|
|
| cusolve_cbs (spchol *_C, const dmat *_Up, const dmat *_Vp) |
| |
|
void | chol_solve (Real *out, const Real *in, stream_t &stream) |
| |
| virtual Real | solve (curcell &xout, const curcell &xin, stream_t &stream) |
| | Apply the following: xout=solve(xin)-Up*(Vp'*xin). Notice that, in single precision mode, the error is large due to accumulation.
|
| |
|
|
cusp | Cl |
| |
|
cuimat | Cp |
| |
|
curmat | Up |
| |
|
curmat | Vp |
| |
|
curmat | y |
| |
|
curmat | Vr |
| |
◆ solve()
| Real cusolve_cbs::solve |
( |
curcell & |
xout, |
|
|
const curcell & |
xin, |
|
|
stream_t & |
stream |
|
) |
| |
|
virtual |
Apply the following: xout=solve(xin)-Up*(Vp'*xin). Notice that, in single precision mode, the error is large due to accumulation.
- Parameters
-
- Returns
- Real
Implements cusolve_l.
The documentation for this class was generated from the following files:
- cuda/recon/solve.h
- cuda/recon/solve.cu