7 1.1174 src/WorkDistrib.Cfixed the usage of USE_TOPOMAP flag
2007/11/01 17:42:35 bhatele +8 -4 1.26 src/Rebalancer.hfixed the usage of USE_TOPOMAP flag
2007/11/01 17:42:35 bhatele +8 -1 1.1024 src/PatchMgr.hfixed the usage of USE_TOPOMAP flag
2007/11/01 17:39:53 bhatele +8 -1 1.55 src/AlgSeven.Cfixed the USE_TOPOMAP flag usage
2007/11/01 06:36:25 johns +74 -11 1.9 src/Settle.CAdded more SSE to the Settle implementation, speeding it up by another10 percent or more. Impact on the full NAMD folding run is minor butbeneficial.
2007/10/31 17:46:28 jordi +4 -4 1.1134 src/ComputeNonbondedBase.hFEP: Fixed a small mismatch when using fepVdwLambdaEnd that would causevdw scaling to not be completely confined in its stated range of lambda.
2007/10/30 17:46:23 dbwells2 +3 -1 1.114 Make.dependsFixed Make.depends for dependence on GridForceGrid.h.
2007/10/30 17:44:40 dbwells2 +3 -1 1.1137 src/Molecule.CFixed missing header includes.
2007/10/30 17:44:40 dbwells2 +2 -1 1.8 src/ComputeGridForce.CFixed missing header includes.
2007/10/30 17:00:23 jordi +20 -13 1.9 ug/ug_alchemy.texDocs: Added docs for fepVdwLambdaEnd and fepElecLambdaStart
2007/10/30 16:59:21 jordi +2 -2 1.1133 src/SimParameters.hFEP: Added fepVdwLambdaEnd and fepElecLambdaStart to control when toscale vdw and electrostatics. This very significantly improves theresults of FEP calcuclations when charges are involved. Also, removedfepVdwScaleExp (which was not that useful in retrospect)
2007/10/30 16:59:21 jordi +11 -4 1.1243 src/SimParameters.CFEP: Added fepVdwLambdaEnd and fepElecLambdaStart to control when toscale vdw and electrostatics. This very significantly improves theresults of FEP calcuclations when charges are involved. Also, removedfepVdwScaleExp (which was not that useful in retrospect)
2007/10/30 16:59:21 jordi +0 -1 1.5 src/DumpBenchParams.hFEP: Added fepVdwLambdaEnd and fepElecLambdaStart to control when toscale vdw and electrostatics. This very significantly improves theresults of FEP calcuclations when charges are involved. Also, removedfepVdwScaleExp (which was not that useful in retrospect)
2007/10/30 16:59:20 jordi +2 -1 1.1057 src/ComputeNonbondedUtil.hFEP: Added fepVdwLambdaEnd and fepElecLambdaStart to control when toscale vdw and electrostatics. This very significantly improves theresults of FEP calcuclations when charges are involved. Also, removedfepVdwScaleExp (which was not that useful in retrospect)
2007/10/30 16:59:20 jordi +12 -10 1.1089 src/ComputeNonbondedUtil.CFEP: Added fepVdwLambdaEnd and fepElecLambdaStart to control when toscale vdw and electrostatics. This very significantly improves theresults of FEP calcuclations when charges are involved. Also, removedfepVdwScaleExp (which was not that useful in retrospect)
2007/10/30 16:59:20 jordi +39 -31 1.26 src/ComputeNonbondedBase2.hFEP: Added fepVdwLambdaEnd and fepElecLambdaStart to control when toscale vdw and electrostatics. This very significantly improves theresults of FEP calcuclations when charges are involved. Also, removedfepVdwScaleExp (which was not that useful in retrospect)
2007/10/30 16:59:20 jordi +20 -5 1.1133 src/ComputeNonbondedBase.hFEP: Added fepVdwLambdaEnd and fepElecLambdaStart to control when toscale vdw and electrostatics. This very significantly improves theresults of FEP calcuclations when charges are involved. Also, removedfepVdwScaleExp (which was not that useful in retrospect)
2007/10/30 05:51:36 dbwells2 +14 -8 1.1136 src/Molecule.CHad deleted calls to distribute Gridforce parameters to nodes... oops. Now repaired.
