Added a skeleton function for updating an arbitrary block inside the computational domain instead of the whole mesh
This commit is contained in:
@@ -269,6 +269,20 @@ acIntegrateStep(const int& isubstep, const AcReal& dt)
|
||||
return AC_SUCCESS;
|
||||
}
|
||||
|
||||
AcResult
|
||||
acIntegrateStepWithOffset(const int& isubstep, const AcReal& dt, const int3& start, const int3& end)
|
||||
{
|
||||
/*
|
||||
// A skeleton function for computing integrations with arbitrary subblocks
|
||||
// Uncommenting the following should work with a single GPU.
|
||||
const int3 start = (int3){STENCIL_ORDER / 2, STENCIL_ORDER / 2, STENCIL_ORDER / 2};
|
||||
const int3 end = (int3){STENCIL_ORDER / 2 + subgrid.n.x, STENCIL_ORDER / 2 + subgrid.n.y,
|
||||
STENCIL_ORDER / 2 + subgrid.n.z};
|
||||
rkStep(devices[0], STREAM_PRIMARY, isubstep, start, end, dt);
|
||||
*/
|
||||
return AC_FAILURE;
|
||||
}
|
||||
|
||||
AcResult
|
||||
acBoundcondStep(void)
|
||||
{
|
||||
|
Reference in New Issue
Block a user