Modified thread counts that were causing failures on lower end boards

plugins/amoeba/platforms/cuda/src/kernels/amoebaCudaGpu.cpp

@@ -1417,7 +1417,7 @@ void gpuMutualInducedFieldAllocate( amoebaGpuContext amoebaGpu )
     amoebaGpu->amoebaSim.pInducedDipolePolar   = amoebaGpu->psInducedDipolePolar->_pDevData;
 
     amoebaGpu->psCurrentEpsilon                = new CUDAStream<float>(5, 1, "CurrentEpsilon");
-    amoebaGpu->epsilonThreadsPerBlock          = 384;
+    amoebaGpu->epsilonThreadsPerBlock          =  amoebaGpu->gpuContext->sim.threads_per_block;
  
     amoebaGpu->psPolarizability                = new CUDAStream<float>(paddedNumberOfAtoms*3, 1, "Polarizability");
     unsigned int offset                        = paddedNumberOfAtoms*3*sizeof( float );

plugins/amoeba/platforms/cuda/src/kernels/kCalculateAmoebaCudaUtilities.cu

@@ -142,14 +142,14 @@ void kClearFields_3( amoebaGpuContext amoebaGpu, unsigned int numberToClear )
     LAUNCHERROR("kClearFields_3_1");
 
     if( numberToClear > 1 ){
-        kClearFields_kernel<<<gpu->sim.nonbond_blocks, 384>>>( gpu->sim.paddedNumberOfAtoms*3*gpu->sim.outputBuffers, amoebaGpu->psWorkArray_3_2->_pDevData );
+        kClearFields_kernel<<<gpu->sim.nonbond_blocks, gpu->sim.threads_per_block>>>( gpu->sim.paddedNumberOfAtoms*3*gpu->sim.outputBuffers, amoebaGpu->psWorkArray_3_2->_pDevData );
         LAUNCHERROR("kClearFields_3_2");
     } else {
         return;
     }
 
     if( numberToClear > 2 ){
-        kClearFields_kernel<<<gpu->sim.nonbond_blocks, 384>>>( gpu->sim.paddedNumberOfAtoms*3*gpu->sim.outputBuffers,
+        kClearFields_kernel<<<gpu->sim.nonbond_blocks, gpu->sim.threads_per_block>>>( gpu->sim.paddedNumberOfAtoms*3*gpu->sim.outputBuffers,
                                                             amoebaGpu->psWorkArray_3_3->_pDevData );
         LAUNCHERROR("kClearFields_3_3");
     } else {
@@ -157,7 +157,7 @@ void kClearFields_3( amoebaGpuContext amoebaGpu, unsigned int numberToClear )
     }
 
     if( numberToClear > 3 ){
-        kClearFields_kernel<<<gpu->sim.nonbond_blocks, 384>>>( gpu->sim.paddedNumberOfAtoms*3*gpu->sim.outputBuffers,
+        kClearFields_kernel<<<gpu->sim.nonbond_blocks, gpu->sim.threads_per_block>>>( gpu->sim.paddedNumberOfAtoms*3*gpu->sim.outputBuffers,
                                                                 amoebaGpu->psWorkArray_3_4->_pDevData );
         LAUNCHERROR("kClearFields_3_4");
     }

plugins/amoeba/platforms/cuda/src/kernels/kCalculateAmoebaCudaVdw14_7.cu

@@ -416,7 +416,7 @@ void kCalculateAmoebaVdw14_7NonReduction_kernel( float* inputForce, float4* outp
 
 static void kCalculateAmoebaVdw14_7NonReduction(amoebaGpuContext amoebaGpu, CUDAStream<float>* vdwOutputArray, CUDAStream<float4>* forceOutputArray )
 {
-    kCalculateAmoebaVdw14_7NonReduction_kernel<<<amoebaGpu->gpuContext->sim.blocks, 384>>>(
+    kCalculateAmoebaVdw14_7NonReduction_kernel<<<amoebaGpu->gpuContext->sim.blocks, amoebaGpu->gpuContext->sim.threads_per_block>>>(
                                vdwOutputArray->_pDevData, forceOutputArray->_pDevData );
     LAUNCHERROR("kCalculateAmoebaVdw14_7MonReduction");
 }
@@ -480,7 +480,7 @@ void kCalculateAmoebaVdw14_7CopyCoordinates_kernel( unsigned int bufferLength, f
 
 void kCalculateAmoebaVdw14_7CopyCoordinates( amoebaGpuContext amoebaGpu, CUDAStream<float4>* toCopy, CUDAStream<float4>* copy )
 {
-    kCalculateAmoebaVdw14_7CopyCoordinates_kernel<<<amoebaGpu->gpuContext->blocksPerSM, 384>>>( amoebaGpu->gpuContext->sim.paddedNumberOfAtoms, 
+    kCalculateAmoebaVdw14_7CopyCoordinates_kernel<<<amoebaGpu->gpuContext->blocksPerSM, amoebaGpu->gpuContext->sim.threads_per_block>>>( amoebaGpu->gpuContext->sim.paddedNumberOfAtoms, 
                       toCopy->_pDevData, copy->_pDevData );
     LAUNCHERROR("kCalculateAmoebaVdw14_7CopyCoordinates");
 }