2007/10/30 05:51:36 dbwells2 +19 -1 1.2 src/GridForceGrid.CHad deleted calls to distribute Gridforce parameters to nodes... oops. Now repaired.
2007/10/30 05:51:36 dbwells2 +1 -1 1.7 src/ComputeGridForce.CHad deleted calls to distribute Gridforce parameters to nodes... oops. Now repaired.
2007/10/30 04:24:01 johns +3 -2 1.1132 src/ComputeNonbondedBase.hprettied up the code a tad
2007/10/30 04:17:14 johns +64 -1 1.1131 src/ComputeNonbondedBase.hAdded another SSE accelerated pairlist loop where we already had anexisting optimized loop. The remaining loops do not appear to beviewed as performance-critical based on their appearance, so furtherSSE pairlist acceleration will require consultation with Jim andmore profiling.
2007/10/30 03:56:53 johns +61 -2 1.1130 src/ComputeNonbondedBase.hAdded an Intel SSE pairlist loop directly based on the existingpairlist loop that was already made to be friendly to the IBM double-hummer.This initial implementation gives a nice little speed boost, but candefinitely be improved by using SSE conditionals and other niceties thatthe present code doesn't bother with.
2007/10/30 02:25:57 dbwells2 +1 -0 1.1228 MakefileUpdated Makefile and Make.depends for the addition ofGridForceGrid.[Ch]
2007/10/30 02:25:57 dbwells2 +11 -0 1.113 Make.dependsUpdated Makefile and Make.depends for the addition ofGridForceGrid.[Ch]
2007/10/30 02:24:47 dbwells2 +1 -1 1.1242 src/SimParameters.C* Drastic structural changes to Gridforce code. There is now a'GridforceGrid' object which includes essentially all the informationthat was previously a 'gridfrc...' variable in the Molecule object.This cleans up Molecule a lot, and more importantly paves the wayfor multiple grids.* Also fixed a fairly major Gridforce bug in which the grid wasshifted slightly from the correct position. This is because the gridorigin was not corrected when adding border grid points.
2007/10/30 02:24:47 dbwells2 +4 -44 1.1070 src/Molecule.h* Drastic structural changes to Gridforce code. There is now a'GridforceGrid' object which includes essentially all the informationthat was previously a 'gridfrc...' variable in the Molecule object.This cleans up Molecule a lot, and more importantly paves the wayfor multiple grids.* Also fixed a fairly major Gridforce bug in which the grid wasshifted slightly from the correct position. This is because the gridorigin was not corrected when adding border grid points.
2007/10/30 02:24:47 dbwells2 +14 -515 1.1135 src/Molecule.C* Drastic structural changes to Gridforce code. There is now a'GridforceGrid' object which includes essentially all the informationthat was previously a 'gridfrc...' variable in the Molecule object.This cleans up Molecule a lot, and more importantly paves the wayfor multiple grids.* Also fixed a fairly major Gridforce bug in which the grid wasshifted slightly from the correct position. This is because the gridorigin was not corrected when adding border grid points.
2007/10/30 02:24:47 dbwells2 None 1.1 src/GridForceGrid.h* Drastic structural changes to Gridforce code. There is now a'GridforceGrid' object which includes essentially all the informationthat was previously a 'gridfrc...' variable in the Molecule object.This cleans up Molecule a lot, and more importantly paves the wayfor multiple grids.* Also fixed a fairly major Gridforce bug in which the grid wasshifted slightly from the correct position. This is because the gridorigin was not corrected when adding border grid points.
2007/10/30 02:24:47 dbwells2 None 1.1 src/GridForceGrid.C* Drastic structural changes to Gridforce code. There is now a'GridforceGrid' object which includes essentially all the informationthat was previously a 'gridfrc...' variable in the Molecule object.This cleans up Molecule a lot, and more importantly paves the wayfor multiple grids.* Also fixed a fairly major Gridforce bug in which the grid wasshifted slightly from the correct position. This is because the gridorigin was not corrected when adding border grid points.
