[Diff since v0.1.4](v0.1.4...v0.2.0)

**Merged pull requests:**
- Update to AMGX_jll 2.4 and later Julia versions (#33) (@moyner)

## AMGX v0.1.4

[Diff since v0.1.3](v0.1.3...v0.1.4)


**Closed issues:**
- Support for CUDA v4? (#26)

**Merged pull requests:**
- Update CUDA to 4 (#27) (@vchuravy)

## AMGX v0.1.3

[Diff since v0.1.2](v0.1.2...v0.1.3)


**Closed issues:**
- Changing config parameters after solver creation (#19)
- Weird error for special matrix values (#21)
- `using AMGX` errors on non-supported machines -- is there a good way to conditionally use AMGX? (#22)

**Merged pull requests:**
- Fix typo in README (#20) (@navidcy)
- Initialize `libAMGX` only if `AMGX_jll.libamgxsh` is available (#24) (@navidcy)
- Run CI tests also on Julia 1.8 (#25) (@navidcy)

## AMGX v0.1.2

[Diff since v0.1.1](v0.1.1...v0.1.2)


**Closed issues:**
- Bump compat for CUDA v3? (#16)

**Merged pull requests:**
- update CUDA compat (#17) (@KristofferC)

## AMGX v0.1.1

[Diff since v0.1.0](v0.1.0...v0.1.1)



**Merged pull requests:**
- loosen CUDA compat (#15) (@KristofferC)

## AMGX v0.1.0

**Closed issues:**
- Set up CI (#9)
- Add nicer show methods. (#11)

**Merged pull requests:**
- CompatHelper: add new compat entry for "JSON" at version "0.21" (#1) (@github-actions[bot])
- use AMGX_jll for binaries (#3) (@KristofferC)
- add buildkite CI (#10) (@KristofferC)
- add a warning if objects are finalized without being destroyed (#12) (@KristofferC)
- add nicer show methods (#13) (@KristofferC)