Commit 461f5b04 authored by Tomas Sobotik's avatar Tomas Sobotik
Browse files

More optimalizations

parent eeb5983b

examples/hamilton-jacobi-parallel/tnlParallelEikonalSolver_impl.h

+16 −5
Original line number Diff line number Diff line
@@ -1096,13 +1096,13 @@ void tnlParallelEikonalSolver<SchemeHost, SchemeDevice, Device, double, int>::ru 
	{

		tmp = 0;

		int centreGID = (blockDim.y*blockIdx.y + (blockDim.y>>1))*(blockDim.x*gridDim.x) + blockDim.x*blockIdx.x + (blockDim.x>>1);

		if(this->unusedCell_cuda[centreGID] == 0)

		if(this->unusedCell_cuda[centreGID] == 0 || boundaryCondition == 0)

			tmp = 1;

	}

	__syncthreads();



	if(!tmp && (u[0]*u[l] <= 0.0))

		atomicMax( &tmp, 1);

	/*if(!tmp && (u[0]*u[l] <= 0.0))

		atomicMax( &tmp, 1);*/



	__syncthreads();

	if(tmp !=1)
@@ -1294,8 +1294,9 @@ void /*tnlParallelEikonalSolver<SchemeHost, SchemeDevice, Device, double, int>:: 
			atomicMax(&newSubgridValue, INT_MAX);

			atomicMax(&boundary[boundary_index], 1);

			cudaSolver->work_u_cuda[gid] = u_cmp;

			u=u_cmp;

		}

		__threadfence();

		//__threadfence();

		if(threadIdx.y == 0 && (blockIdx.y != 0)/* && (cudaSolver->currentStep & 1)*/)

		{

			u_cmp = cudaSolver->work_u_cuda[gid - blockDim.x*gridDim.x];
@@ -1622,7 +1623,7 @@ void /*tnlParallelEikonalSolver<SchemeHost, SchemeDevice, Device, double, int>:: 
	int i = blockIdx.y * gridDim.x + blockIdx.x;

	int l = threadIdx.y * blockDim.x + threadIdx.x;



	if(caller->getSubgridValueCUDA(i) != INT_MAX)

	if(caller->getSubgridValueCUDA(i) != INT_MAX && caller->getSubgridValueCUDA(i) >= 0)

	{

		caller->getSubgridCUDA(i,caller, &u[l]);

		int bound = caller->getBoundaryConditionCUDA(i);
@@ -1714,6 +1715,16 @@ void /*tnlParallelEikonalSolver<SchemeHost, SchemeDevice, Device, double, int>:: 
			caller->updateSubgridCUDA(i,caller, &u[l]);

			__syncthreads();

		}

		/*if( bound )

		{

			caller->runSubgridCUDA(15,u,i);

			__syncthreads();

			//caller->insertSubgridCUDA(u[l],i);

			//__syncthreads();

			//caller->getSubgridCUDA(i,caller, &u[l]);

			caller->updateSubgridCUDA(i,caller, &u[l]);

			__syncthreads();

		}*/





		if(l==0)