2007/10/30 02:24:46 dbwells2 +159 -163 1.6 src/ComputeGridForce.C* Drastic structural changes to Gridforce code. There is now a'GridforceGrid' object which includes essentially all the informationthat was previously a 'gridfrc...' variable in the Molecule object.This cleans up Molecule a lot, and more importantly paves the wayfor multiple grids.* Also fixed a fairly major Gridforce bug in which the grid wasshifted slightly from the correct position. This is because the gridorigin was not corrected when adding border grid points.
2007/10/29 20:41:27 johns +66 -9 1.8 src/Settle.CAdded an SSE2-accelerated version of some of the costlier components ofthe settle algorithm. The new code is currently only enabled when compiledwith the macros NAMD_SSE, __INTEL_COMPILER, __SSE2__ all defined.If any of these are not defined, the plain C code path is used.The current code uses the Intel Class library, but could be convertedto intrinsics or even inline assembler, though the existing class-basedcode is likely the most readable.
2007/10/29 05:01:49 johns +10 -6 1.7 src/Settle.CEliminated another redundant multiply, and identified three areas wherewe're doing math that ends up getting thrown away. This was previouslyobscured by the fact that we used the vector classes that make the codeeasy to write, but also obscure when you're using vector elements or not.A subsequent patch should eliminate 15 FLOPS of wasted arithmetic per water.
2007/10/29 04:11:26 bhatele +31 -110 1.2 src/TorusLB.Csome bug fixes for the new ldb's
2007/10/29 04:11:26 bhatele +51 -117 1.2 src/RefineTorusLB.Csome bug fixes for the new ldb's
2007/10/29 04:11:26 bhatele +1 -0 1.25 src/Rebalancer.hsome bug fixes for the new ldb's
2007/10/29 04:11:25 bhatele +17 -0 1.73 src/Rebalancer.Csome bug fixes for the new ldb's
2007/10/26 21:50:54 johns +9 -4 1.4 src/Settle.hSplit the initialization logic for cached water atom properties into a newsettle1init() function, added a settle1isinitted() query, and eliminatedruntime initialization tests from settle1().Eliminated redundant arithmetic in settlec() and inlined the wholefunction into settle1() since it is not used elsewhere.The new code avoids expensive divisions where possible, precalculatesa couple of additional quantities, but makes no assumptions that theoriginal code wasn't already making. The settle1init() routine iscurrently called on entry into rattle1() if it has not already beeninitialized. The initialization will move to Molecule::build_atom_statusonce that version is debugged.
2007/10/26 21:50:54 johns +48 -53 1.6 src/Settle.CSplit the initialization logic for cached water atom properties into a newsettle1init() function, added a settle1isinitted() query, and eliminatedruntime initialization tests from settle1().Eliminated redundant arithmetic in settlec() and inlined the wholefunction into settle1() since it is not used elsewhere.The new code avoids expensive divisions where possible, precalculatesa couple of additional quantities, but makes no assumptions that theoriginal code wasn't already making. The settle1init() routine iscurrently called on entry into rattle1() if it has not already beeninitialized. The initialization will move to Molecule::build_atom_statusonce that version is debugged.
2007/10/26 21:50:54 johns +20 -3 1.1146 src/HomePatch.CSplit the initialization logic for cached water atom properties into a newsettle1init() function, added a settle1isinitted() query, and eliminatedruntime initialization tests from settle1().Eliminated redundant arithmetic in settlec() and inlined the wholefunction into settle1() since it is not used elsewhere.The new code avoids expensive divisions where possible, precalculatesa couple of additional quantities, but makes no assumptions that theoriginal code wasn't already making. The settle1init() routine iscurrently called on entry into rattle1() if it has not already beeninitialized. The initialization will move to Molecule::build_atom_statusonce that version is debugged.
