# Sparse Array Solver on GPU

**URL:** <https://discourse.julialang.org/t/sparse-array-solver-on-gpu/25780>\
**Category:** New to Julia\
**Tags:** question\
**Created:** [June 27, 2019, 9:13pm UTC](https://discourse.julialang.org/t/sparse-array-solver-on-gpu/25780 "2019-06-27T21:13:28Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![cholt](https://avatars.discourse-cdn.com/v4/letter/c/ba9def/32.png) [@cholt](https://discourse.julialang.org/u/cholt)\
**Post date:** [June 27, 2019, 9:13pm UTC](https://discourse.julialang.org/t/sparse-array-solver-on-gpu/25780/1 "2019-06-27T21:13:28Z")

</div>

So I’m trying to invert a sparse array, A, using a GPU. I would either like to find the true inverse, or find the solution, x, to A\*x = ones(length(x)).

Here’s the code I’ve been using:

 ![image](https://global.discourse-cdn.com/julialang/original/3X/d/8/d82582f4aa56f55922af244e0fbe334c4e6e3cca.png)

I don’t know why x gives me numbers back, but xGPU gives NaN.

Thanks for your help

---

<div class="post-metadata">

**Author:** ![cholt](https://avatars.discourse-cdn.com/v4/letter/c/ba9def/32.png) [@cholt](https://discourse.julialang.org/u/cholt)\
**Post date:** [June 27, 2019, 9:16pm UTC](https://discourse.julialang.org/t/sparse-array-solver-on-gpu/25780/2 "2019-06-27T21:16:44Z")

</div>

Fixed it. I used cuones(size(A,1)) instead of CuVector(y).