Added a note of next potential steps
This commit is contained in:
25
Pipe1D.cs
25
Pipe1D.cs
@@ -183,3 +183,28 @@ namespace FluidSimV2
|
|||||||
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
/*
|
||||||
|
Next potential steps:
|
||||||
|
|
||||||
|
Implement a flux method (Lax Friedrichs)
|
||||||
|
|
||||||
|
Dont update ghost cells in UpdateDerivedValues and UpdateCells
|
||||||
|
|
||||||
|
Ghost cells boundary conditions
|
||||||
|
|
||||||
|
Scrap the arrays _u, _P, _c - instead, compute on the fly (half memory traffic)
|
||||||
|
|
||||||
|
Consider System.Runtime.Intrinsics for finer SIMD tuning
|
||||||
|
|
||||||
|
Fuse UpdateDerivedValues and UpdateCells
|
||||||
|
|
||||||
|
CFL based sub-stepping
|
||||||
|
|
||||||
|
Vectorise the flux calculation
|
||||||
|
|
||||||
|
Process multiple components together (SIMD - complicated)
|
||||||
|
|
||||||
|
Potential multithreading of the SIMD loops
|
||||||
|
|
||||||
|
*/
|
||||||
Reference in New Issue
Block a user