2007/10/26 15:45:06 johns +1 -1 1.3 src/Settle.hReplaced floating point divides in settle1() with multiplies with invdt
2007/10/26 15:45:06 johns +6 -5 1.5 src/Settle.CReplaced floating point divides in settle1() with multiplies with invdt
2007/10/26 15:45:06 johns +1 -1 1.1145 src/HomePatch.CReplaced floating point divides in settle1() with multiplies with invdt
2007/10/26 15:32:17 johns +13 -10 1.1144 src/HomePatch.CReplaced slow floating point divides by dt with multiplies by precalculatedinvdt. Changed the per-water-atom fixed atom checks with a single test ofgroupFixed.
2007/10/25 21:20:49 johns +5 -0 1.16 src/vmdsock.CAdded comments warning about SMP-unsafe global and static variables.
2007/10/25 21:17:57 johns +4 -0 1.1217 src/Controller.CAdded comments warning about SMP-unsafe global and static variables.Signal handlers are also potentially thread-unsafe, though this isextremely platform dependent.
2007/10/25 21:14:10 johns +4 -0 1.8 src/GlobalMasterIMD.CAdded comments warning about SMP-unsafe global and static variables.
2007/10/25 21:12:47 johns +2 -2 1.82 src/LdbCoordinator.CAdded comments warning about SMP-unsafe global and static variables.
2007/10/25 21:06:42 johns +5 -0 1.5 src/pub3dfft.CAdded comments warning about SMP-unsafe global and static variables.The pub3dfft routines cannot be safely used for multithreaded NAMD builds.
2007/10/25 21:05:12 johns +4 -0 1.60 src/ScriptTcl.CAdded comments warning about SMP-unsafe global and static variables.
2007/10/25 21:02:57 johns +7 -0 1.4 src/Settle.CAdded comments warning about SMP-unsafe global and static variables.
2007/10/25 20:57:27 johns +6 -1 1.1056 src/ProxyMgr.CAdded comments warning about the global and static variablesreferenced in the proxy manager code.
2007/10/25 20:41:20 chaomei2 +1 -0 1.12 src/CompressPsf.CAdded an output showing the number of crossterm signatures.
2007/10/25 20:04:47 chaomei2 +36 -1 1.1134 src/Molecule.CAdded the implementation of compressing crossterms and reading crossterms from the compressed psf file.Now the crossterm calculation is supported in the memory optimized version.
2007/10/25 20:04:47 chaomei2 +153 -5 1.11 src/CompressPsf.CAdded the implementation of compressing crossterms and reading crossterms from the compressed psf file.Now the crossterm calculation is supported in the memory optimized version.
2007/10/25 20:03:13 chaomei2 +0 -4 1.1014 src/structures.hRemoved the comments about signatures about crossterms and exclusions since they are no longer true. Crossterms are now supported in the memory optimized version
2007/10/25 18:31:45 chaomei2 +11 -1 1.10 src/CompressPsf.CDo abortion if explicit exclusions and crossterms are encountered in the psf file.The support will be implemented later.
2007/10/18 21:29:09 chaomei2 +5 -5 1.1227 MakefileChanged the way to compile for tracing compute objects
2007/10/16 01:13:25 jim +92 -70 1.110 src/ComputePme.CReworked pencil placement to use bit reversal and patch locationsfollowed by round-robin to avoid sending and receiving in the samephase as much as possible. Only uses node 0 if it has a patch oris very likely to improve performance.
2007/10/15 20:40:14 jim +34 -2 1.20 src/ComputePmeMgr.ciEliminate replies to empty pencil PME messages.
2007/10/15 20:40:14 jim +86 -12 1.109 src/ComputePme.CEliminate replies to empty pencil PME messages.
2007/10/15 19:26:03 chaomei2 +25 -3 1.1226 MakefileAdded compiling target for tracing compute objects
2007/10/15 19:23:58 chaomei2 +5 -0 1.1083 src/Node.CAdded user events for every compute object's work. The user event name of every compute object is "type_computeId". This is useful in two aspects:1. Track the execution time of every compute object.2. In the timeline view in projections (the performance analysis tool), a timeline event will be associated with such a user event which shows the compute object that is directly related with this timeline event.To enable tracing compute objects, one has to type "make tracecomputes CXXOPTS=-DTRACE_COMPUTE_OBJECTS $(CXXOPTS)" to generate the namd binary.
