Can I solve Ax=b1,b2,b3 repeatedly on GPU?

[ztdepztdep]

I need to solve a series of Ax=bi,i=1.....N with the matrix A unchanged.
while(time<FianalTime) { solve Ax=bi; set bi; if(time=SaveTime) copy x to cpu for further treating }
I want to know can i set the bi directly on the GPU without copying data between cpu and gpu.

[upsj]

Yes, what you are describing is possible directly! All data that is allocated on the GPU can be accessed via the corresponding data pointers vector->get_values(). So depending on the complexity of the operations you want to do on b_i, you can write your own CUDA kernels, build the operations out of thrust algorithms or (probably soon) use our common kernel interface to implement simple pointwise kernels in a device-independent fashion.

Let us know if you want some details on any of the 3 approaches. What kinds of calculations do you need to do to compute the new b_i?

[ztdepztdep]

ok, I am solving Navier-Stokes equations, the bi is updated by computing the gradient, and divergence of the velocity vector fields and it is very suited to write a CUDA kernels by myself and transfer the array to the library.