CHANGELOG for 2.26.0 (#750)
This PR contains the CHANGELOG of 2.26.0 (#749) Co-authored-by: Victor Magri <paludettomag1@llnl.gov>
This commit is contained in:
parent
403e16a997
commit
94cb9a3137
16
CHANGELOG
16
CHANGELOG
@ -7,10 +7,24 @@
|
|||||||
# This file chronicles user-level changes, beginning with the most recent.
|
# This file chronicles user-level changes, beginning with the most recent.
|
||||||
# =============================================================================
|
# =============================================================================
|
||||||
|
|
||||||
|
Version 2.26.0 released 2022/10/14
|
||||||
|
|
||||||
|
- Added a new feature of BoomerAMG (L1-Jacobi and Chebyshev smoothers) and
|
||||||
|
Krylov subspace solvers with multiple right-hand-side vectors;
|
||||||
|
- Optimized parallel sparse matrix/multi-component vector multiplication
|
||||||
|
on CPUs and GPUs.
|
||||||
|
- Added SYCL support for the IJ interface and various AMG operators;
|
||||||
|
- Added a runtime switch feature between CPU and GPU;
|
||||||
|
- Added CMAKE support for CUDA;
|
||||||
|
- Optimized the performance of AMG interpolation truncation routines with CUDA/HIP;
|
||||||
|
- Optimized the performance of hypre's SpGEMM kernels on NVIDIA/AMD GPUs;
|
||||||
|
- Various bug fixes
|
||||||
|
|
||||||
|
|
||||||
Version 2.25.0 released 2022/06/15
|
Version 2.25.0 released 2022/06/15
|
||||||
|
|
||||||
- New features in MGR: multilevel assignment of solver options, block Jacobi for
|
- New features in MGR: multilevel assignment of solver options, block Jacobi for
|
||||||
smoothers and integrid transfer, CPR, non-Galerkin.
|
smoothers and inter-grid transfer, CPR, non-Galerkin.
|
||||||
- New factorized sparse approximate inverse (FSAI) solver/preconditioner. Can
|
- New factorized sparse approximate inverse (FSAI) solver/preconditioner. Can
|
||||||
also be used as a smoother in BoomerAMG.
|
also be used as a smoother in BoomerAMG.
|
||||||
- Added various optimizations for Nvidia and AMD GPUs by removing synchronizations,
|
- Added various optimizations for Nvidia and AMD GPUs by removing synchronizations,
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user