2007/10/15 19:23:58 chaomei2 +8 -0 1.16 src/ComputeSelfTuples.hAdded user events for every compute object's work. The user event name of every compute object is "type_computeId". This is useful in two aspects:1. Track the execution time of every compute object.2. In the timeline view in projections (the performance analysis tool), a timeline event will be associated with such a user event which shows the compute object that is directly related with this timeline event.To enable tracing compute objects, one has to type "make tracecomputes CXXOPTS=-DTRACE_COMPUTE_OBJECTS $(CXXOPTS)" to generate the namd binary.
2007/10/15 19:23:58 chaomei2 +8 -0 1.108 src/ComputePme.CAdded user events for every compute object's work. The user event name of every compute object is "type_computeId". This is useful in two aspects:1. Track the execution time of every compute object.2. In the timeline view in projections (the performance analysis tool), a timeline event will be associated with such a user event which shows the compute object that is directly related with this timeline event.To enable tracing compute objects, one has to type "make tracecomputes CXXOPTS=-DTRACE_COMPUTE_OBJECTS $(CXXOPTS)" to generate the namd binary.
2007/10/15 19:23:58 chaomei2 +8 -0 1.1032 src/ComputeNonbondedSelf.CAdded user events for every compute object's work. The user event name of every compute object is "type_computeId". This is useful in two aspects:1. Track the execution time of every compute object.2. In the timeline view in projections (the performance analysis tool), a timeline event will be associated with such a user event which shows the compute object that is directly related with this timeline event.To enable tracing compute objects, one has to type "make tracecomputes CXXOPTS=-DTRACE_COMPUTE_OBJECTS $(CXXOPTS)" to generate the namd binary.
2007/10/15 19:23:58 chaomei2 +8 -0 1.1038 src/ComputeNonbondedPair.CAdded user events for every compute object's work. The user event name of every compute object is "type_computeId". This is useful in two aspects:1. Track the execution time of every compute object.2. In the timeline view in projections (the performance analysis tool), a timeline event will be associated with such a user event which shows the compute object that is directly related with this timeline event.To enable tracing compute objects, one has to type "make tracecomputes CXXOPTS=-DTRACE_COMPUTE_OBJECTS $(CXXOPTS)" to generate the namd binary.
2007/10/15 19:23:58 chaomei2 +567 -408 1.1072 src/ComputeMgr.CAdded user events for every compute object's work. The user event name of every compute object is "type_computeId". This is useful in two aspects:1. Track the execution time of every compute object.2. In the timeline view in projections (the performance analysis tool), a timeline event will be associated with such a user event which shows the compute object that is directly related with this timeline event.To enable tracing compute objects, one has to type "make tracecomputes CXXOPTS=-DTRACE_COMPUTE_OBJECTS $(CXXOPTS)" to generate the namd binary.
2007/10/15 19:23:58 chaomei2 +7 -0 1.1018 src/Compute.hAdded user events for every compute object's work. The user event name of every compute object is "type_computeId". This is useful in two aspects:1. Track the execution time of every compute object.2. In the timeline view in projections (the performance analysis tool), a timeline event will be associated with such a user event which shows the compute object that is directly related with this timeline event.To enable tracing compute objects, one has to type "make tracecomputes CXXOPTS=-DTRACE_COMPUTE_OBJECTS $(CXXOPTS)" to generate the namd binary.
