smyalygames/FiniteVolumeGPU
1
0
Fork 0
mirror of https://github.com/smyalygames/FiniteVolumeGPU.git synced 2026-01-14 15:48:43 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity

Refactoring - broke 2nd order

Browse Source
This commit is contained in:
André R. Brodtkorb
2018-11-09 11:46:34 +01:00
parent 812ebcc9ba
commit e98ae0a979
21 changed files with 499023 additions and 611 deletions
Download Patch File Download Diff File Expand all files Collapse all files

42
GPUSimulators/cuda/SWE2D_KP07_dimsplit.cu
View File

@@ -121,7 +121,8 @@ __global__ void KP07DimsplitKernel(
float theta_,
int step_,
int step_order_,
int boundary_conditions_,
//Input h^n
float* h0_ptr_, int h0_pitch_,
@@ -147,22 +148,14 @@ __global__ void KP07DimsplitKernel(
//Read into shared memory
readBlock<w, h, gc>( h0_ptr_, h0_pitch_, Q[0], nx_+2, ny_+2);
readBlock<w, h, gc>(hu0_ptr_, hu0_pitch_, Q[1], nx_+2, ny_+2);
readBlock<w, h, gc>(hv0_ptr_, hv0_pitch_, Q[2], nx_+2, ny_+2);
__syncthreads();
//Fix boundary conditions
noFlowBoundary<w, h, gc, 1, 1>(Q[0], nx_, ny_);
noFlowBoundary<w, h, gc, -1, 1>(Q[1], nx_, ny_);
noFlowBoundary<w, h, gc, 1, -1>(Q[2], nx_, ny_);
__syncthreads();
readBlock<w, h, gc, 1, 1>( h0_ptr_, h0_pitch_, Q[0], nx_, ny_, boundary_conditions_);
readBlock<w, h, gc, -1, 1>(hu0_ptr_, hu0_pitch_, Q[1], nx_, ny_, boundary_conditions_);
readBlock<w, h, gc, 1, -1>(hv0_ptr_, hv0_pitch_, Q[2], nx_, ny_, boundary_conditions_);
//Step 0 => evolve x first, then y
if (step_ == 0) {
if (getStep(step_order_) == 0) {
//Compute fluxes along the x axis and evolve
minmodSlopeX<w, h, gc, vars>(Q, Qx, theta_);
__syncthreads();
@@ -171,18 +164,9 @@ __global__ void KP07DimsplitKernel(
evolveF<w, h, gc, vars>(Q, F, dx_, dt_);
__syncthreads();
//Set boundary conditions
noFlowBoundary<w, h, gc, 1, 1>(Q[0], nx_, ny_);
noFlowBoundary<w, h, gc, -1, 1>(Q[1], nx_, ny_);
noFlowBoundary<w, h, gc, 1, -1>(Q[2], nx_, ny_);
__syncthreads();
//Compute fluxes along the y axis and evolve
minmodSlopeY<w, h, gc, vars>(Q, Qx, theta_);
__syncthreads();
computeFluxG(Q, Qx, F, g_, dy_, dt_);
__syncthreads();
evolveG<w, h, gc, vars>(Q, F, dy_, dt_);
@@ -198,12 +182,6 @@ __global__ void KP07DimsplitKernel(
evolveG<w, h, gc, vars>(Q, F, dy_, dt_);
__syncthreads();
//Set boundary conditions
noFlowBoundary<w, h, gc, 1, 1>(Q[0], nx_, ny_);
noFlowBoundary<w, h, gc, -1, 1>(Q[1], nx_, ny_);
noFlowBoundary<w, h, gc, 1, -1>(Q[2], nx_, ny_);
__syncthreads();
//Compute fluxes along the x axis and evolve
minmodSlopeX<w, h, gc, vars>(Q, Qx, theta_);
__syncthreads();
@@ -215,9 +193,11 @@ __global__ void KP07DimsplitKernel(
// Write to main memory for all internal cells
writeBlock<w, h, gc>( h1_ptr_, h1_pitch_, Q[0], nx_, ny_);
writeBlock<w, h, gc>(hu1_ptr_, hu1_pitch_, Q[1], nx_, ny_);
writeBlock<w, h, gc>(hv1_ptr_, hv1_pitch_, Q[2], nx_, ny_);
const int step = getStep(step_order_);
const int order = getOrder(step_order_);
writeBlock<w, h, gc>( h1_ptr_, h1_pitch_, Q[0], nx_, ny_, step, order);
writeBlock<w, h, gc>(hu1_ptr_, hu1_pitch_, Q[1], nx_, ny_, step, order);
writeBlock<w, h, gc>(hv1_ptr_, hv1_pitch_, Q[2], nx_, ny_, step, order);
}
} // extern "C"