2007/10/14 16:42:24 jim +4 -12 1.19 src/ComputePmeMgr.cisimple cleanups
2007/10/14 16:42:24 jim +10 -8 1.107 src/ComputePme.Csimple cleanups
2007/10/11 20:59:31 jim +3 -3 1.1173 src/WorkDistrib.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +7 -8 1.1039 src/ProxyPatch.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +3 -2 1.1055 src/ProxyMgr.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +43 -63 1.5 src/Priorities.hCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +6 -7 1.1143 src/HomePatch.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +2 -2 1.15 src/ComputeSelfTuples.hCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +39 -34 1.106 src/ComputePme.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +14 -14 1.1025 src/ComputePatchPair.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +3 -2 1.1024 src/ComputePatch.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +2 -1 1.1047 src/ComputeHomeTuples.hCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +1 -1 1.1015 src/ComputeHomePatches.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +3 -2 1.6 src/ComputeHomePatch.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/11 20:59:31 jim +3 -2 1.1029 src/Compute.CCompletely revamped message priority system. New file Priorities.hcontains high-level priorities and macros for prioritizing messages.
2007/10/10 23:30:50 jim +20 -20 1.18 src/ComputePmeMgr.ciremoved [expedited] to allow prioritization of messages
2007/10/10 23:29:41 jim +17 -7 1.105 src/ComputePme.Cprioritize pencil pme messages
2007/10/09 22:36:53 dbwells2 +3 -1 1.1069 src/Molecule.hUpdated Gridforces:* Fixed way in which continuous grids are handled.* Fixed handling of grids whose unit vectors are not mutuallyorthogonal and/or not parallel to those of the periodic cell.
2007/10/09 22:36:53 dbwells2 +73 -64 1.1133 src/Molecule.CUpdated Gridforces:* Fixed way in which continuous grids are handled.* Fixed handling of grids whose unit vectors are not mutuallyorthogonal and/or not parallel to those of the periodic cell.
2007/10/09 22:36:53 dbwells2 +5 -3 1.5 src/ComputeGridForce.CUpdated Gridforces:* Fixed way in which continuous grids are handled.* Fixed handling of grids whose unit vectors are not mutuallyorthogonal and/or not parallel to those of the periodic cell.
2007/10/08 03:38:55 jim +11 -6 1.104 src/ComputePme.CPut PME grid and transpose slabs on different processors.
2007/10/07 21:12:20 jim +19 -3 1.103 src/ComputePme.CPrioritize slab-based PME messages.
2007/10/07 20:05:31 jordi +3 -3 1.8 ug/ug_alchemy.textypos
2007/10/07 20:03:31 jordi +26 -0 1.7 ug/ug_alchemy.texFEP docs: Added docs for the new fepVdWShiftCoeff and fepVdwScaleExpparameters
2007/10/07 01:03:34 jordi +17 -11 1.1132 src/SimParameters.hTowards a working FEP...Implemented a new FEP vdW scaling function that removes the end-pointcatastrophe that used to happen a lot in NAMD. The vdw potential hasthe form: V(r) = A/r^12 - B/r^6The new scaled V looks like this: V'(r) = lambda A'/r'^12 - B'/r'^6where: A' = lambda^2n * A B' = lambda^n * B r'^2 = r^2 + shift * (1.-lambda)This form scales well and gives accurate and reproducibleresults for tested FEP calculations. Velocity crashes are now almostnon-existent (if Coulomb forces are not significant).Also added two new NAMD config file options: fepVdwShiftCoeff: sets value "shift" [default 5, range >= 0] fepVdwScaleExp: sets the exponent "n" [default 0]Values one can use: shift = 0, n=0: old NAMD way (crashes a lot and bad convergence) shift = 5, n=0: used in CHARMM shift = 1, n=2: this scales quite nicely as well
2007/10/07 01:03:34 jordi +12 -2 1.1241 src/SimParameters.CTowards a working FEP...Implemented a new FEP vdW scaling function that removes the end-pointcatastrophe that used to happen a lot in NAMD. The vdw potential hasthe form: V(r) = A/r^12 - B/r^6The new scaled V looks like this: V'(r) = lambda A'/r'^12 - B'/r'^6where: A' = lambda^2n * A B' = lambda^n * B r'^2 = r^2 + shift * (1.-lambda)This form scales well and gives accurate and reproducibleresults for tested FEP calculations. Velocity crashes are now almostnon-existent (if Coulomb forces are not significant).Also added two new NAMD config file options: fepVdwShiftCoeff: sets value "shift" [default 5, range >= 0] fepVdwScaleExp: sets the exponent "n" [default 0]Values one can use: shift = 0, n=0: old NAMD way (crashes a lot and bad convergence) shift = 5, n=0: used in CHARMM shift = 1, n=2: this scales quite nicely as well
2007/10/07 01:03:34 jordi +3 -1 1.4 src/DumpBenchParams.hTowards a working FEP...Implemented a new FEP vdW scaling function that removes the end-pointcatastrophe that used to happen a lot in NAMD. The vdw potential hasthe form: V(r) = A/r^12 - B/r^6The new scaled V looks like this: V'(r) = lambda A'/r'^12 - B'/r'^6where: A' = lambda^2n * A B' = lambda^n * B r'^2 = r^2 + shift * (1.-lambda)This form scales well and gives accurate and reproducibleresults for tested FEP calculations. Velocity crashes are now almostnon-existent (if Coulomb forces are not significant).Also added two new NAMD config file options: fepVdwShiftCoeff: sets value "shift" [default 5, range >= 0] fepVdwScaleExp: sets the exponent "n" [default 0]Values one can use: shift = 0, n=0: old NAMD way (crashes a lot and bad convergence) shift = 5, n=0: used in CHARMM shift = 1, n=2: this scales quite nicely as well
2007/10/07 01:03:34 jordi +2 -0 1.1056 src/ComputeNonbondedUtil.hTowards a working FEP...Implemented a new FEP vdW scaling function that removes the end-pointcatastrophe that used to happen a lot in NAMD. The vdw potential hasthe form: V(r) = A/r^12 - B/r^6The new scaled V looks like this: V'(r) = lambda A'/r'^12 - B'/r'^6where: A' = lambda^2n * A B' = lambda^n * B r'^2 = r^2 + shift * (1.-lambda)This form scales well and gives accurate and reproducibleresults for tested FEP calculations. Velocity crashes are now almostnon-existent (if Coulomb forces are not significant).Also added two new NAMD config file options: fepVdwShiftCoeff: sets value "shift" [default 5, range >= 0] fepVdwScaleExp: sets the exponent "n" [default 0]Values one can use: shift = 0, n=0: old NAMD way (crashes a lot and bad convergence) shift = 5, n=0: used in CHARMM shift = 1, n=2: this scales quite nicely as well
2007/10/07 01:03:34 jordi +5 -1 1.1088 src/ComputeNonbondedUtil.CTowards a working FEP...Implemented a new FEP vdW scaling function that removes the end-pointcatastrophe that used to happen a lot in NAMD. The vdw potential hasthe form: V(r) = A/r^12 - B/r^6The new scaled V looks like this: V'(r) = lambda A'/r'^12 - B'/r'^6where: A' = lambda^2n * A B' = lambda^n * B r'^2 = r^2 + shift * (1.-lambda)This form scales well and gives accurate and reproducibleresults for tested FEP calculations. Velocity crashes are now almostnon-existent (if Coulomb forces are not significant).Also added two new NAMD config file options: fepVdwShiftCoeff: sets value "shift" [default 5, range >= 0] fepVdwScaleExp: sets the exponent "n" [default 0]Values one can use: shift = 0, n=0: old NAMD way (crashes a lot and bad convergence) shift = 5, n=0: used in CHARMM shift = 1, n=2: this scales quite nicely as well
2007/10/07 01:03:34 jordi +11 -11 1.25 src/ComputeNonbondedBase2.hTowards a working FEP...Implemented a new FEP vdW scaling function that removes the end-pointcatastrophe that used to happen a lot in NAMD. The vdw potential hasthe form: V(r) = A/r^12 - B/r^6The new scaled V looks like this: V'(r) = lambda A'/r'^12 - B'/r'^6where: A' = lambda^2n * A B' = lambda^n * B r'^2 = r^2 + shift * (1.-lambda)This form scales well and gives accurate and reproducibleresults for tested FEP calculations. Velocity crashes are now almostnon-existent (if Coulomb forces are not significan